Guide to LaTeX (4th Edition) (Tools and Techniques

A Guide to 

LAT E X 

and Electronic Publishing 

Fourth edition 

Helmut Kopka 

Patrick W. Daly 

Addison-Wesley 

Harlow, England Reading, Massachusetts Menlo Park, California 

New York Don Mills, Ontario Amsterdam Bonn Sydney Singapore 

Tokyo Madrid San Juan Milan Mexico City Seoul Taipei

© Addison Wesley Longman Limited 2004 

Addison Wesley Longman Limited 

Edinburgh Gate 

Harlow 

Essex CM20 2JE 

England 

and Associated Companies throughout the World. 

The rights of Helmut Kopka and Patrick W. Daly to be identified as authors of 

this Work have been asserted by them in accordance with the Copyright, 

Designs and Patents Act 1988. 

All rights reserved. No part of this publication may be reproduced, stored in a 

retrieval system, or transmitted in any form or by any means, electronic, 

mechanical, photocopying, recording or otherwise, without either the prior 

written permission of the publisher or a licence permitting restricted copying in 

the United Kingdom issued by the Copyright Licensing Agency Ltd, 

90 Tottenham Court Road, London W1P 9HE. 

The programs in this book have been included for their instructional value. 

They have been tested with care but are not guaranteed for any particular 

purpose. The publisher does not offer any warranties or representations nor 

does it accept any liabilities with respect to the programs. 

Many of the designations used by manufacturers and sellers to distinguish their 

products are claimed as trademarks. Addison Wesley Longman Limited has 

made every attempt to supply trademark information about manufacturers and 

their products mentioned in this book. A list of the trademark designations and 

their owners appears on page v. 

Cover designed by Designers & Partners, Oxford 

Typeset by the authors with the L AT E X Documentation System 

Printed in Great Britain by Henry Ling Ltd, at the Dorset Press, Dorchester, 

Dorset 

First published 1993 

Second edition 1995 

Third edition 1999. Reprinted 1999, 2000 

Fourth edition 2004 

ISBN ???????????? 

British Library Cataloguing-in-Publication Data 

A catalogue record for this book is available from the British Library 

Library of Congress Cataloging-in-Publication Data 

Kopka, Helmut. 

A guide to L AT E X : and Electronic Publishing 

/ Helmut Kopka, Patrick W. Daly -- 4th ed. 

p. cm. 

Includes bibliographical references and index. 

ISBN 0-201-39825-7 

1. L AT E X (Computer file) 2. Computerized typesetting. I. Daly, 

Patrick W. II. Title. 

???????????? 

??????????? ??????? 

CIP

Trademark notices 

METAFONT is a trademark of Addison-Wesley Publishing Company. 

T E X, A M S-T E X, and A M S-L AT E X are trademarks of the American 

Mathematical Society. 

Lucida is a trademark of Bigelow & Holmes. 

Microsoft , MS-DOS , Windows , Internet Explorer are registered 

trademarks of Microsoft Corporation. 

PostScript , Acrobat Reader , Acrobat logo are registered trademarks and 

PDF a trademark of Adobe Systems Incorporated. 

UNIX is a registered trademark in the United States and other countries, 

licensed exclusively through X/Open Company, Limited. 

VAX and VMS are trademarks of Digital Equipment Corporation. 

IBM is a registered trademark and techexplorer Hypermedia Browser a 

trademark of International Business Machines Corporation. 

Netscape and Netscape Navigator are trademarks of Netscape 

Communications Corporation. 

TrueType is a trademark and Apple and Macintosh are registered 

trademarks of Apple Computer Inc. 

v

Preface 

A new edition to A Guide to L AT E X begs the fundamental question: Has 

L AT E X changed so much since the appearance of the third edition in 1999 

that a new release of this manual is justified? 

The simple answer to that question is ‘Well . . . .’ In 1994, the L AT E X world 

was in upheaval with the issue of the new version L AT E X 2ε, and the second 

edition of the Guide came out just then to act as the bridge between the 

old and new versions. By 1998, the initial teething problems had been 

worked out and corrected through semi-annual releases, and the third 

edition could describe an established, working system. However, homage 

was still paid to the older 2.09 version since many users still employed its 

familiar syntax, although they were most likely to be using it in a L AT E X 2ε 

environment. L AT E X has now reached a degree of stability that since 2000 

the regular updates have been reduced to annual events, which often 

appear months after the nominal date, something that does not worry 

anyone. The old version 2.09 is obsolete and should no longer play any 

role in such a manual. In this fourth edition, it is reduced to an appendix 

just to document its syntax and usage. 

But if L AT E X itself has not changed substantially since 1999, many of its 

peripherals have. The rise of programs like pdfT E X and dvipdfm for PDF 

output adds new possibilities, which are realized, not in L AT E X directly, but 

by means of more modern packages to extend the basic features. The 

distribution of T E X/L AT E X installations has changed, such that most users 

are given a complete, ready-to-run setup, with all the ‘extras’ that one 

used to have to obtain oneself. Those extras include user-contributed 

packages, many of which are now considered indispensable. Today ‘the 

L AT E X system’ includes much more than the basic kernel by Leslie Lamport, 

encompassing the contributions of hundreds of other people. This edition 

reflects this increase in breadth. 

The changes to the fourth edition are mainly those of emphasis. 

1. The material has been reorganized into ‘Basics’ and ‘Beyond the 

Basics’ (‘advanced’ sounds too intimidating) while the appendices 

contain topics that really can be skipped by most everyday users. 

Exception: Appendix H is an alphabetized command summary that 

many people find extremely useful (including ourselves). 

This reorganizing is meant to stress certain aspects over others. For 

vii

viii Preface 

example, the section on graphics inclusion and color was originally 

treated as an exotic freak, relegated to an appendix on extensions; 

in the third edition, it moved up to be included in a front chapter 

along with the picture environment and floats; now it dominates 

Chapter 6 all on its own, the floats come in the following Chapter 7, 

and picture is banished to the later Chapter 13. This is not to say 

that the picture features are no good, but only that they are very 

specialized. We add descriptions of additional drawing possibilities 

there too. 

2. It is stressed as much as possible that L AT E X is a markup language, 

with separation of content and form. Typographical settings should 

be placed in the preamble, while the body contains only logical 

markup. This is in keeping with the modern ideas of XML, where 

form and content are radically segregated. 

3. Throughout this edition, contributed packages are explained at that 

point in the text where they are most relevant. The fancyhdr 

package comes in the section on page styles, natbib where literature 

citations are explained. This stresses that these ‘extensions’ are part 

of the L AT E X system as a whole. However, to remind the user that 

they must still be explicitly loaded, a marginal note is placed at the 

start of their descriptions. 

4. PDF output is taken for granted throughout the book, in addition 

to the classical DVI format. This means that the added possibilities 

of pdfT E X and dvipdfm are explained where they are relevant. A 

separate Chapter 10 on PostScript and PDF is still necessary, and the 

best interface to PDF output, the hyperref package by Sebastian 

Rahtz, is explained in detail. PDF is also included in Chapter 15 on 

presentation material. 

On the other hand, the other Web output formats, HTML and XML, 

are only dealt with briefly in Appendix E, since these are large topics 

treated in other books, most noticeably the L AT E X Web Companion. 

5. This book is being distributed with the T E XLive CD, with the kind 

permission of Sebastian Rahtz who maintains it for the T E X Users 

Group. It contains a full T E X and L AT E X installation for Windows, 

Macintosh, and Linux, plus many of the myriad extensions that 

exist. 

We once again express our hope that this Guide will prove more than 

useful to all those who wish to find their way through the intricate world 

of L AT E X. And with the addition of the T E XLive CD, that world is brought 

even closer to their doorsteps. 

Helmut Kopka and Patrick W. Daly 

June, 2003

Contents 

Preface vii 

I Basics 1 

1 Introduction 3 

1.1 Just what is L AT E X? . . . . . . . . . . . . . . . . . . . . . . . . . . 3 

1.2 Markup Languages . . . . . . . . . . . . . . . . . . . . . . . . . 4 

1.3 T E X and its offspring . . . . . . . . . . . . . . . . . . . . . . . . 6 

1.4 How to use this book . . . . . . . . . . . . . . . . . . . . . . . . 10 

1.5 Basics of a L AT E X file . . . . . . . . . . . . . . . . . . . . . . . . . 11 

1.6 T E X processing procedure . . . . . . . . . . . . . . . . . . . . . 14 

2 Text, Symbols, and Commands 17 

2.1 Command names and arguments . . . . . . . . . . . . . . . . 17 

2.2 Environments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 

2.3 Declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 

2.4 Lengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 

2.5 Special characters . . . . . . . . . . . . . . . . . . . . . . . . . . 22 

2.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 

2.7 Fine-tuning text . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 

2.8 Word division . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 

3 Document Layout and Organization 37 

3.1 Document class . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 

3.2 Page style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 

3.3 Parts of the document . . . . . . . . . . . . . . . . . . . . . . . 52 

3.4 Table of contents . . . . . . . . . . . . . . . . . . . . . . . . . . 58 

4 Displayed Text 61 

4.1 Changing font . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 

4.2 Centering and indenting . . . . . . . . . . . . . . . . . . . . . . 67 

4.3 Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 

4.4 Generalized lists . . . . . . . . . . . . . . . . . . . . . . . . . . 74 

4.5 Theorem-like declarations . . . . . . . . . . . . . . . . . . . . 80 

ix

x CONTENTS 

4.6 Tabulator stops . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 

4.7 Boxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 

4.8 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 

4.9 Printing literal text . . . . . . . . . . . . . . . . . . . . . . . . . 110 

4.10 Footnotes and marginal notes . . . . . . . . . . . . . . . . . . 112 

4.11 Comments within text . . . . . . . . . . . . . . . . . . . . . . . 118 

5 Mathematical Formulas 119 

5.1 Mathematical environments . . . . . . . . . . . . . . . . . . . 119 

5.2 Main elements of math mode . . . . . . . . . . . . . . . . . . 120 

5.3 Mathematical symbols . . . . . . . . . . . . . . . . . . . . . . . 124 

5.4 Additional elements . . . . . . . . . . . . . . . . . . . . . . . . 130 

5.5 Fine-tuning mathematics . . . . . . . . . . . . . . . . . . . . . 145 

5.6 Beyond standard L AT E X . . . . . . . . . . . . . . . . . . . . . . . 151 

6 Graphics Inclusion and Color 153 

6.1 The graphics packages . . . . . . . . . . . . . . . . . . . . . . 153 

6.2 Adding color . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 

7 Floating tables and figures 169 

7.1 Float placement . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 

7.2 Postponing floats . . . . . . . . . . . . . . . . . . . . . . . . . . 171 

7.3 Style parameters for floats . . . . . . . . . . . . . . . . . . . . 171 

7.4 Float captions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 

7.5 Float examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 

7.6 References to figures and tables in text . . . . . . . . . . . . 177 

7.7 Some float packages . . . . . . . . . . . . . . . . . . . . . . . . 178 

8 User Customizations 181 

8.1 Counters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 

8.2 Lengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 

8.3 User-defined commands . . . . . . . . . . . . . . . . . . . . . . 185 

8.4 User-defined environments . . . . . . . . . . . . . . . . . . . . 195 

8.5 Some comments on user-defined structures . . . . . . . . . 200 

II Beyond the Basics 205 

9 Document Management 207 

9.1 Processing parts of a document . . . . . . . . . . . . . . . . . 207 

9.2 In-text references . . . . . . . . . . . . . . . . . . . . . . . . . . 213 

9.3 Bibliographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 

9.4 Keyword index . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

CONTENTS xi 

10 PostScript and PDF 231 

10.1 L AT E X and PostScript . . . . . . . . . . . . . . . . . . . . . . . . 231 

10.2 Portable Document Format . . . . . . . . . . . . . . . . . . . . 236 

11 Multilingual L AT E X 251 

11.1 The babel system . . . . . . . . . . . . . . . . . . . . . . . . . 252 

11.2 Contents of the language.dat file . . . . . . . . . . . . . . . 256 

12 Math Extensions with A M S-L AT E X 257 

12.1 Invoking A M S-L AT E X . . . . . . . . . . . . . . . . . . . . . . . . 258 

12.2 Standard features of A M S-L AT E X . . . . . . . . . . . . . . . . . 258 

12.3 Further A M S-L AT E X packages . . . . . . . . . . . . . . . . . . . 280 

12.4 The A M S fonts . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 

13 Drawing with L AT E X 287 

13.1 The picture environment . . . . . . . . . . . . . . . . . . . . 287 

13.2 Extended pictures . . . . . . . . . . . . . . . . . . . . . . . . . . 302 

13.3 Other drawing packages . . . . . . . . . . . . . . . . . . . . . . 307 

14 Bibliographic Databases and BIBT E X 309 

14.1 The BIBT E X program . . . . . . . . . . . . . . . . . . . . . . . . . 309 

14.2 Creating a bibliographic database . . . . . . . . . . . . . . . . 311 

14.3 Customizing bibliography styles . . . . . . . . . . . . . . . . 321 

15 Presentation Material 323 

15.1 Slide production with SLIT E X . . . . . . . . . . . . . . . . . . . 324 

15.2 Slide production with seminar . . . . . . . . . . . . . . . . . 330 

15.3 Electronic documents for screen viewing . . . . . . . . . . . 340 

15.4 Special effects with PDF . . . . . . . . . . . . . . . . . . . . . . 343 

16 Letters 351 

16.1 The L AT E X letter class . . . . . . . . . . . . . . . . . . . . . . . 351 

16.2 A house letter style . . . . . . . . . . . . . . . . . . . . . . . . . 356 

16.3 A model letter customization . . . . . . . . . . . . . . . . . . 359 

Appendices 

A The New Font Selection Scheme (NFSS) 367 

A.1 Font attributes under NFSS . . . . . . . . . . . . . . . . . . . . 368 

A.2 Simplified font selection . . . . . . . . . . . . . . . . . . . . . . 370 

A.3 Installing fonts with NFSS . . . . . . . . . . . . . . . . . . . . . 372

xii CONTENTS 

B The L AT E X Clockwork 381 

B.1 Installing L AT E X . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381 

B.2 Obtaining the Adobe euro fonts . . . . . . . . . . . . . . . . . 387 

B.3 T E X directory structure . . . . . . . . . . . . . . . . . . . . . . . 387 

B.4 The CTAN server . . . . . . . . . . . . . . . . . . . . . . . . . . 389 

B.5 Additional standard files . . . . . . . . . . . . . . . . . . . . . 391 

B.6 The various L AT E X files . . . . . . . . . . . . . . . . . . . . . . . 396 

C Error Messages 401 

C.1 Basic structure of error messages . . . . . . . . . . . . . . . . 401 

C.2 Some sample errors . . . . . . . . . . . . . . . . . . . . . . . . 409 

C.3 List of L AT E X error messages . . . . . . . . . . . . . . . . . . . . 415 

C.4 T E X error messages . . . . . . . . . . . . . . . . . . . . . . . . . 424 

C.5 Warnings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429 

C.6 Search for subtle errors . . . . . . . . . . . . . . . . . . . . . . 435 

D L AT E X Programming 437 

D.1 Class and package files . . . . . . . . . . . . . . . . . . . . . . 437 

D.2 L AT E X programming commands . . . . . . . . . . . . . . . . . . 440 

D.3 Sample packages . . . . . . . . . . . . . . . . . . . . . . . . . . 451 

D.4 Changing preprogrammed text . . . . . . . . . . . . . . . . . 459 

D.5 Direct typing of special letters . . . . . . . . . . . . . . . . . . 461 

D.6 Alternatives for special symbols . . . . . . . . . . . . . . . . . 462 

D.7 Managing code and documentation . . . . . . . . . . . . . . . 462 

E L AT E X and World Wide Web 475 

E.1 Converting to HTML . . . . . . . . . . . . . . . . . . . . . . . . 476 

E.2 The Extensible Markup Language: XML . . . . . . . . . . . . . 478 

E.3 The techexplorer Hypermedia Browser . . . . . . . . . . . 481 

F Obsolete L AT E X 483 

F.1 The 2.09 preamble . . . . . . . . . . . . . . . . . . . . . . . . . 483 

F.2 Font selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484 

F.3 Obsolete means obsolete . . . . . . . . . . . . . . . . . . . . . 485 

G T E X Fonts 487 

G.1 Font metrics and bitmaps . . . . . . . . . . . . . . . . . . . . . 487 

G.2 Computer Modern fonts . . . . . . . . . . . . . . . . . . . . . . 488 

G.3 The METAFONT program . . . . . . . . . . . . . . . . . . . . . 497 

G.4 Extended Computer fonts . . . . . . . . . . . . . . . . . . . . . 498 

G.5 PostScript fonts . . . . . . . . . . . . . . . . . . . . . . . . . . . 503 

G.6 Computer Modern as PostScript fonts . . . . . . . . . . . . . 505

LIST OF TABLES xiii 

H Command Summary 507 

H.1 Brief description of the L AT E X commands . . . . . . . . . . . . 507 

H.2 Summary tables and figures . . . . . . . . . . . . . . . . . . . 595 

Bibliography 605 

Index 607 

List of Tables 

10.1 The psnfss packages and their fonts . . . . . . . . . . . . . 234 

10.2 Acrobat menu actions . . . . . . . . . . . . . . . . . . . . . . . 248 

A.1 The NFSS encoding schemes . . . . . . . . . . . . . . . . . . . 368 

A.2 The NFSS series attributes . . . . . . . . . . . . . . . . . . . . . 369 

A.3 Attributes of the Computer Modern fonts . . . . . . . . . . . 370 

D.1 Input coding schemes for inputenc package . . . . . . . . . 462 

D.2 Alternative commands for special symbols . . . . . . . . . . 463 

G.1 Computer Modern text fonts . . . . . . . . . . . . . . . . . . . 491 

G.2 Root names of the 35 standard PostScript fonts . . . . . . . 504 

G.3 Encoding suffixes . . . . . . . . . . . . . . . . . . . . . . . . . . 505 

H.1 Font attribute commands . . . . . . . . . . . . . . . . . . . . . 595 

H.2 Math alphabet commands . . . . . . . . . . . . . . . . . . . . . 595 

H.3 Font sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595 

H.4 L AT E X 2.09 font declarations . . . . . . . . . . . . . . . . . . . . 595 

H.5 Dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 

H.6 Accents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 

H.7 Special letters from other languages . . . . . . . . . . . . . . 596 

H.8 Special symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 

H.9 Command symbols . . . . . . . . . . . . . . . . . . . . . . . . . 596 

H.10 Greek letters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596 

H.11 Binary operation symbols . . . . . . . . . . . . . . . . . . . . . 597 

H.12 Relational symbols . . . . . . . . . . . . . . . . . . . . . . . . . 597 

H.13 Negated relational symbols . . . . . . . . . . . . . . . . . . . . 597 

H.14 Brackets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 597 

H.15 Arrows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598 

H.16 Miscellaneous symbols . . . . . . . . . . . . . . . . . . . . . . 598 

H.17 Mathematical symbols in two sizes . . . . . . . . . . . . . . . 598 

H.18 Function names . . . . . . . . . . . . . . . . . . . . . . . . . . . 598 

H.19 Math accents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599 

H.20 A M S arrows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599 

H.21 A M S binary operation symbols . . . . . . . . . . . . . . . . . 599

xiv CONTENTS 

H.22 A M S Greek and Hebrew letters . . . . . . . . . . . . . . . . . 600 

H.23 A M S delimiters . . . . . . . . . . . . . . . . . . . . . . . . . . . 600 

H.24 A M S relational symbols . . . . . . . . . . . . . . . . . . . . . . 600 

H.25 A M S negated relational symbols . . . . . . . . . . . . . . . . 601 

H.26 Miscellaneous A M S symbols . . . . . . . . . . . . . . . . . . . 601 

List of Figures 

1.1 Sample display with the WinEdt editor . . . . . . . . . . . . . 16 

3.1 Page layout parameters . . . . . . . . . . . . . . . . . . . . . . 48 

3.2 Sample title page . . . . . . . . . . . . . . . . . . . . . . . . . . 53 

4.1 The list parameters . . . . . . . . . . . . . . . . . . . . . . . 76 

6.1 An embellished image file . . . . . . . . . . . . . . . . . . . . . 160 

10.1 Output produced by pdfT E X with the hyperref package . . 240 

13.1 Comparison of eepic with eepicemu . . . . . . . . . . . . . 306 

15.2 Title page of a pdfscreen document . . . . . . . . . . . . . . 341 

B.1 The T E XLive welcome . . . . . . . . . . . . . . . . . . . . . . . . 383 

B.2 The T E XLive documentation browser . . . . . . . . . . . . . . 384 

B.3 The TDS directory tree . . . . . . . . . . . . . . . . . . . . . . . 388 

B.4 Partial directory tree of CTAN servers . . . . . . . . . . . . . 390 

E.1 Example of T E X4ht and techexplorer output . . . . . . . . 482 

H.1 Single column page format . . . . . . . . . . . . . . . . . . . . 602 

H.2 Double column page format . . . . . . . . . . . . . . . . . . . 603 

H.3 Format of the list environment . . . . . . . . . . . . . . . . 604

Part I 

Basics

1 

1.1 Just what is L AT E X? 

To summarize very briefly: 

Introduction 

• L AT E X is a comprehensive set of markup commands used with the 

powerful typesetting program T E X for the preparation of a wide 

variety of documents, from scientific articles, reports, to complex 

books. 

• L AT E X like T E X is an open software system, available free of charge. 

Its core is maintained by the L AT E X3 Project Group but it also benefits 

from extensions written by hundreds of user/contributors, with all 

the advantages and disadvantages of such a democracy. 

• A L AT E X document consists of one or more source files containing 

plain text characters, the actual textual content plus markup 

commands. These include instructions which can insert graphical 

material produced by other programs. 

• It is processed by the T E X program to produce a binary file in DVI 

(device independent) format, containing precise directions for the 

typesetting of each character. This in turn can be viewed on a monitor, 

or converted into printer instructions, or some other electronic 

form such as PostScript, HTML, XML, or PDF. 

• A variant on the T E X program called pdfT E X produces PDF output 

directly from the source file without going through the DVI intermediary. 

With this, L AT E X can automatically include internal links 

and bookmarks with little or no extra effort, plus PDF buttons and 

external links, in addition to graphics in a wide range of common 

formats. 

• T E X activities are coordinated by the T E X Users Group, TUG (www. 

tug.org) who distribute a set of CDs, called T E XLive, annually to its 

3

4 Chapter 1. Introduction 

members, containing a T E X/L AT E X installation for various computer 

types. 

The rest of this book attempts to fill in the gaps in the above summary. 

With the help of the included T E XLive CD for Windows, Macintosh, and 

Linux, which also contains a directory specific to this book (\books\ 

Kopka_and_Daly\), we hope that the user will have additional pleasure 

in learning the joys of L AT E X. 

1.2 Markup Languages 

1.2.1 Typographical markup 

In the days before computers, an author would prepare a manuscript 

either by hand or by typewriter, which he or she would submit to a publisher. 

Once accepted for publication (and after several rounds of corrections 

and modifications, each requiring a rewrite of the paper manuscript), 

it would be sent to a copy editor, a human being who would decorate the 

manuscript with markup, marginal notes that inform the typesetter (another 

human being) which fonts and spacings and other typographical 

features should be used to convert it to the final printed form that one 

expects of books and articles. 

Electronic processing of text today follows a similar procedure, except 

that the humans have been replaced by computer programs. (So far the 

author has avoided this fate, but they are working on it.) The markup 

is normally included directly in the manuscript in such a way that it is 

converted immediately to its output form and displayed on the computer 

monitor. This is known as WYSIWYG, or ‘what you see is what you get’. 

However, what you see is not always what you’ve got. An alternative 

that is used more and more by major publishers is markup languages, 

in which the raw text is interspersed with indicators ‘for the typesetter.’ 

The result as seen on the monitor is much the same as a typewritten 

manuscript, except that the markup is no longer abbreviated marginal 

notes, but cryptic code within the actual text. This source text, which 

can be prepared by a simple, dumb text editor program, is converted into 

typographically set output by a separate program. 

For example, to code the line 

He took a bold step forward. 

with HTML, the classical markup language of the World Wide Web, one 

enters in the source text: 

He took a bold step forward. 

In Plain T E X, the same sentence would be coded as:

He took a {\bf bold step} forward. 

1.2. Markup Languages 5 

The first example is to be processed (displayed) by a Web browser program 

that decides to set everything between and as bold face. The 

second example is intended for the T E X program (Section 1.3). The markup 

in these two examples follow different rules, different syntax, but the 

functionality is the same. 

1.2.2 Logical markup 

The above examples illustrate typographical markup, where the inserted 

commands or tags give direct instructions to alter the appearance of the 

output, here a change of font. An alternative is to indicate the purpose 

of the text. For example, HTML recognizes several levels of headings; to 

place a title into the highest level one enters: 

Logical Markup 

The equivalent L AT E X entry would be: 

\section{Logical Markup} 

With this logical markup, the author concentrates entirely on the content 

and leaves the typographical considerations to the experts. One 

merely marks the structure of the document, and has no means of controlling 

how the logical elements, like section titles, are to be rendered 

typographically. This information is put into HTML style sheets or L AT E X 

classes and packages, which are external to the actual source file. This 

means that the entire layout of a document can be overhauled with only 

minimal or even no alterations to the source file. 

Today much effort is being put into XML, the Extensible Markup Language, 

as the ultimate markup system, since it allows the markup, or 

tags, to be defined as needed, without any indication of how they are to 

be implemented. That is left to XSL, the Extensible Stylesheet Language. 

It must be emphasized that neither XML nor XSL are programs at all; they 

are specifications for how documents and databases may be marked up, 

and how the markup tags may be translated into real output. Programs 

still need to be found to do the actual job. 

And that is the fundamental idea behind markup languages: that the 

source text indicates the logical structure of its contents. Such source 

files, being written in plain ascii text, are extremely robust, not being 

married to any particular software package or computer type. 

What does all this have to do with L AT E X? In the next Section we outline 

the development of T E X and L AT E X, and go on to show that L AT E X, a product 

of the mid 1980’s, is a programmable markup language that is ideally 

suited for the modern world of electronic publishing.


1.3 T E X and its offspring 

The most powerful formatting program for producing book quality text 

of scientific and technical works is that of Donald E. Knuth (Knuth, 1986a, 

1986b, 1986c, 1986d, 1986e). The program is called T E X, which is a 

rendering in capitals of the Greek letters τɛχ. For this reason the last 

letter is pronounced not as an x, but as the ch in Scottish loch or German 

ach, or as the Spanish j or Russian kh. The name is meant to emphasize 

that the printing of mathematical texts is an integral part of the program 

and not a cumbersome add-on. In addition to T E X, the same author has 

developed a further program called METAFONT for the production of 

character fonts. The standard T E X program package contains 75 fonts 

in various design sizes, each of which is also available in up to eight 

magnification steps. All these fonts were produced with the program 

METAFONT. With additional applications, further character fonts have 

been created, such as for Cyrillic, Chinese, and Japanese, with which texts 

in these alphabets can be printed in book quality. 

The T E X program is free, and the source code is readily available. 

Anybody may take it and modify it as they like, provided they call the 

result something other than T E X. This indeed has occurred, and several 

T E X variants do exist, including pdfT E X which we deal with later in this 

Chapter. Only Knuth is allowed to alter T E X itself, which he does only to 

correct any obvious bugs. Otherwise, he considers T E X to be completed; 

the current version number is 3.14159, and with his death, the code will 

be frozen for all time, and the version number will become exactly π. 

1.3.1 The T E X program 

The basic T E X program only understands a set of very primitive commands 

that are adequate for the simplest of typesetting operations and 

programming functions. However, it does allow more complex, higherlevel 

commands to be defined in terms of the primitive ones. In this way, 

a more user-friendly environment can be constructed out of the low-level 

building blocks. 

During a processing run, the program first reads in a so-called format 

file which contains the definitions of the higher-level commands in terms 

of the primitive ones, and which also contains the hyphenation patterns 

for word division. Only then does it read in the author’s source file containing 

the actual text to be processed, including formatting commands 

that are predefined in the format file. 

Creating new formats is something that should be left to very knowledgeable 

programmers. The definitions are written to a source file which 

is then processed with a special version of the T E X program called initex. 

It stores the new format file in a compact manner so that it can be read 

in quickly by the regular T E X program.

1.3. T E X and its offspring 7 

Although the normal user will almost never write such a format, he 

or she may be presented with a new format source file that will need to 

be installed with initex. For example, this is just what must be done to 

upgrade L AT E X periodically. How to do this is described in Appendix B. 

1.3.2 Plain T E X 

1.3.3 L AT E X 

Knuth has provided a basic format named Plain T E X to interact with T E X 

at its simplest level. This is such a fundamental part of T E X processing 

that one tends to forget the distinction between the actual processing 

program T E X and this particular format. Most people who claim to ‘work 

only with T E X’ really mean that they only work with Plain T E X. 

Plain T E X is also the basis of every other format, something that only 

reinforces the impression that T E X and Plain T E X are one and the same. 

The emphasis of Plain T E X is still very much at the typesetter’s level, 

rather than the author’s. Furthermore, the exploitation of all its potential 

demands considerable experience with programming techniques. Its 

application thus remains the exclusive domain of typographic and programming 

professionals. 

For this reason, the American computer scientist Leslie Lamport has 

developed the L AT E X format (Lamport, 1985), which provides a set of 

higher-level commands for the production of complex documents. With 

it, even the user with no knowledge of typesetting or programming is in 

a position to take extensive advantage of the possibilities offered by T E X, 

and to be able to produce a variety of text outputs in book quality within 

a few days, if not hours. This is especially true for the production of 

complex tables and mathematical formulas. 

As pointed out in Section 1.2.2, L AT E X is very much more a logical 

markup language than the original Plain T E X, on which it is based. It 

contains provisions for automatic running heads, sectioning, tables of 

contents, cross-referencing, equation numbering, citations, floating tables 

and figures, without the author having to know just how these are to be 

formatted. The layout information is stored in additional class files which 

are referred to but not included in the input text. The predefined layouts 

may be accepted as they are, or replaced by others with minimal changes 

to the source file. 

Since its introduction in the mid-1980s, L AT E X has been periodically 

updated and revised, like all software products. For many years the 

version number was fixed at 2.09 and the revisions were only identified 

by their dates. The last major update occurred on December 1, 1991, with 

some minor corrections up to March 25, 1992, at which point L AT E X 2.09 

became frozen.


1.3.4 L AT E X 2ε 

The enormous popularity of L AT E X and its expansion into fields for which 

it was not originally intended, together with improvements in computer 

technology, especially dealing with cheap but powerful laser printers, had 

created a diversity of formats bearing the L AT E X label. In an effort to 

re-establish a genuine, improved standard, the L AT E X3 Project was set up 

in 1989 by Leslie Lamport, Frank Mittelbach, Chris Rowley, and Rainer 

Schöpf. Their goal was to construct an optimized and efficient set of 

basic commands complemented by various packages to add specific functionality 

as needed. 

As the name of the project implies, its aim is to achieve a version 3 

for L AT E X. However, since that is the long-term goal, a first step towards it 

was the release of L AT E X 2ε in mid-1994 together with the publication of 

the second edition of Lamport’s basic manual (Lamport, 1994) and of an 

additional book (Goossens et al., 1994) describing many of the extension 

packages available and L AT E X programming in the new system. Since then, 

two further books have appeared, Goossens et al. (1997) dealing with the 

inclusion of graphics and color, and Goossens and Rahtz (1999) explaining 

how L AT E X may be used with the World Wide Web. Both these topics are 

also dealt with in this Guide. 

Initially updates to L AT E X 2ε were issued twice a year, in June and 

December, but it has now become so stable that since 2000 the changes 

are released only once a year, nominally in June. 

L AT E X 2ε is now the standard version, and L AT E X 2.09 is considered 

obsolete, although source files intended for the older version may still be 

processed with the newer one. In this book, unless otherwise indicated, 

‘L AT E X’ will always mean L AT E X 2ε. 

1.3.5 T E X fonts 

T E X initially made use of its own set of fonts, called Computer Modern 

generated by Knuth’s METAFONT program. The reason for doing this 

was that printers at that time (and even today) may contain their own 

preloaded fonts, but they are often slightly different from printer to 

printer. Furthermore, they lacked the mathematical character sets that 

are essential to T E X’s main hallmark, mathematical typesetting. So Knuth 

created pixel fonts that could be sent to every printer ensuring the same 

results everywhere. 

Today the situation with fonts has changed dramatically. Outline fonts 

(also known as type 1 fonts) are more compact and versatile than the pixel 

fonts (type 3). They also have a far superior appearance and are drawn 

much faster in PDF files. The original Computer Modern fonts have been 

converted to outline fonts, but there is no reason to stick with them, 

except possibly for the mathematical symbols. It is L AT E X 2ε with its New

1.3. T E X and its offspring 9 

Font Selection Scheme that freed T E X from its rigid marriage to Computer 

Modern. 

Fonts are discussed in more detail in Appendix G. 

1.3.6 The L AT E X bazaar: user contributions 

Like the T E X program on which is relies, L AT E X is freeware. There may be 

a prejudice that what is free in not worth anything, but there are other 

examples in the computer world to contradict this statement. And since 

the L AT E X macros are provided in files containing plain text, there is no 

problem to exchange, modify, and supplement them. In other words, the 

user can participate in extending the basic L AT E X system. 

Taking advantage of a mechanism in L AT E X 2.09 that allowed options 

to the default layouts to be contained in so-called style option files, many 

users began writing their own ‘options’ to provide additional features to 

the basic L AT E X. They then made these available to other users via the 

Internet. Many were intended for very specific problems, but many more 

proved to be of such general usefulness that they have become part of the 

standard L AT E X installation. In this way, the users themselves have built 

up a system that meets their needs. 

With L AT E X 2ε, these user contributions acquired official status: they 

became known as packages, they could be entered directly into the document 

and not by the back door, guidelines were issued for writing them, 

and additional commands were introduced to assist package programming. 

Package files bear the extension.sty from L AT E X 2.09 days, so that 

the older style option files may still function as packages today. 

Those packages that have established themselves as indispensable 

for sophisticated L AT E X processing are described in this book in those 

sections where they are most relevant. This does not imply that other 

packages are less worthwhile, but simply that this book does have to 

make a selection. Many other packages are described fully in The L AT E X 

Companion (Goossens et al., 1994) and it would go beyond the bounds of 

this book to reproduce it here. 

1.3.7 L AT E X and electronic publishing 

The most significant development in computer usage in the last decade 

is the rise of the World Wide Web (or the hijacking of the Internet by the 

glitzy society). L AT E X makes its own contribution here with 

• programs to convert L AT E X files to HTML (Appendix E); 

• means of creating PDF output, with hypertext features such as links, 

bookmarks, active buttons (Chapter 10);


• interfacing to XML both by acting as an engine to render XML documents 

and with programs to convert L AT E X to XML and vice versa 

(Appendix E). 

All these forms of electronic publishing are alternatives to traditional 

paper output. We do not expect paper to disappear entirely so quickly, 

but it is rapidly being replaced by electronic forms, which can always 

reproduce the paper whenever needed. 

1.4 How to use this book 

This Guide is meant to be a mixture of textbook and reference manual. 

It explains all the essential elements of the current standard L AT E X 2ε, but 

compared to Lamport (1985, 1994), it goes into more detail, offers more 

examples and exercises, and describes many ‘tricks’ based on the authors’ 

experiences. It explains not only the core L AT E X installation, but also many 

of the contributed packages that have become essential to modern L AT E X 

processing, and thus quasi-standard. We necessarily have to be selective, 

for we cannot go to the same extend as The L AT E X Companion (Goossens et 

al., 1994), The L AT E X Graphics Companion (Goossens et al., 1997), and The 

L AT E X Web Companion (Goossens and Rahtz, 1999), which are still valid 

companions to this book. 

The first part of the book is entitled The Basics, and deals with the more 

fundamental aspects of L AT E X: inputting text and symbols, document organization, 

lists and tables, entering mathematics, and customizations by 

the user. The second part is called Beyond the Basics, meaning it presents 

concepts which may be more advanced but which are still essential to 

producing complex, sophisticated documents. The distinction is rather 

arbitrary. Finally, the appendices contain topics that are not directly part 

of L AT E X itself, but useful for understanding its applications: installation, 

error messages, creating packages, World Wide Web, fonts. Appendix H 

is an alphabetized summary of most of the commands and their use, 

cross-referenced to their locations in the main text. 

1.4.1 Some conventions 

In the description of command syntax, typewriter type is used to indicate 

those parts that must be entered exactly as given, while italic is reserved 

for those parts that are variable or for the text itself. For example, the 

command to produce tables is presented as follows: 

\begin{tabular}{col form} lines \end{tabular} 

The parts in typewriter type are obligatory, while col form stands for the 

definition of the column format that must be inserted here. The allowed

1.5. Basics of a L AT E X file 11 

values and their combinations are given in the detailed descriptions of 

the commands. In the above example, lines stands for the line entries in 

the table and are thus part of the text itself. 

Sections describing a package, an extension to basic LAT Package: 

EX, have the 

sample name of that package printed as a marginal note, as demonstrated here 

for this paragraph. In this way, you are reminded that you must include 

it with \usepackage (Section 3.1.2) in order to obtain the additional 

features. Without it, you are likely to get an error message about undefined 

commands. 

Sections of text that are printed in a smaller typeface together with the boxed 

! exclamation mark at the left are meant as an extension to the basic description. 

They may be skipped over on a first reading. This information presents deeper 

insight into the workings of LAT EX than is necessary for everyday usage, but which 

is invaluable for creating more refined control over the output. 

1.5 Basics of a L AT E X file 

1.5.1 Text and commands 

The source file for L AT E X processing, or simply the L AT E X file, contains the 

source text that is to be processed to produce the printed output. Splitting 

the text up into lines of equal width, formatting it into paragraphs, and 

breaking it into pages with page numbers and running heads are all 

functions of the processing program and not of the input text itself. 

For example, words in the source text are strings of letters terminated 

by some non-letter, such as punctuation, blanks, or end-of-lines (hard 

end-of-lines, ones that are really there, not the soft ones that move with 

the window width); whereas punctuation marks will be transferred to the 

output, blanks and end-of-lines merely indicate a gap between words. 

Multiple blanks in the input, or blanks at the beginning of a line, have no 

effect on the interword spacing in the output. 

Similarly, a new paragraph is indicated in the input text by an empty 

line; multiple empty lines have the same effect as a single one. In the 

output, the paragraph may be formatted either by indentation of the first 

line, or by extra interline spacing, but this is not affected in any way by 

the number of blank lines or extra spaces in the input. 

The source file contains more than just text, however; it is also interspersed 

with markup commands that control the formatting or indicate 

the structure. It is therefore necessary for the author to be able to recognize 

what is text and what is a command. Commands consist either 

of certain single characters that cannot be used as text characters, or of 

words preceded immediately by a special character, the backslash (\). 

The syntax of source text is explained in detail in Chapter 2.


1.5.2 Contents of a L AT E X source file 

Every L AT E X file contains a preamble and a body. 

The preamble is a collection of commands that specify the global 

processing parameters for the following text, such as the paper format, 

the height and width of the text, the form of the output page with its 

pagination and automatic page heads and footlines. As a minimum, the 

preamble must contain the command \documentclass to specify the 

document’s overall processing type. This is the first command in the 

preamble. 

If there are no other commands in the preamble, L AT E X selects standard 

values for the line width, margins, paragraph spacing, page height and 

width, and much more. By default, these specifications are tailored to 

the American norms. For European requirements, built-in options exist to 

alter the text height and width to the A4 standard. Furthermore, there are 

language-specific packages to translate certain headings such as ‘Chapter’ 

and ‘Abstract’. 

The preamble ends with \begin{document}. Everything that follows 

this command is interpreted as body. It consists of the actual text mixed 

with markup commands. In contrast to those in the preamble, these 

commands have only a local effect, meaning they apply only to a part of 

the text, such as indentation, equations, temporary change of font, and so 

on. The body ends with the command \end{document}. This is normally 

the end of the file as well. 

The general syntax of a L AT E X file is as follows: 

\documentclass[options]{class} 

Further global commands and specifications 

\begin{document} 

Text mixed with additional commands of local effect 

\end{document} 

The possible options and classes that may appear in the \documentclass 

command are presented in Section 3.1.1. 

A minimal L AT E X file named hi.tex contains just the following lines: 

\documentclass{article} 


Hi! 


1.5.3 Extending L AT E X with packages 

Packages are a very important feature of L AT E X. These are extensions to 

the basic L AT E X commands that are written to files with names that end 

in.sty and are loaded with the command \usepackage in the preamble. 

Packages can be classified by their origin:

1.5. Basics of a L AT E X file 13 

core packages are an integral part of the L AT E X basic installation and are 

therefore fully standard; 

tools packages are a set written by members of the L AT E X3 Team, and 

should always be in the installation; 

graphics packages are a standardized set for including pictures generated 

by other programs, and for handling color; they are on the same 

level as the tools packages; 

A M S-L AT E X packages published by the American Mathematical Society, 

should be in any installation; 

contributed packages have been submitted by actual users; certain of 

these have established themselves as ‘essential’ to standard L AT E X 

usage, but all are useful. 

Only a limited number of these packages are described in this book, those 

that we consider indispensable. However, there is nothing to prevent 

the user from obtaining and incorporating any others that should prove 

beneficial for his or her purposes. 

There are over 1000 contributed packages on the included T E XLive CD. 

How can one begin to get an overview of what they offer? Graham Williams 

has compiled a list of brief descriptions which can be found online and 

on the T E XLive CD at 

\texmf\doc\html\catalogue\catalogue.html 

How to load packages into the L AT E X source file is explained in Section 

3.1.2. 

Documentation of contributed packages is somewhat haphazard, depending 

on how much the author has put into it. The preferred method 

for distributing packages is to integrate the documentation with the 

code into a single file with extension .dtx. A special program DocStrip 

(Section D.7.1) is used to extract the actual package file or files, while 

L AT E Xing the original .dtx file produces the instruction manual. Most 

ready-to-run installations will already have done all this for the user, 

with the resulting manuals stored as DVI or PDF files somewhere in 

\texmf\doc\latex\. . . . However, you might have to generate the documentation 

output yourself by processing the.dtx file, which should be 

found in \texmf\source\latex\. . . . (Section B.3 explains the organization 

of the T E X directory system.) 

Some package authors write their manuals as an extra .tex file, the 

output of which may or may not be prestored in DVI or PDF form. Others 

provide HTML files. And still others simply add the instructions as 

comments in the package file itself. (This illustrates some of the joys of 

an open system.)


1.6 T E X processing procedure 

Since L AT E X is a set of definitions for the T E X program, L AT E X processing itself 

is in fact T E X processing with the L AT E X format. What T E X does with this is 

the same as for any other of the many formats available (of which L AT E X is 

perhaps the most popular). All the typesetting work is done by T E X, while 

L AT E X handles the conversion from the logical markup to the typesetting 

commands. It also enables cross-referencing, running headlines, table 

of contents, literature citations and bibliography, indexing, and more. 

However, the processing of the source file to final output is T E X’s task, 

regardless of the format being used. 

1.6.1 In the good old days 

T E X arose over 20 years ago before there were such things as PCs, graphical 

displays, and before computers were infected with windows or mice. T E X 

and its support programs were invoked from a command line, not with 

a mouse click. This may sound very old fashioned, but it did guarantee 

portability to all computer types. 

The processing steps that were taken in those days still exist with 

today’s graphical interfaces, but are now executed more conveniently. 

One can still open a ‘command prompt window’ and run them from the 

command line. 

The first step is of course to use a text editor program to write the 

source file containing the actual text and markup. The rules for entering 

this source text are explained in Chapter 2. It goes into a text file, or what 

is often called an ‘ascii’ file containing only standard punctuation marks, 

numbers, unaccented letters, upper and lower case. In other words, the 

text is that which can be produced from a standard English typewriter. 

The name of the source file normally has the extension.tex; it is then 

processed by T E X to produce a new file with the same base name and the 

extension.dvi, for device independent file. This is a binary file (all codes 

possible, not a text file) containing precise instructions for the selection 

and placement of every symbol, a coded description of the final printed 

page. The command to invoke T E X with the source file hi.tex is 

tex &latex hi 

meaning run the T E X program with the format latex. Usually the installation 

has defined a shortcut named latex to do this, so 

latex hi 

should be sufficient. It is only necessary to specify the extension of the 

source file name if it is something other than.tex. 

During the processing, T E X writes information, warnings, error messages 

to the computer monitor, and to a transcript file with the extension 

.log. It is well worth inspecting this file when unexpected results appear.

1.6. T E X processing procedure 15 

The final step is to produce the printed pages from the DVI file. This 

requires another program, a driver, to generate the instructions specific 

to the given printer. For example, to produce a PostScript file, one runs 

dvips hi 

to obtain hi.ps from hi.dvi. And then one sends hi.ps to the PostScript 

printer with the regular command for that computer system. 

Previewing the DVI file on a computer monitor before printing was 

a later development, requiring high quality graphics displays. These 

programs are essentially special drivers that send the output directly to 

the monitor rather than to a printer or printer file. One very popular 

previewer is called with 

xdvi hi 

to view hi.dvi before committing it to paper. 

1.6.2 And today 

The various steps for L AT E X processing described above are still necessary 

today, and one can open up a command prompt window and carry them 

out just as before. However, there now exist intelligent editors with L AT E Xsavvy 

that not only assist writing the source text, but also will call the 

various programs, T E X, previewer, printer driver, BIBT E X, MakeIndex (these 

are explained later) with a mouse click. 

One such editor for Windows, available on the enclosed T E XLive CD in 

the support directory, is called WinShell, written by Ingo H. de Boer (www. 

winshell.de). Although free of charge, its author appreciates donations 

to offset his expenses. 

Another such editor and L AT E X interface is WinEdt by Aleksander Simonic 

(www.winedt.com). A sample window with the opening text of this 

chapter is shown in Figure 1.1. This program is available for a 30-day trial 

period, after which one must pay a nominal fee to obtain a licence. It is 

the editor that we ourselves use and we can highly recommend it. 

An alternative is LyX, a free, open source software for document processing 

in near WYSIWYG, acting as a front-end to L AT E X, where the user 

need not know anything about L AT E X. See its home page at www.lyx.org. 

It must be stressed that all the above are interfaces to an existing L AT E X 

installation. On the other hand, there are also commercial packages which 

include both the T E X/L AT E X installation and a graphics interface. These are 

listed in Section B.1.1. 

1.6.3 Alternative to T E X: pdfT E X 

As we mentioned earlier, it is permitted to use the T E X source code to 

generate something else, as long as it bears another name. One such


Figure 1.1: Sample display with the WinEdt editor for interfacing to L AT E X. 

modification is called pdfT E X, created by Hàn Thê ´ Thành. This program 

does everything T E X does, but it optionally writes its output directly to a 

PDF file, bypassing the DVI output of regular T E X. It therefore combines 

the T E X program with a DVI-to-PDF driver program. Normally this option 

is also the default. 

There are many advantages to producing PDF output directly this way, 

apart from saving a step. The PDF file is generated in exactly the same way 

as the DVI file with T E X, and can be viewed immediately with the Acrobat 

Reader or other PDF viewer. The results can be sent directly to a printer 

without going through the DVI-to-Printer program. It is also much easier 

to include the hypertext features of a true active PDF file, as we explain in 

Section 10.2.4. 

Adding the L AT E X macros to pdfT E X produces something one could call 

pdfL AT E X. This distinction is only meaningful for invoking the programplus-format 

to process the L AT E X source file. Except for some things that 

we note in Section 10.2.3, L AT E X commands are identical whether used with 

T E X or with pdfT E X. This makes the conversion extremely easy. 

The rest of this book deals essentially with L AT E X itself, regardless of 

what the end product is to be: paper, HTML, XML, or PDF.

2 

Text, Symbols, and 

Commands 

The text that is to be the input to a L AT E X processing run is written to a 

source file with a name ending in.tex, the file name extension. This file is 

prepared with a text editor, either one that handles straightforward plain 

text, or one that is configured to assist the writing and processing of L AT E X 

files. In either case, the contents of this file are plain ascii characters 

only, with no special symbols, no accented letters, preferably displayed in 

a fixed width typewriter font, with no frills like bold or italics, all in one 

size. All these aspects of true typesetting are produced afterwards by 

the T E X processing program with the help of markup commands inserted 

visibly into the actual text. It is therefore vital to know how commands 

are distinguished from text that is to be printed, and, of course, how they 

function. 

(However, for languages other than English, native keyboard input may 

indeed be used, as shown in Section 2.5.9.) 

2.1 Command names and arguments 

A command is an instruction to L AT E X to do something special, like print 

some symbol or text not available to the restricted character set used 

in the input file, or to change the current typeface or other formatting 

properties. There are three types of command names: 

• the single characters # $ & ˜ _ ˆ % { } all have special meanings 

that are explained later in this chapter; 

• the backslash character \ plus a single non-letter character; for 

example \$ to print the $ sign; all the special characters listed above 

have a corresponding two-character command to print them literally; 

• the backslash character \ plus a sequence of letters, ending with the 

first non-letter; for example, \large to switch to a larger typeface. 

17

18 Chapter 2. Text, Symbols, and Commands 

Command names are case sensitive, so \large, \Large and \LARGE 

are distinct commands. 

Many commands operate on some short piece of text, which then 

appears as an argument in curly braces following the command name. 

For example, \emph{stress} is given to print the word stress in an 

emphasized typeface (here italic) as stress. Such arguments are said to be 

mandatory because they must always be given. 

Some commands take optional arguments, which are normally employed 

to modify the effects of the command somehow. The optional 

arguments appear in square braces. 

In this book we present the general syntax of commands as 

\name[optional]{mandatory} 

where typewriter characters must be typed exactly as illustrated and 

italic text indicates something that must be substituted for. Optional 

arguments are put into square brackets [ ] and the mandatory ones into 

curly braces { }. A command may have several optional arguments, each 

one in its set of brackets in the specified sequence. If none of the optional 

arguments is used, the square brackets may be omitted. Any number of 

blanks, or even a single new line, may appear between the command name 

and the arguments, to improve legibility. 

Some commands have several mandatory arguments. Each one must 

be put into a { } pair and their sequence must be maintained as given in 

the command description. For example, 

\rule[lift]{width}{height} 

produces a black rectangle of size width and height, raised by an amount 

lift above the current baseline. A rectangle of width 10 mm and height 

3 mm is made with \rule{10mm}{3mm}. Since the optional argument lift 

is omitted, the rectangle is set on the baseline with no lifting, as . 

The arguments must appear in the order specified by the syntax and may 

not be interchanged. 

Some commands have a so-called *-form in addition to their normal 

appearance. A * is added to their name to modify their functionality 

somehow. For example, the \section command has a *-form \section* 

which, unlike the regular form, does not print an automatic section number. 

For each such command, the difference between the normal and 

*-form will be explained in the description of the individual commands. 

Command names consist only of letters, with the first non-letter indicating 

the end of the name. If there are optional or mandatory arguments 

following the command name, then it ends before the [ or { bracket, 

since these characters are not letters. Many commands, however, possess 

no arguments and are composed of only a name, such as the command 

\LaTeX to produce the L AT E X logo. If such a command is followed by

2.2. Environments 19 

a punctuation mark, such as comma or period, it is obvious where the 

command ends. If it is followed by a normal word, the blank between 

the command name and the next word is interpreted as the command 

terminator: The \LaTeX logo results in ‘The L AT E Xlogo’, that is, the blank 

was seen only as the end of the command and not as spacing between 

two words. This is a result of the special rules for blanks, described in 


In order to insert a space after a command that consists only of a name, 

either an empty structure {} or a space command (\ and blank) must be 

placed after the command. The proper way to produce ‘The L AT E X logo’ is 

to type either The \LaTeX{} logo or The \LaTeX\ logo. Alternatively, 

the command itself may be put into curly braces, as The {\TeX} logo, 

which also yields the right output with the inserted blank: ‘The T E X logo’. 

Incidentally, the L AT E X 2ε logo is produced with \LaTeXe. Can you see why 

this logo command cannot be named \LaTeX2e? 

2.2 Environments 

An environment is initiated with the command \begin{name} and is 

terminated by \end{name}. 

An environment has the effect that the text within it is treated differently 

according to the environment parameters. It is possible to alter 

(temporarily) certain processing features, such as indentation, line width, 

typeface, and much more. The changes apply only within the environment. 

For example, with the quote environment, 

previous text 

\begin{quote} 

text1 \small text2 \bfseries text3 

\end{quote} 

following text 

the left and right margins are increased relative to those of the previous 

and following texts. In the example, this applies to the three texts text1, 

text2, and text3. After text1 comes the command \small, which has the 

effect of setting the next text in a smaller typeface. After text2, there is an 

additional command \bfseries to switch to bold face type. Both these 

commands only remain in effect up to the \end{quote}. 

The three texts within the quote environment are indented on 

both sides relative to the previous and following texts. The 

text1 appears in the normal typeface, the same one as outside 

the environment. The text2 and text3 appear in a smaller typeface, 

and text3 furthermore appears in bold face. 

After the end of the quote environment, the subsequent text appears in 

the same typeface that was in effect beforehand.


Note that if the names of the environment in the \begin{..} \end{..} 

pair do not match, an error message will be issued on processing. 

Most declaration command names (see next section) may also be used 

as environment names. In this case the command name is used without 

the preceding \ character. For example, the command \em switches to 

an emphatic typeface, usually italic, and the corresponding environment 

\begin{em} will set all the text in italic until \end{em} is reached. 

A nameless environment can be simulated by a {...} pair. The effect 

of any command within it ends with the closing curly brace. 

The user can even create his or her own environments, as described in 

Section 8.4. 

2.3 Declarations 

A declaration is a command that changes the values or meanings of 

certain parameters or commands without printing any text. The effect of 

the declaration begins immediately and ends when another declaration 

of the same type is encountered. However, if the declaration occurs 

within an environment or a {...} pair, its scope extends only to the 

corresponding \end command, or to the closing brace }. The commands 

\bfseries and \small mentioned in the previous section are examples 

of such non-printing declarations that alter the current typeface. 

Some declarations have associated arguments, such as the command 

\setlength which assigns a value to a length parameter (see Sections 2.4 

and 8.2). 

Examples: 

{\bfseries This text appears in bold face} The \bfseries declaration 

changes the typeface: This text appears in bold face. The 

effect of this declaration ends with the closing brace }. 

\setlength{\parindent}{0.5cm} The paragraph indentation is set to 

0.5 cm. The effect of this declaration ends with the next encounter 

of the command \setlength{\parindent}, or at the latest with 

the \end command that terminates the current environment. 

\pagenumbering{roman} The page numbering is to be printed in Roman 

numerals. 

Some declarations, such as the last example, are global, that is, their 

effects are not limited to the current environment. The following declarations 

are of this nature, the meanings of which are given later: 

\newcounter \pagenumbering \newlength 

\setcounter \thispagestyle \newsavebox 

\addtocounter

2.4. Lengths 21 

Declarations made with these commands are effective right away and 

remain so until they are overridden by a new declaration of the same type. 

In the last example above, page numbering will be done in Roman numerals 

until countermanded by a new \pagenumbering{arabic} command. 

2.4 Lengths 

2.4.1 Fixed lengths 

Lengths consist of a decimal number with a possible sign in front (+ or 

-) followed by a mandatory dimensional unit. Permissible units and their 

abbreviated names are: 

cm centimeter, 

mm millimeter, 

in inch (1 in = 2.54 cm), 

pt point (1 in = 72.27 pt), 

bp big point (1 in = 72 bp), 

pc pica (1 pc = 12 pt), 

dd didôt point (1157 dd = 1238 pt), 

cc cicero (1 cc = 12 dd), 

em a font-specific size, the width of the capital M, 

ex another font-related size, the height of the letter x. 

Decimal numbers in T E X and L AT E X may be written in either the English 

or European manner, with a period or a comma: both 12.5cm and 12,5cm 

are permitted. 

Note that 0 is not a legitimate length since the unit specification is 

missing. To give a zero length it is necessary to add some unit, such as 

0pt or 0cm. 

Values are assigned to a length parameter by means of the L AT E X command 

\setlength, which is described in Section 8.2 along with other 

commands for dealing with lengths. Its syntax is: 

\setlength{\length name}{length spec} 

For example, the width of a line of text is specified by the parameter 

\textwidth, which is normally set to a default value depending on the 

class, paper type, and font size. To change the line width to be 12.5 cm, 

one would give: 

2.4.2 Rubber lengths 

\setlength{\textwidth}{12.5cm} 

Some parameters expect a rubber length. These are lengths that can be 

stretched or shrunk by a certain amount. The syntax for a rubber length 

is:


nominal value plus stretch value minus shrink value 

where the nominal value, stretch value, and shrink value are each a length. 

For example, 

\setlength{\parskip}{1ex plus0.5ex minus0.2ex} 

means: the extra line spacing between paragraphs, called \parskip, is to 

be the height of the x in the current font, but it may be increased to 1.5 

or reduced to 0.8 times that size. 

One special rubber length is \fill. This has the natural length of 

zero but can be stretched to any size. 

2.5 Special characters 

2.5.1 Spaces 

The space or blank character has some properties different from those 

of normal characters, some of which have already been mentioned in 

Section 2.1. During processing, blanks in the input text are replaced by 

rubber lengths (Section 2.4.2) in order to allow the line to fill up to the 

full line width. As a result, some peculiar effects can occur if one is not 

aware of the following rules: 

• one blank is the same as a thousand, only the first one counts; 

• blanks at the beginning of an input line are ignored; 

• blanks terminating a command name are removed; 

• the end of a line is treated as a blank. 

Some of the consequences of these rules are that there may be as many 

blanks as desired between words or at the beginning of a line (to make 

the input text more legible) and that a word may come right at the end of 

a line without the spacing between it and the next word disappearing. To 

force a space to appear where it would otherwise be ignored, one must 

give the command \ (a \ followed by a space character, made visible here 

by the symbol ). 

To ensure that certain words remain together on the same line, a protected 

space is inserted between them with the ˜ character (Section 2.7.1, 

page 28). Multiple protected spaces are all printed out, in contrast to 

normal spaces. 

Sometimes it is necessary to suppress the space that appears because 

of the new line. In this case, the last character in the line must be the 

comment character % (Section 4.11). 

Paragraphs are separated in the source text by blank lines. As for 

blank characters, one blank line is the same as a thousand.

2.5. Special characters 23 

Instead of a blank line, the command \par may also be used to indicate 

the end of a paragraph. 

2.5.2 Quotation marks 

The quotation marks found on the typewriter " are not used in book 

printing. Instead different characters are used at the beginning and end, 

such as ‘single quotes’ and “double quotes”. Single quotes are produced 

with ‘ and ’, while double quotes are made by typing the respective 

characters twice: ‘‘ for “ and ’’ for ”. Furthermore the typewriter 

character " will also generate the double closing quote ”. However, it 

should be avoided since it can lead to confusion. 

2.5.3 Hyphens and dashes 

In book printing, the character that appears on the typewriter as - comes 

in various lengths: -, –, —. The smallest of these, the hyphen, is used 

for compound words such as father-in-law and for word division at the 

end of a line; the middle-sized one, the en dash, is used in ranges of 

numbers, for example, pages 33–36; and the largest, the em dash, is used 

as punctuation—what is normally called the dash. These are generated 

by typing the hyphen character one, two, or three times, so that - yields 

-, while -- makes –, and --- produces —. A fourth type of dash is the 

minus sign −, which is entered in math mode as $-$ (Chapter 5). 

2.5.4 Printing command characters 

As mentioned in Section 2.1, the characters # $ ˜ _ ˆ % { } are interpreted 

as commands. To print them as text, one must give a command 

consisting of \ plus that character. 

$ = \$ & = \& % = \% # = \# = \_ { = \{ } = \} 

2.5.5 The special characters , , , , and 

These special characters do not exist on the computer keyboard. They 

can however be generated by special commands as follows: 

§ = \S † = \dag ‡ = \ddag = \P © = \copyright £ = \pounds 

The production of Greek letters and other mathematical symbols is 

described in Chapter 5.


2.5.6 Non-English letters 

Special letters that exist in languages other than English can also be 

generated with T E X. These are: 

œ={\oe} Œ={\OE} æ={\ae} Æ={\AE} ˚a={\aa} ˚A ={\AA} ¡ =!‘ 

ø ={\o} Ø ={\O} ł ={\l} Ł ={\L} ß={\ss} SS={\SS} ¿=?‘ 

˚Angstrøm may be written as {\AA}ngstr{\o}m while Karlstraße can be 

input as Karlstra{\ss}e. The ‘letter’ \SS is the upper case equivalent 

of \ss, used for automatic conversion between upper and lower case. 

However, see Section 2.5.9 for the possibility of entering such characters 

directly. 

2.5.7 Accents 

In non-English languages, there is a multiplicity of diacritical marks or 

accents, most of which can be printed with T E X: 

ò =\‘{o} ó=\’{o} ô=\ˆ{o} ö=\"{o} õ=\˜{o} 

ō =\={o} ˙o=\.{o} ŏ=\u{o} ǒ=\v{o} ő=\H{o} 

oo=\t{oo} o¸=\c{o} o . =\d{o} ō=\b{o} ˚o=\r{o} 

The o above is given merely as an example: any letter may be used. With 

i and j it should be pointed out that the dot must first be removed. This 

is carried out by prefixing these letters with a backslash: the commands 

\i and \j yield ı and j. In this way ĭ and ˝j are formed by typing \u{\i} 

and \H{\j}. 

The accent commands consisting of a non-letter may also be given 

without the curly braces: 

ò=\‘o ó=\’o ô=\ô ö=\"o õ=\õ ō=\=o ˙o=\.o 

The letter accent commands should always be used with the curly braces. 

2.5.8 The euro symbol 

Package: 

textcomp 

The euro symbol € (or e) is too new to be part of the original L AT E X, but it 

can be produced with the help of some additional fonts and contributed 

packages. Just which package you may use depends on your installation, 

and whether you have access to these additional fonts. 

The Text Companion fonts, described in Section G.4.4, do contain a 

euro symbol. Since these fonts should be part of every modern L AT E X 

installation, you should be able to use their euro symbol if all else fails. 

The package textcomp must be loaded in the preamble with 

\usepackage{textcomp}

2.5. Special characters 25 

which defines many commands including \texteuro to print the symbol 

€. Since the European Commission originally dictated that it should only 

be printed in a sans serif font, it is better to issue \textsf{\texteuro} to 

produce €. (The font selection commands are described in Section 4.1.4.) 

If you are going to use this very frequently, you might want to define a 

shortcut named \euro with 

\newcommand{\euro}{\textsf{\texteuro}} 

as described in Section 8.3 on defining commands. 

Package: A better solution is presented by the eurosym package by Henrik 

eurosym Theiling and the associated fonts that come with it, which bear the names 

feymr10, feybr10, and so on. This package defines the \euro command 

to print e, which changes automatically to bold e and slanted e as 

needed. 

Package: The europs package by Joern Clausen interfaces to the type 1 (Posteurops 

Script) euro fonts published by Adobe. For licensing reasons, these fonts 

may only be obtained from Adobe directly, even though free of charge 

(see Section B.2). This package provides the command \EUR for a symbol 

that varies with font family (Roman €25, sans serif €25, and typewriter 

€25) as well as for bold €25 and slanted €25. There is also a command 

\EURofc for the invariable symbol €. 

Package: Finally, the package eurosans by Walter Schmidt also addresses the 

eurosans Adobe euro fonts, again with the command \euro, with the same behavior 

as that of eurosym: always sans serif family, but changes with the other 

font attributes. 

The table below summarizes the above packages: 

Package Command Fonts Notes 

textcomp \texteuro Text Companion Non standard symbol 

eurosym \euro Eurosym Sans serif, variable 

europs \EUR PostScript Varies with font family 

\EURofc Invariable, official 

eurosans \euro PostScript Sans serif, variable 

So which package should one use? That really depends on the fonts 

available. Since the Adobe fonts can never be distributed with a T E X 

installation, they must be actively fetched and installed. However, it is 

worth doing so, because the European Commission has revised its initial 

directive and now allows the euro symbol to be typographically matched to 

the text, which is also standard practice in Europe today. This strengthens 

the case for the europs package and the \EUR command for €, at least for 

Roman fonts.


2.5.9 Typing special symbols directly 

The commands for producing the special characters and accented letters in 

! the previous sections may be suitable for typing isolated ‘foreign’ words, but 

become quite tedious for inputting large amounts of text making regular use of 

such characters. Most computer systems provide non-English keyboards with 

appropriate fonts for typing these national variants directly. Unfortunately, the 

coding of such extra symbols is by no means standard, depending very much on 

the computer system. 

For example, the text Gauß meets Ampère entered with an MS-DOS editor 

(code page 437 or 850) appears in a Windows application as Gauá meets AmpˇSre 

and on a Macintosh as Gau meets Ampäre. Since LAT EX is intended to run on 

all systems, it simply ignores all such extra character codes on the grounds that 

they are not properly defined. 

The inputenc package solves this problem. It not only informs LAT Package: 

EX which 

inputenc input coding scheme is being used, it also tells it what to do with the extra 

characters. One invokes it with 

\usepackage[code]{inputenc} 

where code is the name of the coding scheme to be used. The current list of 

allowed values for code (more are added with each L AT E X update) can be found in 

Table D.1 on page 462. For most users, the most interesting codes are: 

cp437 IBM code page 437 (DOS, North America) 

cp850 IBM code page 850 (DOS, Western Europe) 

applemac Macintosh encoding 

ansinew Windows ANSI encoding 

In short, you should select applemac for a Macintosh, and ansinew for Windows, 

and one of the others if you are working with DOS. 

Documents making use of this package are fully portable to other computer 

systems. The source text produced with a DOS editor may still look very strange 

to a human user reading it on a Macintosh, but when the Macintosh L AT E X processes 

it, the proper DOS interpretations will be applied so that the end result is what 

the author intended. 

See Section D.5 for more details. 

2.5.10 Ligatures 

In book printing, certain combinations of letters are not printed as individuals 

but as a single symbol, a so-called ligature. T E X processes the 

letter combinations ff, fi, fl, ffi, and ffl not as 

ff, fi, fl, ffi, ffl but rather as ff, fi, fl, ffi, ffl 

Ligatures may be broken, that is, forced to be printed as separate letters, 

by inserting \/ between the letters. This is sometimes desired for such 

words as shelfful (shelf\/ful), which looks rather strange when printed 

with the normal ff ligature, shelfful.

2.5.11 The date 

2.6. Exercises 27 

The current date can be placed at any point in the text with the command 

\today. The standard form for the date is the American style of month, 

day, year (for example, January 15, 2004). The British form (15th January 

2004) or the date in other languages can be generated with the help of the 

T E X commands \day, \month, and \year, which return the current values 

of these parameters as numbers. Examples of how such a new \today 

command may be made are shown on page 461 in Section D.4.2. 

It is in fact better to enter the date explicitly, rather than to rely 

on \today. Reprocessing a two-year-old L AT E X source file will yield a 

document with the processing date, not the date when the text was 

written. 

2.6 Exercises 

Exercise 2.1: This exercise tests the basic operations of running the L AT E X program 

with a short piece of text. A few simple commands are also included. Use a text 

editor to produce the following source text and store it in a file named exer.tex. 



Today (\today) the rate of exchange between the British 

pound and American dollar is \pounds 1 = \$1.58, an 

increase of 1\% over yesterday. 


Process this source file with L AT E X by clicking the appropriate icon, or by 

issuing latex exer in a command window. If the processing occurs without any 

error messages, the .dvi file exer.dvi will have been successfully created and 

may be viewed by a dvi previewer or sent to a printer. The final printed result 

should look as follows except that your current date will appear: 

Today (January 15, 2004) the rate of exchange between the British pound and 

American dollar is £1 = $1.58, an increase of 1% over yesterday. 

Note the following points about the commands used: 

no blank is necessary after \today because the ) suffices to terminate it; 

the blank after \pounds is optional and it is not printed in the output; 

the commands \$ and \% do not require blanks to terminate them; if blanks 

are given, they will be printed. 

Exercise 2.2: Take some text of about 3/4 of a page long out of a book or journal 

article and type it into a L AT E X source file. Pay attention that the paragraphs are 

separated by blank lines. Use the same set of commands as in Exercise 2.1, that 

is, put the text between the commands \begin{document}...\end{document} 

and repeat the procedures for obtaining the output.


Exercise 2.3: If you are likely to need the euro symbol in your work, try redoing 

Exercise 2.1 as follows: 


\usepackage{eurosym} 


Today (\today) the rate of exchange between the British 

pound and European euro is \pounds 1 = \euro1.46, an 

increase of 1\% over yesterday. 


If this fails, try one of the other packages described in Section 2.5.8, substituting 

\textsf{\texteuro} or \EUR for \euro as required. 

2.7 Fine-tuning text 

The subject of the section concerns pure typographical markup, and has 

nothing to do with the logical markup that we wish to stress in this book. 

Unfortunately, there are times when the author or editor does have to 

help the typesetting program to achieve good appearance. 

2.7.1 Word and character spacing 

The spacing between words and characters is normally set automatically 

by T E X, which not only makes use of the natural width of the characters 

but also takes into account alterations for certain character combinations. 

For example, an A followed by a V does not appear as AV but rather as AV; 

that is, they are moved together slightly for a more pleasing appearance. 

Interword spacing within one line is uniform, and is chosen so that the 

right and left ends match exactly with the side margins. This is called 

left and right justification. T E X also attempts to keep the word spacing for 

different lines as nearly the same as possible. 

Words that end with a punctuation mark are given extra spacing, 

depending on the character: following a period ‘.’ or exclamation mark ‘!’, 

there is more space than after a comma ‘,’. This corresponds to the rule in 

English typesetting that there should be extra spacing between sentences. 

In certain cases, the automatic procedures do not work properly, or it is 

desirable to override them, as described in the next sections. 

Sentence termination and periods 

T E X interprets a period following a lower case letter to be the end of a 

sentence where additional interword spacing is to be inserted. This leads 

to confusion with abbreviations such as i. e., Prof. Jones, or Phys. Rev., 

where the normal spacing is required. This can be achieved by using 

the characters ˜ or \ instead of the normal blank. (The character is

2.7. Fine-tuning text 29 

simply a symbol for the blank which is otherwise invisible.) Both these 

methods insert the normal interword spacing; in addition, ˜ is a protected 

space that prevents the line from being broken at this point. The above 

examples should be typed in as i.˜e., Prof.˜Jones, and Phys.\ Rev., 

producing i. e., Prof. Jones, and Phys. Rev. with the correct spacing and 

forcing the first two to be all on one line. In the third case, there is nothing 

wrong with putting Phys. and Rev. on different lines. 

A period following an upper case letter is not interpreted as the end 

of a sentence, but as an abbreviation. If it really is the end of a sentence, 

then it is necessary to add \@ before the period in order to achieve the 

extra spacing. For example, this sentence ends with NASA. It is typed in 

as This sentence ends with NASA\@. 

French spacing 

The additional interword spacing between sentences can be switched 

off with the command \frenchspacing, which remains in effect until 

countermanded with \nonfrenchspacing. In this case, the command \@ 

is ignored and may be omitted. This paragraph has been printed with 

\frenchspacing turned on, so that all word spacings within one line are 

the same. It corresponds to the normal rule for non-English typesetting. 

Character combinations “ ‘ and ’ ” 

A small spacing is produced with the command \,. This may be used, for 

example, to separate the double quotes “ and ” from the corresponding 

single quotes ‘ and ’ when they appear together. For example, the text 

‘‘\,‘Beginning’ and ‘End’\,’’ produces “ ‘Beginning’ and ‘End’ ”. 

Inserting arbitrary spacing 

Spacing of any desired size may be inserted into the text with the commands 

\hspace{space} 

\hspace*{space} 

where space is the length specification for the amount of spacing, for 

example 1.5cm or 3em. (Recall that one em is the width of the letter M in 

the current typeface.) 

This command puts blank space of width space at that point in the 

text where it appears. The standard form (without *) has no effect if it 

should come at the beginning of an output line, just as normal blanks are 

removed at the beginning of lines. The *-form, on the other hand, inserts 

the spacing no matter where it occurs. 

A blank before or after the command will also be included:


This is\hspace{1cm}1cm This is 1cm 

This is \hspace{1cm}1cm This is 1cm 

This is \hspace{1cm} 1cm This is 1cm 

The length specification may be negative, in which case the command 

works as a backspace for overprinting characters with other ones, or 

moving them closer together. For example, there is an energy unit in 

physics called electron volt, abbreviated ‘eV’, which looks much better if 

the two letters are nearer together, as ‘eV’, with e\hspace{-.12em}V. 

The command \hfill is an abbreviation for \hspace{\fill} (see 

Section 2.4.2). It inserts enough space at that point to force the text 

on either side to be pushed over to the left and right margins. With 

Left\hfill Right one produces 

Left Right 

Multiple occurrences of \hfill within one line will each insert the 

same amount of spacing so that the line becomes left and right justified. 

For example, the text Left\hfill Center\hfill Right generates 

Left Center Right 

If \hfill comes at the beginning of a line, the spacing is suppressed 

in accordance with the behavior of the standard form for \hspace. If 

a rubber space is really to be added at the beginning or end of a line, 

\hspace*{\fill} must be used instead. However, L AT E X also offers a 

number of commands and environments to simplify most such applications 

(see Section 4.2.2). 

A number of other fixed horizontal spacing commands are available: 

\quad and \qquad 

The command \quad inserts a horizontal space equal to the current type 

size, that is, 10 pt for a 10 pt typeface, whereas \qquad inserts twice as 

much. 

Inserting variable . . . . . . and sequences 

Two commands that work exactly the same way as \hfill are 

\dotfill and \hrulefill 

Instead of inserting empty space, these commands fill the gap with dots 

or a ruled line, as follows: 

Start \dotfill\ Finish\\ and 

Left \hrulefill\ Center \hrulefill\ Right\\ produce 

Start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Finish 

Left Center Right


Any combination of \hfill, \dotfill, and \hrulefill may be given 

on one line. If any of these commands appears more than once at one 

location, the corresponding filling will be printed that many more times 

than for a single occurrence. 

Departure \dotfill\dotfill\dotfill\ 8:30 \hfill\hfill 

Arrival \hrulefill\ 11:45\\ 

Departure . . . . . . . . . . . . . . . . . . . . . 8:30 Arrival 11:45 

2.7.2 Line breaking 

Breaking text into lines is done automatically in T E X and L AT E X. However, 

there are times when a line break must be forced or encouraged, or when 

a line break is to be suppressed. 

The command \\ 

A new line with or without additional line spacing can be achieved with 

the command \\. Its syntax is 

\\[space] 

\\*[space] 

The optional argument space is a length that specifies how much additional 

line spacing is to be put between the lines. If it is necessary to start 

a new page, the additional line spacing is not included and the new page 

begins with the next line of text. The *-form prevents a new page from 

occurring between the two lines. 

With \\*[10cm], the current line is ended and a vertical spacing of 

10 cm is inserted before the next line, which is forced to be on the same 

page as the current line. If a page break is necessary, it will be made 

before the current line, which is then positioned at the top of the new 

page together with the 10 cm vertical spacing and the next text line. 

The command \newline is identical to \\ without the option space. 

That is, a new line is started with no additional spacing and a page break 

is possible at that point. 

Both commands may be given only within a paragraph, and not between 

them where they would be meaningless. 

Further line-breaking commands 

The command \linebreak is used to encourage or force a line break at 

a certain point in the text. Its form is 

\linebreak[num]


where num is an optional argument, a number between 0 and 4 that 

specifies how important a line break is. The command recommends a line 

break, and the higher the number the stronger the recommendation. A 

value of 0 allows a break where it otherwise would not occur (like in the 

middle of a word), whereas 4 compels a line break, as does \linebreak 

without num. The difference between this command and \\ or \newline 

is that the current line will be fully justified, that is, interword spacing will 

be added so that the text fills the line completely. With \\ and \newline, 

however, the line is filled with empty space after the last word and the 

interword spacing remains normal. 

The opposite command 

\nolinebreak[num] 

discourages a line break at the given position, with num specifying the 

degree of discouragement. Again, \nolinebreak without a num argument 

has the same effect as \nolinebreak[4], that is, a line break is 

absolutely impossible here. 

Another way of forcing text to stay together on one line is with the command 

\mbox{text}. This is convenient for expressions such as ‘Voyager-1’ 

to stop a line break at the hyphen. 

2.7.3 Vertical spacing 

It is possible to add extra vertical spacing of amount space between 

particular paragraphs using the commands 

\vspace{space} 

\vspace*{space} 

The *-form will add the extra space even when a new page occurs, or 

when the command appears at the top of a new page. The standard form 

ignores the extra vertical spacing in these situations. 

If these commands are given within a paragraph, the extra space is 

inserted after the current line, which is right and left justified as usual. 

The space parameter may even be negative, in order to move the 

following text higher up the page than where it would normally be printed. 

The command \vfill is an abbreviation for \vspace{\fill} (see 

Section 2.4.2). This is the equivalent of \hfill for vertical spacing, 

inserting enough blank vertical space to make the top and bottom of the 

text match up exactly with the upper and lower margins. The comments 

on multiple occurrences of \hfill also apply to \vfill. If this command 

is given at the beginning of a page, it is ignored, just like the standard 

form of \vspace{\fill}. If a rubber space is to be put at the top of a 

page, the *-form \vspace*{\fill} must be used. 

Further commands for increasing the spacing between paragraphs are

\bigskip \medskip \smallskip 


which add vertical spacing depending on the font size declared in the 

document class. 

2.7.4 Page breaking 

Breaking text into pages occurs automatically in T E X and L AT E X, just as 

for line breaking. Here again, it may be necessary to interfere with the 

program’s notion of where a break should take place. 

Normal pages 

The commands 

\pagebreak[num] 

\nopagebreak[num] 

are the equivalents of \linebreak and \nolinebreak for page breaking. 

If \pagebreak appears between two paragraphs, a new page will be forced 

at that point. If it comes within a paragraph, the new page will be 

implemented after the current line is completed. This line will be right 

and left justified as usual. 

The command \nopagebreak has the opposite effect: between paragraphs, 

it prevents a page break from occurring there, and within a 

paragraph, it stops a page break that might take place at the end of the 

current line. 

Optional numbers between 0 and 4 express the degree of encouragement 

or discouragement for a page break. The analogy with the command 

\linebreak goes further: just as the line before the break is left and right 

justified with extra interword spacing, in the same way the page before 

the break is expanded with interline spacing to make it top and bottom 

justified. 

The proper command to end a page in the middle, fill it with blank 

spacing, and go on to a new page is 

\newpage 

which is equivalent to \newline with regard to page breaking. 

Pages with figures and tables 

If the text contains tables, pictures, or reserved space for figures, these 

are inserted at the location of the corresponding command, provided that 

there is enough room for them on the current page. If there is not enough 

space, the text continues and the figure or table is stored to be put on a 

following page. 

The command


\clearpage 

ends the current page like \newpage and in addition outputs all the 

pending figures and tables on one or more extra pages (Chapter 7). 

Two-column pages 

If the document class option twocolumn has been chosen, or the command 

\twocolumn is in effect, then the two commands \pagebreak and 

\newpage end the current column and begin a new one, treating columns 

as pages. On the other hand, \clearpage and \cleardoublepage (see 

below) terminate the current page, inserting an empty right column if 

necessary. 

Two-sided pages 

An additional page-breaking command is available when the document 

class option twoside has been selected: 

\cleardoublepage 

which functions exactly the same as \clearpage (the current page is 

terminated, all pending figures and tables are output) but in addition, the 

next text will be put on to an odd-numbered page. If necessary, an empty 

page with an even number is printed to achieve this. 

Controlling page breaks 

L AT E X provides the possibility of increasing the height of the current page 

slightly with commands 

\enlargethispage{size} 

\enlargethispage*{size} 

which add the length size to \textheight for this one page only. Sometimes 

the difference of a few points is all that is necessary to avoid a 

bad page break. The *-form of the command also shrinks any interline 

spacing as needed to maximize the amount of text on the page. 

2.8 Word division 

When a line is to be right and left justified, it often turns out that the 

break cannot be made between whole words without either shoving the 

text too close together or inserting huge gaps between the words. It is 

then necessary to split a word. This fundamental task is performed by 

T E X, the underlying basis of L AT E X, by means of a word-dividing algorithm

2.8. Word division 35 

that works (almost) perfectly for English text, which is more than can 

be said for most authors. Nevertheless, even it makes mistakes at times 

which need to be corrected by human intervention. 

If normal T E X/L AT E X is used for other languages, or if foreign words appear 

in English text, incorrect hyphenations are very likely to appear. (See 

Section 2.8.4 and Chapter 11 for more about L AT E X with other languages.) 

In these cases too, something must be done to override T E X’s hyphenation 

rules, as described below. 

2.8.1 Manual hyphenation 

The simplest way to correct a wrongly divided word is to include a \command 

at the right place within the word. The word manuscript, for 

example, will not be hyphenated at all, so if it causes problems with 

breaking a line, write it as man\-u\-script. This tells T E X to divide the 

word as necessary either as man-uscript or as manu-script, and to ignore 

its normal rules. 

The \- command merely makes hyphenation possible at the indicated 

locations; it does not force it. If the author absolutely insists in dividing a 

word at a certain point, say between the u and s in manuscript, he or she 

can type manu-\linebreak script to achieve this. However, this brute 

force method is not recommended, because the line break will always 

occur here even if the text is later changed. 

For English text, the spelling of a word remains the same when it is 

hyphenated, something that is not true in other languages. In traditional 

German spelling, for example, if ck is split, it becomes k-k. T E X allows 

such behavior with the general hyphenation command 

\discretionary{before}{after}{without} 

where before and after are the letters (with hyphen) that come on either 

side of the break if division takes place, and without is the normal text 

with no hyphenation. Thus Boris Becker’s name should be typed as 

Boris Be\discretionary{k-}{k}{ck}er 

something that one only wants to do in exceptional situations. Incidentally, 

the \- command is shorthand for \discretionary{-}{}{}. 

Note: In today’s new German spelling, ck is never split. This reformed 

spelling is still controversial however. 

2.8.2 Hyphenation list 

Words that are incorrectly hyphenated and that appear frequently within 

the document can be put into a list of exceptions in the preamble, to avoid 

laboriously inserting \- every time:


\hyphenation{list} 

The list consists of a set of words, separated by blanks or new lines, with 

the allowed division points indicated by hyphens. For example, 

\hyphenation{man-u-script com-pu-ter gym-na-sium 

coun-try-man re-sus-ci-tate ... } 

The list may contain only words with the normal letters a–z, with no 

special characters or accents. However, if the inputenc package is loaded, 

then the directly typed letters are also be included in the automatic 

hyphenation. 

2.8.3 Suppressing hyphenation 

Another means of avoiding bad word divisions is to turn hyphenation off, 

at least for a paragraph or two. Actually the command 

\begin{sloppypar} paragraph text \end{sloppypar} 

does not prevent word division, but does permit larger interword spacings 

without giving a warning message. This means that practically all lines are 

broken between words. It is also possible to put the command \sloppy 

in the preamble or in the current environment to reduce the number of 

word divisions in the whole document or within the environment scope. 

This is recommended when the line width is rather narrow. 

When the command \sloppy is in effect, it is possible to undo it 

temporarily and to turn hyphenation back on with the command \fussy. 

2.8.4 Word division with multilingual text 

Multiple hyphenation lists may be included in the T E X format, making it possible 

! to switch hyphenation schemes within one document, using the T E X command 

\language. This command may be used as part of language-specific adaptations 

to translate certain explicit English words in the output (such as ‘Contents’), to 

simplify accents or punctuation, and to alter the definition of the date command 

\today. This topic is treated in more detail in Chapter 11.

3 

3.1 Document class 

Document Layout and 

Organization 

The first command in the preamble of a L AT E X file determines the global 

processing format for the entire document. Its syntax is: 

\documentclass[options]{class} 

where some value of class must be given, while [options] may be omitted 

if default values are acceptable. 

The standard values of class, of which one and only one may be given, 

are: book, report, article, or letter. (The properties of the letter 

class are explained in Chapter 16.) The basic differences between these 

classes lie not only in the page layouts, but also in the organization. An 

article may contain parts, sections, subsections, and so on, while a report 

can also have chapters. A book also has chapters, but treats even and 

odd pages differently; also, it prints running heads on each page with the 

chapter and section titles. 

Other classes besides the standard ones exist, as contributions for 

specific journals, or for book projects. These will have their own set of 

options and additional commands, which should be described in separate 

documentation or instructions. However, since they are normally modifications 

of one of the standard classes, most of the following options 

apply to them too. 

3.1.1 Standard class options 

The options available allow various modifications to be made to the formatting. 

They can be grouped as follows. 

Selecting font size 

The basic font size is selected with one of the options 

37

38 Chapter 3. Document Layout and Organization 

10pt 11pt 12pt 

This is the size of the font in which the normal text in the document will 

be set. The default is 10pt, which means that this is the value assumed if 

no size option is specified. All other font size declarations are relative to 

this standard size, so that the section titles, footnotes, and so on will all 

change size automatically if a different basic font size is selected. 

Specifying paper size 

L AT E X calculates the text line width and lines per page according to the 

selected font size and paper mode. It also sets the margins so that the 

text is centered both horizontally and vertically. In order to do this, it 

needs to know which paper format is being used. This is specified by one 

of the following options: 

letterpaper (11 × 8.5 in) a4paper (29.7 × 21 cm) 

legalpaper (14 × 8.5 in) a5paper (21 × 14.8 cm) 

executivepaper (10.5 × 7.25 in) b5paper (25 × 17.6 cm) 

The default is letterpaper, American letter size paper, 11 × 8.5 in. 

Normally, the paper format is such that the longer dimension is the 

vertical one, the so-called portrait mode. With the option 

landscape 

the shorter dimension becomes the vertical one, the landscape mode. 

(One still has to ensure that the output is printed as landscape; see for 

example page 232.) 

Page formats 

The text on the page may be formatted into one or two columns with the 

options 

onecolumn twocolumn 

The default is onecolumn. In the case of the twocolumn option, the 

separation between the columns as well as the width of any rule between 

them may be specified by \columnsep and \columnseprule, described 

below. 

The even- and odd-numbered pages may be printed differently according 

to the options 

oneside twoside 

With oneside, all pages are printed the same; however, with twoside, the 

running heads are such that the page number appears on the right on odd

3.1. Document class 39 

pages and on the left on even pages. It does not force the printer to output 

double-sided. The idea is that when these are later printed back-to-back, 

the page numbers are always on the outside where they are more easily 

noticed. This is the default for the book class. For article and report, 

the default is oneside. 

With the book class, chapters normally start on a right-hand, odd-numbered 

page. The options 

openright openany 

control this feature: with openany a chapter always starts on the next 

page, but with openright, the default, a blank page may be inserted if 

necessary. 

Normally the title of a book or report will go on a separate page, while 

for an article, it is placed on the same page as the first text. With the 

options 

notitlepage titlepage 

this standard behavior may be overruled. See Sections 3.3.1 and 3.3.2. 

Further options 

The remaining standard options are: 

leqno Equation numbers in displayed formulas will appear on the left 

instead of the normal right side (Section 5.1). 

fleqn Displayed formulas will be set flush left instead of centered 

(Section 5.1). The amount of indentation may be set with the 

parameter \mathindent described below. 

openbib 

The format of bibliographies may be changed so that segments 

are set on new lines. By default, the texts for each entry are run 

together. 

draft If the L AT E X line-breaking mechanism does not function properly 

and text must stick out into the right margin, then this is marked 

with a thick black bar, to make it noticeable. 

final The opposite of draft, and the default. Lines of text that are too 

wide are not marked in any way. 

If multiple options are to be given, they are separated by commas, as 

for example, \documentclass[11pt,twoside,fleqn]{article}. The 

order of the options is unimportant. If two conflicting options are specified, 

say oneside and twoside, it is not obvious which one will be 

effective. That depends entirely on the definitions in the class file itself, 

so it would be best to avoid such situations.


Parameters associated with some options 

Some options make use of parameters that have been given certain default 

values: 

\mathindent 

specifies the indentation from the left margin for the equation 

numbers when fleqn is selected (Section 5.1); 

\columnsep 

specifies the space between the two columns for the twocolumn 

option (see Figure H.2 on page 603); 

\columnseprule 

determines the width of the vertical line between the two columns 

for the twocolumn option. The default is zero width, that is, no 

vertical rule (see Figure H.2). 

The standard values of these parameters may be changed with the 

L AT E X command \setlength. For example, to change \mathindent to 

2.5 cm, give 

\setlength{\mathindent}{2.5cm} 

These parameters may be assigned values either in the preamble or 

at any place in the document. Parameters in the preamble apply to the 

entire document, whereas those within the text are in effect until the next 

change or until the end of the environment in which they were made 

(Section 2.3). In the latter case, the previous values become effective once 

more. 

Exercise 3.1: Take your text file from Exercise 2.2 and change the initial command 

\documentclass{article} first to \documentclass[11pt]{article} 

and then to \documentclass[12pt]{article} and print the results of each 

L AT E X processing. Compare the line breaking of these outputs with that of Exercise 

2.2. 

Note: if there are some improper word divisions, you call tell L AT E X where the correct 

division should occur with the command \-, for example, man\-u\-script. 

(This is one of the few words that the T E X English word divider does not handle 

properly.) Additional means of modifying word division are given in Section 2.8. 

If there are warnings of the sort Overfull \hbox ... during the L AT E X processing, 

T E X was not able to break the lines cleanly. In the output, these lines will 

extend beyond the right margin. The usual cause is that T E X was not able to 

divide some word, either because it is indivisible or because T E X’s word division 

routines were not adequate. Here again a suggested hyphenation in the text can 

solve the problem. Other solutions will be given shortly. 

Exercise 3.2: Now employ \documentclass[twocolumn]{article} in your text 

file. If you now receive a number of warnings with Underfull \hbox ..., then 

these lines will indeed be left and right justified but will have too much empty

3.1. Document class 41 

space between the words. Check the output yourself to see whether the word 

spacing is acceptable. If not, try giving some hyphenation suggestions in the first 

words of the next line. 

Note: if you use the classes book or report instead of article in the preceding 

exercises, you will notice no difference in the outputs. These classes affect 

the subsequent structural elements of the document. Basically, you should use 

article for short articles (say 10–20 pages) and report for longer reports that 

are to be organized into chapters. The chapters always begin on a new page. The 

class book is available for producing books. 

3.1.2 Loading packages 

In Section 1.5.3 we explained how L AT E X can be extended by packages which 

are either part of the core installation or contributed by engaged users. 

How to write your own packages is described in Appendix D. 

A package is nothing more than a set of L AT E X (or T E X) commands 

stored in a file with the extension .sty, although there are some special 

commands that may only appear within them. To invoke a package, simply 

call 

\usepackage{package} 

in the preamble, where package is the root name of the file. More than one 

package may be loaded with one call to \usepackage. For example, two 

packages provided with standard L AT E X are stored in files makeidx.sty 

(Section 9.4.3) and ifthen.sty (Section 8.3.5). They may be read in 

together with 

\usepackage{makeidx,ifthen} 

A package may have options associated with it, which may be selected 

in the same way as for document classes: by including the option names 

within square braces. The general syntax is thus: 

\usepackage[opt1,opt2. . . ]{package1,package2,. . . } 

where all the listed options will be applied to all the selected packages. 

If any of the packages does not understand one of the options, a warning 

message is output to the monitor. 

3.1.3 Global and local options 

One interesting feature about options specified with the \documentclass com- 

! mand is that they also apply to any packages that follow. This means that 

if several packages all take the same option, it is only necessary to declare it 

once in \documentclass. For example, one might design a package to modify 

article for generating a local house style that might do different things for 

single or double column text; this package could make use of the class options 

onecolumn and twocolumn to achieve this. Or it could elaborate on the draft


option to produce double line spacing, as for a manuscript. Alternatively, several 

packages might have language-dependent features that could be activated 

with options like french or german; it is sufficient to list such options only in 

\documentclass to apply them to all packages. Such options are called global, 

for they are passed on to all subsequent packages automatically. 

Global options need not be limited to the standard class options listed in 

Section 3.1.1. A warning message is printed only if neither the class nor any 

of the packages understand one or more of them. By contrast, any options 

specified with \usepackage will be applied only to those packages listed in that 

one command; and it is applied to all of them. A warning is printed if one or 

more of those packages does not recognize any one of these local options. 

3.1.4 Class and package versions 

Class and package files normally have an internal version specification in the 

! form of their release date, as yyyy/mm/dd. If you wish to make use of some 

feature that you know was added on a certain date, you include that date in 

square brackets after the class or package name. An example of this is shown in 

Section 3.2.4 on page 46. 

The version date may also be added to the \documentclass command to 

ensure that the right version of the class file is being employed. The reason for 

doing this is to ensure that the source files are processed properly, say on other 

systems. 

3.2 Page style 

The basic page format is determined by the page style. With one exception, 

this command is normally given in the preamble. Its form is: 

\pagestyle{style} 

The mandatory argument style takes on one of the following values: 

plain The page head is empty, the foot contains the centered page 

number. This is the default for the article and report classes 

when no \pagestyle is given in the preamble. 

empty Both head and footlines are empty; no page numbers are printed. 

headings 

The head contains the page number as well as title information 

(chapter and section headings); the foot is empty. This is the 

default for book class. 

myheadings 

The same as headings except that the page titles in the head are 

not chosen automatically but rather are given explicitly by the 

commands \markright or \markboth (see below).

The command 

\thispagestyle{style} 

3.2. Page style 43 

functions exactly as \pagestyle except that it affects only the current 

page. For example, the page numbering may be suppressed for just the 

current page with the command \thispagestyle{empty}. It is only the 

printing of the page number that is suppressed; the next page will be 

numbered just as though the command had never been given. 

3.2.1 Heading declarations 

For the page styles headings and myheadings, the information appearing 

in the headline may be given with the declarations 

\markright{right head} 

\markboth{left head}{right head} 

The declaration \markboth is used with the document class option 

twoside, with even-numbered pages considered to be on the left and 

odd-numbered pages on the right. Furthermore, the page number is 

printed on the left side of the head for a left page and on the right side 

for a right page. 

For one-sided output, all pages are considered to be right-handed. In 

this case, the declaration \markright is appropriate. It may also be used 

with two-sided output to overwrite the right head given in \markboth. 

With the page style headings, the standard titles in the page headline 

are the chapter, section, or subsection headings, depending on the 

document and page style, according to the following scheme: 

Style Left Page Right Page 

book, report 

article 

one-sided — Chapter 

two-sided Chapter Section 

one-sided — Section 

two-sided Section Subsection 

If there are more than one \section or \subsection on a page, it is 

the heading of the last one that appears in the page head. 

3.2.2 Customized head and footlines 

Package: The standard page styles described in Section 3.2 select how the head and 

fancyhdr footlines are to appear, and what information they contain. This is a very 

limited choice, and the fancyhdr package by Piet van Oostrum offers the 

user considerable more flexibility. 

This package makes available an additional page style named fancy 

which the user can easily redefine. Head and footlines consist each of


three parts, left, center, right, each of which can be individually defined 

with 

\lhead{Left head} \chead{Center head} \rhead{Right head} 

\lfoot{Left foot} \cfoot{Center foot} \rfoot{Right foot} 

where the various texts may be explicit, or a command like \thepage to 

print the current page number. Both head and footlines may be decorated 

with a rule, the widths of which are set by commands \headrulewidth 

and \footrulewidth. By default, the fancy head and footlines are much 

the same as for the headings page style, but the head rule is set to 0.4 pt 

and the foot rule set to 0 (no rule). The rules may be redefined with, for 

example, 

\renewcommand{\footrulewidth}{0.4pt} 

to turn the foot rule on. 

The above defining commands are in fact specific examples of the 

more general commands \fancyhead and \fancyfoot, where 

\lhead{..} is \fancyhead[L]{..} 

\cfoot{..} is \fancyfoot[C]{..} 

and so on, with L C R standing for ‘left’, ‘center’, ‘right’. 

For two-sided output with the twoside option, one normally wants 

the left and right parts to alternate with page number. The easiest way to 

do this is with 

\fancyhead[LE,RO]{Text 1} \fancyhead[LO,RE]{Text 2} 

to put the same Text 1 in the left part of even pages, and right part of 

odd pages, and Text 2 for the other way round. With \fancyhead{}, all 

head line parts are set to blanks, something that should be done before 

resetting them explicitly. Similarly \fancyfoot{} sets all foot entries to 

blank. 

The default (two-sided) definitions for the fancy page style are 

\fancyhead[EL,OR]{\textsl{\rightmark}} 

\fancyhead[ER,OL]{\textsl{\leftmark}} 

where \rightmark and \leftmark contain the automatic texts for the 

headings page style generated by the \chapter, \section, \subsection 

commands (Section 3.3.3), while \textsl (Section 4.1.4) sets its argument 

in a slanted typeface. The user may also make use of these to redefine 

the head line with automatic texts. 

There is also the most general \fancyhf command taking optional 

arguments [H] and [F] to apply to head or footlines. Thus \fancyhf[HL] 

{..} is the same as \fancyhead[L]{..}. Clearly, \fancyhf{} resets 

everything.


In many classes, the first page of a chapter, or the very first page 

of the document, is switched to plain automatically. If the user wants 

to change this, he or she must redefine that page style. The fancyhdr 

package simplifies this task with 

\fancypagestyle{plain}{definitions} 

where definitions consist of \fancyhead, \fancyfoot, and/or rule redefinitions 

that are to apply to the revised plain style. In fact, any existing 

page style can be redefined in this way. 

3.2.3 Page numbering 

The declaration that specifies the style of the page numbering has the 

form 

\pagenumbering{num style} 

The allowed values of num style are: 

arabic for normal (Arabic) numerals, 

roman for lower case Roman numerals, 

Roman for upper case Roman numerals, 

alph for lower case letters, 

Alph for upper case letters. 

The standard value is arabic. This declaration resets the page counter 

to 1. In order to paginate the foreword of a document with Roman numerals 

and the rest with Arabic numbers beginning with page 1 for chapter 1, 

one must declare \pagenumbering{roman} at the start of the foreword 

and then reset the page numbering with \pagenumbering{arabic} immediately 

after the first \chapter command. (See Section 3.3.5 for the 

preferred method.) 

Pages may be numbered starting with a value different from 1 by giving 

the command 

\setcounter{page}{page num} 

where page num is the number to appear on the current page. 

Exercise 3.3: Expand your exercise text file so that it fills more than one page of 

output and include the following preamble: 


\pagestyle{myheadings} \markright{Exercises} 

\pagenumbering{Roman} 

\begin{document}


3.2.4 Paragraph formatting 

The following parameters affect the appearance of a paragraph and may 

be given new values with \setlength as explained in Section 8.2: 

\parskip 

The distance between paragraphs, expressed in units of ex so 

that it will automatically change with character font size. This 

should be a rubber length. 

\parindent 

The amount of indentation for the first line of a paragraph. 

\baselinestretch 

This is a number that magnifies the normal distance between 

baselines, the line on which the letters sit. This number is initially 

1, for standard line spacing. It may be changed to another 

number with 

\renewcommand{\baselinestretch}{factor} 

where factor is any decimal number, such as 1.5 for a 50% increase. 

This then applies to all font sizes. If this command is 

given outside the preamble, it does not come into effect until 

another font size has been selected (Section 4.1.2). 

These parameters may be set either in the preamble or anywhere in the 

text of the document. In the latter case, the changes remain in effect until 

the next change or until the end of the environment in which they were 

made (Section 2.3). 

To suppress indentation for one paragraph, or to force it where it 

would otherwise not occur, place 

\noindent or \indent 

at the beginning of the paragraph to be affected. 

Package: Normally, the first paragraph of a section is not indented, not even with 

indent- \indent. However, by including the package indentfirst one ensures 

first 

that all paragraphs are indented. 

By default, LAT Package: 

EX indicates paragraphs by indenting the first line. 

parskip An alternative is without indentation but with extra spacing between 

paragraphs. One could redefine \parindent and \parskip accordingly, 

or one could employ one of the oldest and simplest packages dating 

back to LAT EX 2.09 days: parskip, written by H. Partl. For consistency, this 

package also makes some changes in the parameters for lists (Section 4.3). 

One loads this package with 

\usepackage{parskip}


There is a recent update to the parskip package by Robin Fairbairns 

that includes the option parfill, given as 

\usepackage[parfill]{parskip} 

that avoids ugly-looking rectangular paragraphs by ensuring that there 

is always space at the end of the last line. This update is dated April 9, 

2001, so to be sure that this package version is loaded, add this date as 

\usepackage[parfill]{parskip}[2001/04/09] 

as explained in Section 3.1.4. A warning will be issued on processing if the 

actual version of parskip is earlier than this. There will also be warning 

about the unknown option parfill. 

Exercise 3.4: Add the following to the preamble of your exercise file: 

\usepackage{parskip} 

\renewcommand{\baselinestretch}{1.2} 

After processing this exercise, repeat it with another value for the parameter 

\baselinestretch, say 1.5, in order to get a feeling for how it works. Remove 

these lines from the exercise file afterwards. 

3.2.5 Page format 

Each page consists of a head, the body containing the actual text, and a 

foot. The selection of the page style determines what information is to be 

found in the head and footlines. 

L AT E X uses default values for the distances between the head, body, and 

foot, for the upper and left margins, and for the text line width and heights 

of the head, body, and foot. These formatting lengths are illustrated in 

Figure 3.1 on the next page. They may be changed by declaring new values 

for them, preferably in the preamble, with the command \setlength 

(Section 8.2). For example, give 

\setlength{\textwidth}{12.5cm} 

to make the text line width to be 12.5 cm. 

There is also a parameter \linewidth equal to the text line width in 

whatever environment one is currently in. This must never be changed, 

but is used when one needs to know this width. 

More detailed diagrams of the page formats for one- and two-column 

outputs are shown at the end of Appendix H in Figures H.1 and H.2. 

Package: You can examine your own page layout with the layout package from 

layout the tools collection. Simply issue the command \layout and a diagram 

similar to that in Figure 3.1 will be drawn at that point, together with a 

list of the current values of the layout parameters. Naturally, you would 

not do this in the middle of the final version of a document, but only as 

diagnostic check.


\oddsidemargin 

left margin for odd pages, 

\evensidemargin 

left margin for even pages, 

\topmargin 

upper margin to top of head, 

\headheight 

height of head, 

\headsep 

distance from the bottom of 

headline to top of body, 

\topskip 

distance from top of body to 

baseline of first line of text, 

\textheight, \textwidth 

height and width of main text, 

\footskip 

distance from bottom of body 

to bottom of foot, 

\paperwidth, \paperheight 

total width and height of paper 

as given by paper size option, 

including all margins. 

✻ 

✲✛ 

\paperheight 

❄ 

1 inch 

• ❄✻ 1 inch 

❄ 

Head 


✻❄ \headheight 

❄ \headsep 

✻ \topskip 

❄ First line of text 

Body 

✛ ✲ \textheight 

✻✛ 

\textwidth 

❄ 

✝ 

✲ 

\footskip 

Foot ❄ 

✞ 

\oddsidemargin 

✝\evensidemargin 

✛ \paperwidth 

Figure 3.1: Page layout parameters 

In order to calculate the page layout precisely, one must realize that L AT E X 

! measures all distances from a point one inch from the top of the paper and one 

inch from the left edge. Thus the total left margin is \oddsidemargin plus one 

inch. The L AT E X parameters \paperwidth and \paperheight, which include this 

extra inch, are given their values by the paper size option in the \documentclass 

command; they are used internally to calculate the margins so that the text is 

centered. The user may also take advantage of them for calculations. 

With document class book or with the option twoside, the bottom edge of 

! the body will always appear at exactly the same position on every page. In the 

other classes or options, it will vary slightly. In the first two cases, the constant 

bottom edge is produced by the internal command \flushbottom, whereas the 

varying bottom is produced by the command \raggedbottom. The user may 

apply these declarations to change the behavior of the bottom edge at any time, 

independent of document class and options. 

Exercise 3.5: You can change the page format of your text by altering the above 

parameters. Add the following to the preamble of your text: 

\setlength{\textwidth}{13cm} \setlength{\textheight}{20.5cm} 

The upper and left margins of your output will now seem too small. Select 

new values for \oddsidemargin and \topmargin to correct this. Note: do not 

✲


forget the 1 inch margin at the left and top from which additional margins are 

measured. You must take this into account when you select \oddsidemargin 

and \topmargin. 

Exercise 3.6: Expand your text so that the output requires more than two full 

pages with the reduced page format. Add \flushbottom to the preamble and 

check that the last line of all pages is at exactly the same location. 

Exercise 3.7: Remove the command \flushbottom and select the document 

class \documentclass[twoside]{article}. Now the last lines are at the same 

location without the \flushbottom command. On the other hand, the left margin 

of the odd pages probably does not agree with the right margin of the even pages. 

Adjust the value of \evensidemargin to correct this. 

3.2.6 Simplified page formatting 

Package: Getting the page layout to be exactly the way you want it can be very 

geometry tedious. Just centering the text on the page involves a complex set of 

settings that are not at all intuitive. The geometry package by Hideo 

Umeki offers considerable assistance. 

With this package, one can easily give values for some of the layout 

parameters, and the rest will be set automatically, taking into account 

the total paper size. For example, to set \textwidth to 15 cm and 

\textheight to 25 cm on A4 paper, one gives 

\usepackage{geometry} 

\geometry{a4paper,textwidth=15cm,textheight=25cm} 

which will also automatically set \oddsidemargin and \topmargin so 

that the text is centered horizontally and vertically, including the head 

and footlines. Or, one can set all the margins to be 1 inch on US letter 

paper with 

\geometry{letterpaper,margin=1in} 

Rather than using the \geometry command, one may also place the 

parameters as options to \usepackage, for example as 

\usepackage[a4paper,left=3cm,right=2cm]{geometry} 

to set the left and right margins to definite values, and \textwidth to 

what is left over. 

In general, all the parameters in Figure 3.1 may be specified by 

geometry by giving their names (without the backslash character). However, 

the package is far more powerful than that. Here we describe the 

essential features of version 2.3 from 2000/06/28. 

• The paper size is either inherited from the \documentclass option, 

given as a predefined option like a4paper, given explicitly as


paperwidth=pwidth and paperheight=pheight, or as papersize= 

{pwidth,pheight}. 

• By default, the other layout parameters are set so that \textwidth is 

80% of \paperwidth and \textheight + \headheight + \headsep 

+ \footskip is 90% of \paperheight, centered horizontally and 

vertically. 

• The text width and height may be set explicitly with width=width 

and height=height, or with body={width,height}. The margins are 

then set to center. Here height means total text height including 

head and footlines. 

• Margins may be explicitly given with left=lmarg and right=rmarg, 

or with hmargin={lmarg,rmarg}. Similarly with top=tmarg and 

bottom=bmarg, or vmargin={tmarg,bmarg}. If only one value of a 

pair is given, the other is set to that same value, unless the corresponding 

text height or width has been given explicitly. All margins 

may be set to a common value with margin=marg. 

Unlike L AT E X standard \oddsidemargin and \topmargin, geometry’s 

margins are measured from the edge of the paper, and not from a 

point 1 inch removed. 

• With nohead, nofoot, noheadfoot, one tells geometry not to include 

the corresponding head or footline in the height calculation. 

Thus with noheadfoot, the total height is identical to \textheight. 

• With includemp, the marginal note parameters \marginparwidth 

and \marginparsep (Section 4.10.5) are included in the total width 

calculation, which is then less than \textwidth. With reversemp, 

the marginal notes appear in the left margin. 

• The text width and height may be set to a fraction of the paper 

size with hscale=h and vscale=v, or scale={h,v}. With scale=s, 

both h and v are set to s. For example, \geometry{scale=0.8} 

sets width and height to 80% of \paperwidth and \paperheight 

respectively. 

• For two-sided output, add the option twoside. In this case, the values 

of the left and right margins will switch for even page numbers. 

In addition, a quantity of 20 pt is added to and subtracted from 

the left margin of odd and even pages, respectively. This value may 

be changed with twosideshift=shift, which also sets the twoside 

option automatically. 

• With the verbose option, the calculated values of all the layout 

parameters are printed to the monitor and to the transcript file.


As you see, the geometry package is an enormous aid to setting the 

page layout, working fairly intuitively, automatically setting values for the 

unspecified parameters in a way that is normally what one would want 

anyway. There are many more aspects explained in the accompanying 

documentation, but the above synopsis should cover most requirements. 

3.2.7 Single and double column pages 

The document class option twocolumn sets the entire document in two 

columns per page. The default is one column per page. Individual pages 

may be output in one or two columns with the declarations: 

\twocolumn[header text] 

Terminates the current page, starting a new one with two columns 

per page. The optional header text is written at the top of the 

page in one column with the width of the whole page. 

\onecolumn 

Terminates the current two-column page and continues with one 

column per page. 

The option twocolumn automatically changes certain page style parameters, 

! such as indentation, compared with the one-column format. This does not occur 

with the command \twocolumn. These additional changes must be made with 

the corresponding \setlength declarations if they are desired. If the bulk of 

the document is in two-column format, the class option is to be preferred. 

An additional page style parameter is \columnwidth, the width of one column 

! of text. For single column text, this is the same as \textwidth, but when 

twocolumn has been selected, L AT E X calculates it from the values of \textwidth 

and \columnsep. The author should never change this parameter, but he or she 

may make use of it, for example to draw a rule the width of a column of text. 

The length \linewidth is even more general, always containing the text line 

! width in the current environment, minipage, parbox. Within a single column, it 

is the same as \columnwidth. It too may never be changed. 

3.2.8 Multicolumn text 

Package: The commands \twocolumn and \onecolumn always start a new page, 

multicol and when two-column text is terminated, the two columns are of unequal 

length. These problems are solved with the multicol package in the 

tools collection, written by Frank Mittelbach, which also allows up to 10 

columns of text. Once this package has been loaded, one can switch the 

number of columns in the middle of a page with 

\begin{multicols}{num cols}[header text][pre space] 

Text set in num cols columns 

\end{multicols}


where the optional header text is written across all the columns before 

switching to multicolumns. 

Some automatic control for page breaking before and after switching 

to multicolumns is offered by the two lengths \premulticols and 

\postmulticols: if the remaining space on the current page is less 

than \premulticols, a new page is started before switching to multicolumns. 

Similarly, if at the end of the environment, there is less than 

\postmulticols on the page, a page break is inserted before continuing. 

The standard values of these lengths may be altered by the user with 

\setlength, or in the case of \premulticols may be overridden by the 

second optional argument pre space. 

The lengths \columnsep and \columnseprule that apply to twocolumn 

texts are also in effect for the multicols environment, to set 

the widths of the gap between columns and a possible separating rule, 

respectively. 

There is also a starred version \begin{multicols*}{num cols}. . . 

\end{multicols*} for which the columns on the last page are not balanced, 

with all the remaining space put into the final column. 

3.3 Parts of the document 

Every document is subdivided into chapters, sections, subsections, and 

so on. There can be an appendix at the end and at the beginning a title 

page, table of contents, an abstract, etc. L AT E X has a number of markup 

commands available to indicate these structures. In addition, sequential 

numbering and sub-numbering of headings take place automatically. Even 

a table of contents may be produced with a single command. 

The effects of some sectioning commands depend on the selected 

document class and not all commands are available in every class. 

3.3.1 Title page 

A title page can be produced either unformatted with the environment 

\begin{titlepage} Title page text \end{titlepage} 

or with the commands 

\title{Title text} 

\author{Author names and addresses} 

\date{Date text} 

\maketitle 

in the preprogrammed L AT E X style. 

In the standard L AT E X layout for the title page, all entries are centered 

on the lines in which they appear. If the title is too long, it will be broken

\title{% 

How to Write DVI Drivers} 

\author{% 

Helmut Kopka\thanks{Tel. 

[+49] 5556--401--451}\\ 

Max--Planck--Institut\\ 

f\"ur Aeronomie 

\and 

Phillip G. Hardy 

\thanks{Tel. 

[+1] 319--824--7134}\\ 

University\\of Iowa} 

\maketitle 

3.3. Parts of the document 53 

How to Write DVI Drivers 

Helmut Kopka 1 

Max–Planck–Institut 

für Aeronomie 

1 Tel. [+49] 5556–401–451 

2 Tel. [+1] 319–824–7134 

January 15, 2004 

Phillip G. Hardy 2 

University 

of Iowa 

Figure 3.2: Sample title page and the text that produced it 

up automatically. The author may select the break points himself with 

the \\ command, that is, by giving \title{...\\...\\...}. 

If there are several authors, their names may be separated with \and 

from one another, such as \author{G. Smith \and J. Jones}. These 

names will be printed next to each other in one line. The sequence 

\author{Author1\\Institute1\\Address1 

\and Author2\\Institute2\\Address2} 

separately centers the entries, one per line, in each of the sets Author1, 

Institute1, Address1 and Author2, Institute2, Address2 and places the two 

blocks of centered entries beside each other on the title page. 

Instead of printing the author names next to each other, one may 

position them on top of one another by replacing \and with the \\ 

command. In this case, the vertical spacing may be adjusted with an 

optional length specification [space] following the \\. 

If the command \date is omitted, the current date is printed automatically 

below the author entries on the title page. On the other hand, 

the command \date{Date text} puts the text Date text in place of the 

current date. Any desired text may be inserted here, including line break 

commands \\ for more than one line of centered text.



\thanks{Footnote text} 

may be given at any point in the title, author, or date text. This puts a 

marker at that point where the command appears and writes footnote text 

as a footnote on the title page. 

The title page is created using the entries in \title, \author, \date, 

and \thanks when the command 

\maketitle 

is issued. The title page itself does not possess a page number and 

the first page of the following document is number 1. (For book, the 

page numbering is controlled by the special commands in Section 3.3.5.) 

A separate title page is only produced for document classes book and 

report. For article, the command \maketitle creates a title heading 

on the first page using the centered entries from the \title, \author, 

and, if present, \date and \thanks declarations. If the document class 

option titlepage has been given, the title appears on a separate page 

even for the article class. 

An example of a title page in the standard L AT E X format is shown in 

Figure 3.2 on the previous page. Note that the current date appears 

automatically since the command \date is missing in the definition of 

the title page. This command may be used to put any desired text in place 

of the date. 

For the unformatted title page produced with the titlepage environment, 

the commands \title and \author are left out and the entire 

title page is designed according to the author’s specifications within the 

environment. To this end he or she may make use of all the structuring 

commands described in Chapter 4. In this case the printing of the title 

page is implemented at the end of the titlepage environment, so the 

command \maketitle is also left out. 

Exercise 3.8: Remove the declarations for changing the page format in Exercises 

3.5–3.7. Add to your exercise text a title heading with the title ‘Exercises’, 

your name as author, and your address, together with a date entry in the form 

‘place, date’. To do this, write the following commands after \begin{document}: 

\title{Exercises} \author{Your name\\Your address} 

\date{Your town, \today} \maketitle 

Make sure that you have selected document class article. After printing the 

document, change the document class command to 

\documentclass[titlepage]{article} 

to put the title information on to a title page instead of a title heading. Deactivate 

these commands by putting the comment character % at the beginning of each of 

the lines. In this way you avoid getting a title page in the following exercises but 

you can easily reactivate the commands simply by removing the % characters.

3.3.2 Abstract 

The abstract is produced with the command 


\begin{abstract} Text for the abstract \end{abstract} 

In document class report, the abstract appears on a separate page without 

a page number; in article, it comes after the title heading on the 

first page, unless the document class option titlepage has been selected, 

in which case it is also printed on a separate page. An abstract is not 

possible in document class book. 

3.3.3 Sections 

The following commands are available for producing automatic, sequential 

sectioning: 

\part \chapter \subsection \paragraph 

\section \subsubsection \subparagraph 

With the exception of \part, these commands form a sectioning hierarchy. 

In document classes book and report, the highest sectioning level 

is \chapter. The chapters are divided into sections using the \section 

command, which are further subdivided by means of \subsection, and 

so on. In document class article, the hierarchy begins with \section 

since \chapter is not available. 

The syntax of all these commands is 

\sec command[short title]{title} or 

\sec command*{title} 

In the first case, the section is given the next number in the sequence, 

which is then printed together with a heading using the text title. The text 

short title becomes the entry in the table of contents (Section 3.4) and the 

page head (provided that page style headings has been selected). If the 

optional short title is omitted, it is set equal to title; this is the normal 

situation unless title is too long to serve for the other entries. 

In the second (*-form) case, no section number is printed and no entry 

in the table of contents is made (however, see Section 3.4.3). 

The size of the title heading and the depth of the numbering depend 

on the position of the sectioning command within the hierarchy. For 

document class article, the \section command generates a single 

number (say 7), the \subsection command a double number with a 

period between the two parts (say 7.3), and so on. 

In document classes book and report, the chapter headings are given a 

single number with the \chapter command, the \section command creates 

the double number, and so on. Furthermore, the command \chapter 

always starts a new page and prints Chapter n over the chapter title, where


n is the current chapter number. At this point in the present book, we are 

in Chapter 3, Section 3.3, Subsection 3.3.3. 

For each sectioning command there is an internal counter that is 

incremented by one every time that command is called, and reset to zero 

on every call to the next higher sectioning command. These counters are 

not altered by the *-forms, a fact that can lead to difficulties if standard 

and *-forms of the commands are mixed such that the *-forms are higher 

in the hierarchy than the standard forms. There are no problems, however, 

if the *-forms are always lower than the standard forms. The sequence 

\section ... \subsection ... \subsubsection* ... 

numbers the headings for \section and \subsection while leaving the 

headings for \subsubsection without any numbering. 

The sectioning command \part is a special case and does not affect 

the numbering of the other commands. 

The automatic numbering of sections means that the numbers might 

not necessarily be known at the time of writing. The author may be writing 

them out of their final order, or might later introduce new sections or even 

remove some. If he or she wants to refer to a section number in the text, 

some mechanism other than typing the number explicitly will be needed. 

The L AT E X cross-reference system, described in detail in Section 9.2.1, 

accomplishes this task with the two basic commands 

\label{name} \ref{name} 

the first of which assigns a keyword name to the section number, while the 

second may be used as reference in the text for printing that number. The 

keyword name may be any combination of letters, numbers, or symbols. 

For example, in this book the command \label{sec:xref} has been 

typed in at the start of Section 9.2.1, so that this sentence contains the 

input text at the start of Section \ref{sec:xref}. 

A second referencing command is \pageref for printing the page 

number where the corresponding \label is defined. 

The referencing commands may be used in many other situations for 

labeling items that are numbered automatically, such as figures, tables, 

equations. 

Every sectioning command is assigned a level number such that \section 

! is always level 1, \subsection level 2, . . . \subparagraph level 5. In document 

class article, \part is level 0 while in book and report classes, \part is level 

−1 and \chapter becomes level 0. Section numbering is carried out down to 

the level given by the number secnumdepth. This limit is set to 2 for book 

and report, and to 3 for article. This means that for book and report, the 

section numbering extends only to the level of \subsection and for article to 

\subsubsection. 

In order to extend (or reduce) the level of the section numbering, it is necessary 

to change the value of secnumdepth. This is done with the command 

\setcounter{secnumdepth}{num}


(The command \setcounter is explained in Section 8.1.) In article, num may 

take on values from 0 to 5, and in book and report from −1 to 5. 

It is possible to change the initial value of a sectioning command within a 

document with the command 

\setcounter{sec name}{num} 

where sec name is the name of the sectioning command without the preceding 

\ character. This procedure may be useful when individual sections are to be 

processed by L AT E X as single files. For example, 

\setcounter{chapter}{2} 

sets the \chapter counter to 2. The counter will be incremented on the next call 

to \chapter which then produces Chapter 3. 

3.3.4 Appendix 

An appendix is introduced with the declaration 

\appendix 

It has the effect of resetting the section counter for article and the 

chapter counter for book and report and changing the form of the numbering 

for these sectioning commands from numerals to capital letters A, 

B, . . . . Furthermore, the word ‘Chapter’ is replaced by ‘Appendix’ so that 

subsequent chapter headings are preceded by ‘Appendix A’, ‘Appendix B’, 

etc. The numbering of lower sectioning commands contains the letter in 

place of the chapter number, for example A.2.1. 

3.3.5 Book structure 

To simplify the structuring of a book, the commands 

\frontmatter 

preface, table of contents 

\mainmatter 

main body of text 

\backmatter 

bibliography, index, colophon 

are provided in the book class. The \frontmatter command switches 

page numbering to Roman numerals and suppresses the numbering of 

chapters; \mainmatter resets the page numbering to 1 with Arabic numbers 

and reactivates the chapter numbering; this is once again turned off 

with \backmatter. 

Exercise 3.9: Insert at the beginning of your exercise text the command 

\section{Title A} and at some appropriate place near the middle \section{Title 

B}. Select some suitable text for Title A and Title B. Insert at appropriate places


some \subsection commands with reasonable subtitles. Remove the commands 

included from Exercise 3.3: 

\pagestyle{myheadings} \markright{Exercises} 

\pagenumbering{Roman} 

and print the results. 

Exercise 3.10: Include the additional command \chapter{Chapter title} with 

your own appropriate Chapter title before your first \section command. Change 

the document class command to \documentclass[twoside]{report} and call 

the page style command \pagestyle{headings} in the preamble. Note the 

twofold effect of the sectioning commands in the headings and in the page 

headlines. Compare the results with the table in Section 3.2.1. 

Exercise 3.11: Change the chapter command to 

\chapter[Short form]{Chapter title} 

by putting an abbreviated version of Chapter title for Short form. Now the page 

head contains the shortened title where the full chapter title previously appeared. 

3.4 Table of contents 

3.4.1 Automatic entries 

L AT E X can prepare and print a table of contents automatically for the 

whole document. It will contain the section numbers and corresponding 

headings as given in the standard form of the sectioning commands, 

together with the page numbers on which they begin. The sectioning 

depth to which entries are made in the table of contents can be set in the 

preamble with the command 

\setcounter{tocdepth}{num} 

The value num has exactly the same meaning and effect as it does for the 

counter secnumdepth described above, by which the maximum level of 

automatic subsectioning is fixed. By default, the depth to which entries 

are included in the table of contents is the same as the standard level 

to which automatic sectioning is done: that is, to level \subsection for 

book and report and to level \subsubsection for article. 

3.4.2 Printing the table of contents 

The table of contents is generated and printed with the command 

\tableofcontents

3.4. Table of contents 59 

given at the location where the table of contents is to appear, which is 

normally after the title page and abstract. 

This leads to a paradox, for the information in the table of contents is 

to be printed near the beginning of the document, information that cannot 

be known until the end. L AT E X solves this problem as follows: the first 

time the document is processed, no table of contents can be included but 

instead L AT E X opens a new file with the same name as the source file but 

with the extension .toc; the entries for the table of contents are written 

to this file during the rest of the processing. 

The next time L AT E X is run on this document, the \tableofcontents 

command causes the .toc file to be read and the table of contents is 

printed. As the processing continues, the .toc file is updated in case 

there have been major changes since the previous run. This means that 

the table of contents that is printed is always the one corresponding to the 

previous version of the document. For this reason, it may be necessary to 

run L AT E X more than once on the final version. 

3.4.3 Additional entries 

The *-form sectioning commands are not entered automatically in the table of 

! contents. To insert them, or any other additional entry, the commands 

\addcontentsline{toc}{sec name}{entry text} 

\addtocontents{toc}{entry text} 

may be used. 

With the first command, the entries will conform to the format of the table of 

contents, whereby section headings are indented more than those for chapter 

but less than those for subsection. This is determined by the value of the 

argument sec name, which is the same as one of the sectioning commands 

without the \ character (for example, section). The entry text is inserted in the 

table of contents along with the page number. This command is most useful to 

enter unnumbered section headings into the table of contents. For example, 

\section*{Author addresses} 

\addcontentsline{toc}{section}{Author addresses} 

The \addtocontents command puts any desired command or text into the 

.toc file. This could be a formatting command, say \newpage, which takes effect 

when the table of contents is printed. 

3.4.4 Other lists 

In addition to the table of contents, lists of figures and tables can also be 

! generated and printed automatically by L AT E X. The commands to produce these 

lists are 

\listoffigures reads and/or produces file.lof 

\listoftables reads and/or produces file.lot


The entries in these lists are made automatically by the \caption command 

in the figure and table environments (see Section 7.4). Additional entries are 

made with the same commands as for the table of contents, the general form of 

which is 

\addcontentsline{file}{format}{entry} 

\addtocontents{file}{entry} 

where file stands for one of the three types toc (table of contents), lof (list of 

figures), or lot (list of tables). The argument format is one of the sectioning 

commands for the table of contents, as described above, or figure for the list 

of figures, or table for the list of tables. The argument entry stands for the text 

that is to be inserted into the appropriate file. 

Exercise 3.12: In your exercise file, insert after the deactivated title page commands 

\pagenumbering{roman} 

\tableofcontents \newpage 

\pagenumbering{arabic} 

Process your exercise file twice with L AT E X and print out the second results. 

Deactivate the above commands with % before doing the next run.

4 

Displayed Text 

There are a variety of ways to display or emphasize the text: changing 

font style or font size, centering, indentation, marking the paragraphs, 

and so on. L AT E X supplies us with commands for the most common forms 

of display. 

Many parts of this chapter violate the concept of logical markup, 

especially those dealing with selection of fonts. The author should not 

attempt to decorate the document with arbitrary switches of font size and 

style, but should pack his or her source text into a structure that indicates 

its purpose. The exercises in this book are an example of this. Rather 

than starting each one with the word ‘Exercise’ in bold face followed by 

an explicit number, and then shifting to a slanted font, we defined an 

exercise environment to do all that automatically. This not only ensures 

consistency, it also allows a change of style to be easily implemented, 

simply by redefining the environment. This is where the typographical 

font commands come into play. They should not appear in the main text 

at all, but rather in the preamble as part of the definitions of environments 

and commands. 

On the other hand, many of the topics in this chapter really do involve 

logical markup, such as the verse, quote, and quotation environments, 

lists, bibliographies, theorems, and tables. 

4.1 Changing font 

In typography, a set of letters, numbers, and characters of a certain 

size and appearance is called a font. The standard font in L AT E X for the 

main body of text is an upright, Roman one of medium weight, in the 

size specified in the \documentclass statement at the start. The three 

possible basic sizes are 10, 11, and 12 pt, depending on the size options 

10pt (default), 11pt, and 12pt. (Recall, there are 72.27 points per inch 

or about 28.45 pt per cm.) The parenthesis characters ( ) extend the full 

height and depth of the font size. 

61

62 Chapter 4. Displayed Text 

The differences in the visual appearance of the three standard sizes 

are greater than would be expected from the ratios of the numbers: 

4.1.1 Emphasis 

This is an example of the 10 pt font. ( ) 

And this is the 11 pt font for comparison. ( ) 

And finally this is a sample of 12 pt font. ( ) 

The usual way to emphasize text in a typewritten manuscript is by 

underlining. The typesetter will transform underlined text into italics 

for the printed version. Switching from standard to emphasized text is 

carried out in L AT E X with the command \emph or the declaration \em. 

The \em declaration functions just as the other font declarations described 

below: the change of font remains in effect until negated by 

another appropriate declaration (which can be \em itself), or until the end 

of the current environment (Section 2.2). An environment may also be 

created with a pair of curly braces {...}. The command \emph, on the 

other hand, operates only on the text in the following argument. This is 

easiest way to emphasize short pieces of text, as for example: 

This is the easiest way to \emph{emphasize} short ... 

The \em declaration is more appropriate for longer text that is enclosed 

in an environment, named or nameless. 

...enclosed in an environment, {\em named or nameless.} 

Note carefully the difference between the declaration that remains in 

effect until the local environment is ended with the closing curly brace, 

and the command that operates on an argument enclosed in curly braces. 

Another more subtle difference is that the command \emph automatically 

inserts extra spacing at the end if necessary, the so-called italic correction, 

to improve the appearance at the interface between sloping and upright 

fonts. 

Both the declaration and the command switch to an emphasizing font. 

That means, if the current font is upright it switches to italics, whereas if 

the text is already slanted, an upright font is selected. 

Nested emphasis is possible and is simple to understand: 

The \emph{first}, second, and \emph{third font switch} 

The {\em first, {\em second, and {\em third font switch}}} 

both produce ‘The first, second, and third font switch’. 

4.1.2 Choice of font size 

The following declarations are available in L AT E X for changing the font size:

\tiny smallest 

\scriptsize very small 

\footnotesize smaller 

\small small 

\normalsize normal 

\large large 

4.1. Changing font 63 

\Large larger 

\LARGE even larger 

\huge still larger 

\Huge largest 

all of which are relative to the standard size selected in the document 

class option. In this book, the standard size is 10 pt, which is then the 

size selected with \normalsize. 

The font size declarations behave as all other declarations: they make 

an immediate change that remains in effect until counteracted by another 

size declaration, or until the current environment comes to an end. If 

issued within curly braces {..}, the effect of the declaration extends only 

to the closing brace, as in a nameless environment: 

normal {\large large \Large larger} normal again 

normal large larger normal again 

Changing the font size with one of the above commands also automatically 

! changes the interline spacing. For every font size, there is a corresponding natural 

line spacing \baselineskip. This may be altered at any time. If the natural 

line spacing is 12 pt, the command \setlength{\baselineskip}{15pt} will 

increase it to 15 pt. 

The value of \baselineskip that is effective at the end of the paragraph 

is used to make up the whole paragraph. This means that if there are several 

changes to \baselineskip within a paragraph, only the last value given will be 

taken into account. 

With every change in font size, \baselineskip is reset to its natural value 

for that size. Any previous setting with \setlength will be nullified. 

In order to create a change in the line spacing that is valid for all font sizes, 

one must make use of the factor \baselinestretch, which has a normal value 

of 1. The true interline spacing is really 

\baselinestretch×\baselineskip 

which maintains the same relative spacing for all font sizes. The user may change 

this spacing at any time with: 


where factor is any decimal number. A value of 1.5 increases the interline spacing 

(baseline to baseline) by 50% over its natural size for all font sizes. 

The new value of \baselinestretch does not take effect until the next 

change in font size. In order to implement a new value in the current font size, it 

is necessary to switch to another size and back again immediately. If the present 

font size is \normalsize, the sequence 

\small\normalsize 

will do the trick. Any size command may be used in place of \small.


4.1.3 Font attributes 

The size of a font is only one of several attributes that may be used 

to describe it. With the New Font Selection Scheme (NFSS), which was 

introduced as part of L AT E X 2ε, it is possible to select fonts strictly by 

these attributes, as described in Appendix A. However, for normal usage, 

there are some declarations and corresponding commands to simplify 

this procedure. 

For the Computer Modern fonts provided with T E X and L AT E X, the following 

attributes and values exist: 

Family: for the general overall style. Traditional typographical families 

have names like Baskerville, Bodoni, Times Roman, Helvetica, and 

so on. The standard L AT E X installation provides three families with 

declarations 

\rmfamily to switch (back) to a Roman font; 

\ttfamily to switch to a typewriter font; 

\sffamily to select a sans serif font. 

Shape: for the form of the font. The shape declarations available with 

the standard installation are 

\upshape to switch (back) to an upright font; 

\itshape to select an italic shape; 

\slshape to choose a font that is slanted; 

\scshape to switch to Caps and Small Caps. 

Series: for the width and/or weight (boldness) of the font. The declarations 

possible are 

\mdseries to switch (back) to medium weight; 

\bfseries to select a bold face font. 

These do not exhaust all the possible attribute settings, but they do cover 

the most standard ones, especially for the Computer Modern fonts. For 

other fonts, especially PostScript ones, additional attribute values exist. 

See Section A.1 for more details. 

These declarations are used just like any others, normally enclosed 

in a pair of curly braces {...}, such as {\scshape Romeo and Juliet} 

producing Romeo and Juliet. For longer sections of text, an environment 

is preferable: 

\begin{font style} . . . text in new font . . . \end{font style} 

This keeps better track of the beginning and end of the switch-over. For 

font style, any of the above font commands may be used, leaving off the 

initial \ character. 

Since changing any one attribute leaves the others as they were, all 

possible combinations may be obtained. (However, this does not mean

4.1. Changing font 65 

that a font exists for each possible combination; if not, a substitution 

will be made.) If we select first a bold series with \bfseries, and then a 

slanted shape with \slshape, we obtain a bold, slanted font. 

normal and {\bfseries bold and 

{\slshape slanted} and back} again. 

produces: normal and bold and slanted and back again. 

Finally, the declaration \normalfont resets all the attributes (except 

size) back to their defaults: Roman, upright, medium weight. It is often 

useful to issue this command just to be sure of the font in effect. 

4.1.4 Font commands 

For each of the font declarations listed above, there is a corresponding 

font command that sets its argument in a font with the specified attribute. 

Family: \textrm{text} \texttt{text} \textsf{text} 

Shape: \textup{text} \textit{text} \textsl{text} 

\textsc{text} 

Series: \textmd{text} \textbf{text} 

Default: \textnormal{text} 

Emphasis: \emph{text} 

Note that the \emph command is included here, corresponding to the 

declaration \em. The argument of \textnormal is set in the standard 

font selected with \normalfont. 

The use of such commands to change the font for short pieces of text, 

or single words, is much more logical than placing a declaration inside an 

implied environment. The previous example now becomes 

normal and \textbf{bold and \textsl{slanted} 

and back} again. 

to make: normal and bold and slanted and back again. 

As for the \emph command, these font commands automatically add 

any necessary italic correction between upright and slanted/italic fonts. 

The old two-letter T E X declarations such as \bf and \tt, which were 

part of L AT E X 2.09, are still available but are now considered obsolete and 

should be avoided. They are listed for reference in Appendix F. 

4.1.5 Additional fonts 

It is likely that your computing center or your T E X installation has even more 

! fonts and sizes than those listed above. If so, they may be made available for 

use within a L AT E X document either by referring to them by name, or by their 

attributes, if they have been set up for NFSS. 

To load a new font explicitly by name, the command


\newfont{\fnt}{name scaled factor} or 

\newfont{\fnt}{name at size} 

is given, which assigns the font to the new command named \fnt. In the first case, 

factor is a number 1000 times the scaling factor that is to be used to magnify 

or reduce the font from its basic or design size. In the second case, the font 

is scaled to be of the size specified. To install a slanted, sans serif font of size 

20.74 pt, as \sss, we load cmssi17 at 20.74pt with 

\newfont{\sss}{cmssi17 at 20.74pt} 

Now the declaration \sss switches directly to this font but without altering the 

baseline separation. 

Alternatively, the new font declaration can be made by attributes with (see 

Section A.3.2) 

\DeclareFixedFont{\sss}{OT1}{cmss}{m}{sl}{20.74} 

Indeed, if one wants to use the current encoding and \sffamily without knowing 

what they are, or without worrying so precisely what size must be stated, it is 

also possible to give 

\DeclareFixedFont{\sss}{\encodingdefault}{\sfdefault} 

{m}{sl}{20} 

(The defaults are explained in Section A.3.1.) 

4.1.6 Character sets and symbols 

The individual character sets are each stored in their own files. The names of the 

! 75 standard T E X fonts are listed in Section G.2.2, where many of them are also 

printed out. 

Each symbol within a character set is addressed by means of a number 

between 0 and 127 (or 255). The command 

\symbol{num} 

will produce that symbol with the internal identification number num in the 

current font. The symbol ¿ in the present font has the internal number 62 

and can be printed with the command \symbol{62}. The identification number 

may also be given as an octal (prefix ’) or hexadecimal (prefix ") number. Thus 

the symbol commands \symbol{28}, \symbol{’34}, and \symbol{"1C} are all 

identical, producing ‘ø’. 

The \symbol command may also be used to generate symbols for which 

no other command has been defined: for example, {\ttfamily\symbol{’40} 

\symbol{’42} \symbol{’134}} produces " \ in typewriter font. 

Section G.2.2 presents the assignments of the identification numbers with 

the characters for the different symbol families.

4.2 Centering and indenting 

4.2.1 Centered text 

The environment 

4.2. Centering and indenting 67 

\begin{center} line 1 \\ line 2 \\ . . . line n \end{center} 

centers the sections of text that are separated by the \\ command. (An 

optional additional line spacing may be inserted with \\[len].) If the text 

is too long for one line, it is split over several lines using uniform word 

spacing, filling the whole line width as best it can, except for the last line. 

Word division does not occur. 

Within an environment, the command \centering may be used to 

center the following text, again with \\ as the line divider. The effect of 

this declaration lasts until the end of that environment. 

A single line may be centered by typing its text as the argument of the 

T E X command \centerline{text}. 

4.2.2 One-sided justification 

The environments 

\begin{flushleft} line 1 \\ line 2 \\ . . . line 2 \end{flushleft} 

\begin{flushright} line 1 \\ line 2 \\ . . . line 2 \end{flushright} 

produce text that is left (flushleft) or right (flushright) justified. If a 

section of text does not fit on to one line, it is spread over several with 

fixed word spacing, the same as for the center environment. Again, word 

division does not occur. 

The same results may be produced within an environment with the 

declarations 

\raggedright replacing the flushleft environment, and 

\raggedleft replacing the flushright environment. 

4.2.3 Two-sided indentation 

A section of text may be displayed by indenting it by an equal amount on 

both sides, with the environments 

\begin{quote} text \end{quote} 

\begin{quotation} text \end{quotation} 

Additional vertical spacing is inserted above and below the 

displayed text to separate it visually from the normal text. 

The text to be displayed may be of any length; it can be part of 

a sentence, a whole paragraph, or several paragraphs.


Paragraphs are separated as usual with an empty line, although 

no empty lines are needed at the beginning and end of the 

displayed text since additional vertical spacing is inserted here 

anyway. 

The difference between the above two forms is thus: 

In the quotation environment, paragraphs are marked by 

extra indentation of the first line, whereas in the quote environment, 

they are indicated with more vertical spacing between 

them. 

The present text is produced within the quotation environment, 

while the sample above was done with the quote 

environment. 

The quotation environment is only really meaningful when 

the regular text makes use of first-line indentation to show off 

new paragraphs. 

4.2.4 Verse indentations 

For indenting rhymes, poetry, verses, etc. on both sides, the environment 

\begin{verse} poem \end{verse} 

is more appropriate. 

Stanzas are separated by blank lines 

while the individual lines of the stanza are divided by the \\ 

command. 

If a line is too long for the reduced text width, it will be left 

and right justified and continued on the next line, which is 

indented even further. 

The above indenting schemes may be nested inside one another. Within 

a quote environment there may be another quote, quotation, or verse 

environment. Each time, additional indentations are created on both sides 

of the text and vertical spacing is added above and below; these quantities 

however decrease as the depth of nesting increases. A maximum of six 

such nestings is allowed. 

Exercise 4.1: Put some appropriate sections of text in your exercise file into the 

quote and quotation environments, that is, enclose these sections within 

\begin{quote} . . . . . . . \end{quote} or 

\begin{quotation} . . . . . \end{quotation} 

commands.

4.3 Lists 

4.3. Lists 69 

Exercise 4.2: Make up a new file with the name poem.tex and type your favorite 

poem in the verse environment. Select 12pt as the standard font size and italic 

as the typeface. Put the title of the poem before the verse environment in a 

larger bold typeface, such as \Large\bfseries. Include the name of the poet 

right justified. 

Note: remember that you may include declarations to change the font style or 

size within an environment and that these remain in effect only until the end of 

that environment. 

Exercise 4.3: Make up another file with the name title.tex. Do you recall 

the titlepage environment for producing a free-form title page? If not, refer 

to Section 3.3.1. Create a title page with this environment using font sizes and 

styles of your choice, centering all the entries. 

Note: within the titlepage environment you may of course make use of the 

center environment, but it is also sufficient to give the \centering declaration 

instead, since this will remain in effect only until the end of the titlepage 

environment. 

Choose the individual line spacings with the command \\[len] using an 

appropriate value for the spacing len. Remember that vertical spacing before the 

first line of text must be entered with the *-form of the command \vspace*[len] 

(see Section 2.7.3). 

Experiment with different font sizes and styles for the various parts of the 

title page, such as title, author’s name, address, until you are satisfied with the 

results. 

Compare your own title page with that of Exercise 3.8. If your creation appeals 

to you more, include it in your standard exercise file by replacing the commands 

\title, \author, \date, and \maketitle with the titlepage environment and 

your own entries. 

There are three environments available for producing formatted lists: 

\begin{itemize} list text \end{itemize} 

\begin{enumerate} list text \end{enumerate} 

\begin{description} list text \end{description} 

In each of these environments, the list text is indented from the left margin 

and a label, or marker, is included. What type of label is used depends 

on the selected list environment. The command to produce the label is 

\item. 

4.3.1 Sample itemize 

• The individual entries are indicated with a black dot, a so-called 

bullet, as the label.


• The text in the entries may be of any length. The label appears at 

the beginning of the first line of text. 

• Successive entries are separated from one another by additional 

vertical spacing. 

The above text was produced as follows: 

\begin{itemize} 

\item The individual entries are indicated with a black dot, a 

so-called \emph{bullet}, as the label. 

\item The text in the entries may be of any length. The label 

appears at the beginning of the first line of text. 

\item Successive entries are separated from one another by 

additional vertical spacing. 

\end{itemize} 

4.3.2 Sample enumerate 

1. The labels consist of sequential numbers. 

2. The numbering starts at 1 with every call to the enumerate environment. 

The above example was generated with the following text: 

\begin{enumerate} 

\item The labels consist of sequential numbers. 

\item The numbering starts at 1 with every call to the 

\texttt{enumerate} environment. 

\end{enumerate} 

4.3.3 Sample description 

purpose This environment is appropriate when a number of words or 

expressions are to be defined. 

example A keyword is used as the label and the entry contains a clarification 

or explanation. 

other uses It may also be used as an author list in a bibliography. 

The above sample was created using the following: 

\begin{description} 

\item[purpose] This environment is appropriate when a number of 

words or expressions are to be defined. 

\item[example] A keyword is used as the label and the entry 

contains a clarification or explanation. 

\item[other uses] It may also be used as an author list in a 

bibliography. 

\end{description}

4.3. Lists 71 

The \item[option] command contains an optional argument that appears 

in bold face as the label. 

4.3.4 Nested lists 

The above lists may be included within one another, either mixed or of 

one type, to a depth of four levels. The type of label used depends on the 

depth of the nesting. The indentation is always relative to the left margin 

of the enclosing list. A fourfold nesting of the itemize environment 

appears as follows: 

• The label for the first level is a black dot, a bullet. 

– That of the second level is a long dash. 

∗ That of the third level is an asterisk. 

· And the label for the fourth level is a simple dot. 

· At the same time, the vertical spacing is decreased with 

increasing depth. 

∗ Back to the third level. 

– Back to the second level. 

• And here we are at the first level of itemize once again. 

Similarly for the enumerate environment, where the style of the numbering 

changes with the nesting level: 

1. The numbering at the first level is with Arabic numerals followed by 

a period. 

(a) At the second level, it is with lower case letters in parentheses. 

i. The third level is numbered with lower case Roman numerals 

with a period. 

A. At the fourth level, capital letters are used. 

B. The label style can be changed, as described in the next 

section. 

ii. Back to the third level. 

(b) Back to the second level. 

2. And the first level of enumerate again. 

An example of a nested list with mixed types: 

• The itemize label at the first level is a bullet. 

1. The numbering is with Arabic numerals since this is the first 

level of the enumerate environment.


– This is the third level of the nesting, but the second itemize 

level. 

(a) And this is the fourth level of the overall nesting, but 

only the second of the enumerate environment. 

(b) Thus the numbering is with lower case letters in parentheses. 

– The label at this level is a long dash. 

2. Every list should contain at least two points. 

• Blank lines ahead of an \item command have no effect. 

The above mixed list was produced with the following text: 


\item The \texttt{itemize} label at the first level is a ... 


\item The numbering is with Arabic numerals since this ... 


\item This is the third level of the nesting, but the ... 


\item And this is the fourth level of the overall ... 

\item Thus the numbering is with lower case letters ... 


\item The label at this level is a long dash. 

\end{itemize} 

\item Every list should contain at least two points. 


\item Blank lines ahead of an \verb+\item+ command ... 

\end{itemize} 

Exercise 4.4: Produce a nested list using the itemize and enumerate environments 

as in the above example, but with a different sequence of these commands. 

Exercise 4.5: Prepare a list of conference participants with their place of residence 

using the description environment, where the name of the participant appears 

as the argument in the \item command. 

Note: for all three types of lists, any text before the first \item command will 

yield an error message on processing. 

4.3.5 Changing label style 

The labels, or markers, used in the itemize and enumerate environments 

can be easily changed by means of the optional argument in the \item 

command. With \item[+] the label becomes +, and with \item[2.1:] 

it is 2.1:. The optional argument takes precedence over the standard

4.3. Lists 73 

label. For the enumerate environment, this means that the corresponding 

counter is not automatically incremented and the user must do the 

numbering manually. 

The optional label appears right justified within the area reserved for 

the label. The width of this area is the amount of indentation at that level 

less the separation between label and text; this means that the left edge 

of the label area is flush with the left margin of the enclosing level. 

It is also possible to change the standard labels for all or part of the 

document. The labels are generated with the internal commands 

\labelitemi , \labelitemii , \labelitemiii , \labelitemiv 

\labelenumi , \labelenumii , \labelenumiii , \labelenumiv 

The endings i, ii, iii, and iv refer to the four possible levels. 

These commands may be altered with \renewcommand. For example, 

to change the label of the third level of the itemize environment from ∗ 

to +, give 

\renewcommand{\labelitemiii}{+} 

Similarly the standard labels for the enumerate environment may be 

changed. However, here there is an additional complication that there 

is a counter for each enumerate level, named enumi, enumii, enumiii, 

and enumiv. As explained in Section 8.1.4, the value of a counter can be 

printed using one of the commands \arabic, \roman, \Roman, \alph, 

or \Alph, where the style of each command should be obvious from its 

name. That is, \Roman{xyz} prints the current value of the counter xyz 

in upper case Roman numerals, whereas \alph{xyz} prints it as a lower 

case letter (with a corresponding to 1 and z to 26). 

These counters, together with the counter style commands, must be 

used in the redefinitions of the label commands. For example, to change 

the second-level label to Arabic numerals followed by ‘.)’, it is necessary 

to give 

\renewcommand{\labelenumii}{\arabic{enumii}.)} 

which redefines \labelenumii to the value of counter enumii printed in 

Arabic, plus the characters ‘.)’. In this way, all the numbering levels may 

be changed. It is even possible to include more than one counter: 

\renewcommand{\labelenumii}{\Alph{enumi}.\arabic{enumii}} 

which will produce for every call to \item at level two the value of the 

counter enumi as a capital letter followed by the value of counter enumii 

as a number: that is, in the form A.1, A.2, . . . , B.1, B.2, . . . and so on. 

If the new standard labels are to apply to the whole document, the 

redefining commands should be included in the preamble. Otherwise, 

they are valid only within the environment in which they appear.


Package: 

enumerate 

Exercise 4.6: Change the standard labels for the itemize environment into a 

long dash — (written ---) for the first level, to a medium dash – (--) for the 

second level, and to a hyphen - for the third level. 

Exercise 4.7: Change the standard labels for the enumerate environment for the 

first level to (I), (II), . . . , and for the second level to the Roman numerals of the 

first level followed by the number for the second level in the form I–1:, I–2:, . . . , 

II–1:, II–2:, . . . . 

An alternative method of customizing the enumeration labels is with 

the enumerate package in the tools collection (Section B.5.4). Once this 

package has been loaded, the enumerate environment accepts an optional 

argument specifying the text of the label. The characters A a I i 1 

represent the number in alphabetical, Roman, Arabic styles. If these 

characters appear elsewhere in the label text, they must be in {}. For 

example, 

\begin{enumerate}[{Case} A] 

\item Witness tells the truth 

\item Witness is lying 


4.4 Generalized lists 

=⇒ 

Case A Witness tells the truth 

Case B Witness is lying 

Lists such as those in the three environments itemize, enumerate, and 

description can be formed in a quite general way. The type of label and 

its width, the depth of indentation, spacings for paragraphs and labels, 

and so on, may be wholly or partially set by the user by means of the list 

environment: 

\begin{list}{stnd lbl}{list decl} item list \end{list} 

Here item list consists of the text for the listed entries, each of which 

begins with an \item command that generates the corresponding label. 

The stnd lbl contains the definition of the label to be produced by the 

\item command when the optional argument is missing (see below). 

The list parameters described in Section 4.4.2 are set by list decl to 

whatever new values the user wishes. 

4.4.1 Standard label 

The first argument in the list environment defines the stnd lbl, that 

is, the label that is produced by the \item command when it appears 

without an argument. In the case of an unchanging label, such as for the 

itemize environment, this is simply the desired symbol. If this is to be a 

mathematical symbol, it must be given as symbol name, enclosed in 

signs. For example, to select ⇒ as the label, stnd lbl must be defined to 

be $\Rightarrow$.

4.4. Generalized lists 75 

However, the label is often required to contain a sequential numeration. 

For this purpose, a counter must be created with the \newcounter{name} 

command, where name is its designation. This command must appear 

before the first application of the counter in a list environment. Suppose 

a counter named marker has been defined for this use, then the argument 

stnd lbl could be any of the commands for printing counters described in 

Section 4.3.5: for example, \arabic{marker} produces a running Arabic 

number. 

Even more complex labels can be made up in this way. If the sequential 

labels are to be A–I, A–II, . . . , stnd lbl is set to A--\Roman{marker}. 

Before a counter can function properly within the standard label, it 

must be associated with that list by including in the list decl the command 

\usecounter{counter}, where counter is the name of the counter to be 

assigned (marker in the above example). 

The standard label is actually generated by the command \makelabel{label}, 

! which is called by the \item command. The user can redefine \makelabel with 

the aid of the \renewcommand in the list declaration: 

\renewcommand{\makelabel}{new definition} 

If the standard label is defined in this manner, the corresponding entry in the 

list environment is left blank. This is because \makelabel is the more general 

command and overrides the other definition. 

4.4.2 List style parameters 

There are a number of style parameters used for formatting lists that 

are set by L AT E X to certain standard values. These values may be altered 

by the user in the list decl for that particular list. The assignment is 

made in the usual way with the \setlength command. However, if the 

assignment is made outside the list environment, in most cases it will 

simply be ignored. This is because there are preset default values for 

each parameter at each level that can only be overridden by list decl. 

The style parameters are listed below and are also illustrated in Figure 

4.1 on the next page, which is based on one taken from Lamport (1985, 

1994). 

\topsep 

is the vertical spacing in addition to \parskip that is inserted 

between the list and the enclosing text above and below. Its 

default value is set at each list level and cannot be globally 

redefined outside the list decl. 

\partopsep 

is the vertical spacing in addition to \topsep + \parskip that 

is inserted above and below the list when a blank line precedes 

the first or follows the last \item entry. It may be redefined 

globally, but only for the first and second levels.


Preceding text 

✻ 

\topsep + \parskip [+ \partopsep] 

\labelsep 

✛ ✲ 

❄ 

✛ ✲ 

Label 

\rightmargin 

✛ ✲ ✛ ✲ 

Item 1 

✛ 

\labelwidth 

✲ \itemindent Paragraph 1 

\leftmargin 

\listparindent ✻\parsep 

✛ ✲ ❄ 

Item 1 

Paragraph 2 

Label 

✻\itemsep 

+ \parsep 

❄ 

Item 2 

✻ 


❄ 

Following text 

Figure 4.1: The list parameters 

\parsep 

is the vertical spacing between paragraphs of a single \item. Its 

default value is reset at each level, as for \topsep. 

\itemsep 

is the vertical spacing in addition to \parsep that is inserted 

between two \item entries. As for \topsep and \parsep, its 

default value is reset at each level and cannot be globally changed. 

\leftmargin 

is the distance from the left edge of the current environment 

to the left margin of the list text. There are default values for 

it at each level that may be globally redefined, as described in 


\rightmargin 

is the distance from the right edge of the current environment to 

the right margin of the list text. Its standard value is 0 pt, which 

can only be altered in list decl. 

\listparindent 

is the indentation depth of the first line of a paragraph within 

an \item with respect to the left margin of the list text. It is


normally set to 0 pt so that no indentation occurs. This can only 

be changed in list decl. 

\labelwidth 

is the width of the box reserved for the label. The text of the 

label is printed right justified within this space. A new default 

value may be set globally which then applies to all list levels. 

\labelsep 

is the spacing between the label box and the list text. A new value 

may be assigned globally, but it is only effective at the first level. 

\itemindent 

is the distance by which the label and the first line of text in an 

\item are indented to the right. It is normally set to 0 pt and so 

has no effect. This value can only be redefined in list decl. 

When changing the vertical spacings from their standard values, it is 

recommended that a rubber length (Section 2.4.2) be used. 

The label created by the \item command normally appears right justified 

within a box of width \labelwidth. It is possible to make it left 

justified, as in the following list of parameters, by putting \hfill at the 

end of the definition of the standard label or in the \makelabel command. 

4.4.3 Example of a user’s list 

List of Figures: 

Figure 1: Page format with head, body, and foot, showing the 

meaning of the various elements involved. 

Figure 2: Format of a general list showing its elements. 

Figure 3: A demonstration of some of the possibilities for 

drawing pictures with L AT E X. 

This list was produced with the following input: 

\newcounter{fig} 

\begin{list}{\bfseries\upshape Figure \arabic{fig}:} 

{\usecounter{fig} 

\setlength{\labelwidth}{2cm}\setlength{\leftmargin}{2.6cm} 

\setlength{\labelsep}{0.5cm}\setlength{\rightmargin}{1cm} 

\setlength{\parsep}{0.5ex plus0.2ex minus0.1ex} 

\setlength{\itemsep}{0ex plus0.2ex} \slshape} 

\item Page format with head, body, and foot, showing the 

meaning of the various elements involved. 

\item Format of a general list showing its elements. 

\item A demonstration of some of the possibilities for 

drawing pictures with \LaTeX. 

\end{list}


The command \newcounter{fig} sets up the counter fig. The standard 

label is defined to be the word Figure in upright, bold face, followed 

by the running Arabic number, terminated by :. This label is printed for 

each \item command. 

The list declaration contains \usecounter{fig} as its first command, 

which makes the counter fig operational within the list. The width of 

the label box (\labelwidth) is set to 2.0 cm, the left margin of the list 

text (\leftmargin) to 2.6 cm, the distance between the label and the text 

(\labelsep) to 0.5 cm, and the right edge of the list (\rightmargin) is 

set to be 1 cm from that of the enclosing text. 

The vertical spacing between paragraphs within an item (\parsep) 

is 0.5 ex but can be stretched an extra 0.2 ex or shrunk by 0.1 ex. The 

additional spacing between items (\itemsep) is 0 ex, stretchable to 0.2 ex. 

Standard values are used for all the other list parameters. The last 

command in the list declaration is \slshape, which sets the list text in a 

slanted typeface. 

Note: If \upshape were not given in the label definition, the text of each 

\item would also be slanted, as Figure 1:. 

4.4.4 Lists as new environments 

If a particular type of list is employed several times within a document, it 

can become tiresome typing the same stnd lbl and list decl into the list 

environment every time. L AT E X offers the possibility of defining a given list 

as an environment under its own name. This is achieved by means of the 

\newenvironment command. 

For example, the list in the above example can be stored so that it may 

be called at any time with the name figlist: 

\newenvironment{figlist}{\begin{list} 

{\bfseries\upshape Figure \arabic{fig}:} 

{\usecounter{fig} ... {0ex plus0.2ex}\slshape}} 

{\end{list}} 

It can then be called with 

\begin{figlist} item list \end{figlist} 

so that it behaves as a predefined list environment. 

Exercise 4.8: Define a new environment with the name sample that produces a 

list in which every call to \item prints labels Sample A, Sample B, and so on. The 

labels are to be left justified within a box of width 20mm, and the distance between 

the label box and the item text is to be 2mm, with a total left margin of 22mm. The 

right edge of the text is to be moved in 5mm from that of the enclosing text. The 

extra vertical spacing between two items is to be 1ex plus0.5ex minus0.4ex 

in addition to the normal paragraph spacing. Secondary paragraphs within an


item are to be indented by 1em. The normal paragraph separation should be 0ex, 

expandable to 0.5ex. 

L AT E X itself makes frequent use of the list environment to define a number of 

! other structures. For example, the quote environment is defined as 

\newenvironment{quote}{\begin{list}{} 

{setlength{\rightmargin}{\leftmargin}} 

\item[]}{\end{list}} 

This environment is thus a list in which the value of \rightmargin is set to the 

current value of \leftmargin which has a default value of 2.5 em. The list itself 

consists of a single \item call with an empty label, a call that is automatically 

included in the definition of quote with the entry \item[]. 

In the same way, L AT E X defines the quotation and verse environments internally 

as special list environments. The left margins and the vertical spacings 

around the structures are left as the standard values for the list environment, 

and are therefore changed only when the standard values themselves are altered. 

Finally, as an example of a possible user-defined special list we offer 

\newenvironment{lquote}{\begin{list}{}{}\item[]}{\end{list}} 

which creates an lquote environment that does nothing more than indent its 

enclosed text by the amount \leftmargin, with the right edge flush with that of 

the normal text, since \rightmargin has the standard value of 0 pt. 

4.4.5 Trivial lists 

! 

L AT E X also contains a trivlist environment, with syntax 

\begin{trivlist} enclosed text \end{trivlist} 

in which the arguments stnd lbl and list decl are omitted. This is the same as 

a list environment for which the label is empty, \leftmargin, \labelwidth, 

and \itemindent are all assigned the value 0 pt, while \listparindent is set 

equal to \parindent and \parsep to \parskip. 

L AT E X uses this environment to create further structures. For example, the call 

to the center environment generates internally the sequence 

\begin{trivlist} \centering \item[] enclosed text \end{trivlist} 

The environments flushleft and flushright are similarly defined. 

4.4.6 Nested lists 

Lists can be nested within one another with the list environments itemize, 

enumerate, and description, to a maximum depth of six. At each level, 

the new left margin is indented by the amount \leftmargin relative to 

that of the next higher one. 

As mentioned earlier, it is only possible to change the standard values of a 

! limited number of the list parameters with declarations in the preamble. One 

exception is the indentations of the left margins for the different nesting levels.


These are set internally by the parameters \leftmarginn, where n stands for i, 

ii, iii, iv, v, or vi. These values can be changed by the user; for example, by 

declaring \setlength{\leftmarginiv}{12mm}, the left margin of the fourthlevel 

list is shifted 12 mm from that of the third. These declarations must be 

made outside of the list environments and not in the list decl. 

At each level of list nesting, the internal macro \@listn (n being i to vi) is 

called. This sets the value of \leftmargin equal to that of the corresponding 

\leftmarginn, unless \leftmargin is explicitly declared in the list environment. 

That is, there does not exist a single standard value for \leftmargin 

externally, but rather six different ones. The parameter \leftmargin has meaning 

only within a list environment. 

4.5 Theorem-like declarations 

In scientific literature one often has text structures like 

Theorem 1 (Balzano–Weierstrass) Every infinite set of bounded points 

possesses at least one maximum point. 

or 

Axiom 4.1 The natural numbers form a set S of distinct elements. For any 

two elements a, b, they are either identical, a = b, or different from one 

another, a ≠ b. 

Similar structures frequently appear with names such as Definition, 

Corollary, Declaration, Lemma, instead of Theorem or Axiom. What they 

have in common is that a keyword and a running number are printed in 

bold face and the corresponding text in italic. 

Of course, these could be generated by the user by explicitly giving the 

type styles and appropriate number, but if a new structure of that type is 

later inserted in the middle of the text, the user would have the tedious 

job of renumbering all the following occurrences. With the command 

\newtheorem{struct type}{struct title}[in counter] 

L AT E X will keep track of the numbering automatically. Here struct type is 

the user’s arbitrary designation for the structure, while struct title is the 

word that is printed in bold face followed by the running number (for 

example, Theorem). If the optional argument in counter is missing, the 

numbering is carried out sequentially throughout the entire document. 

However, if the name of an existing counter, such as chapter, is given for 

in counter, the numbering is reset every time that counter is augmented, 

and both are printed together, as in Axiom 4.1 above. 

The predefined structures are called with the command 

\begin{struct type}[extra title] text \end{struct type}

4.6. Tabulator stops 81 

which also increments the necessary counter and generates the proper 

number. The above examples were produced with 

\newtheorem{theorem}{Theorem} \newtheorem{axiom}{Axiom}[chapter] 

. . . . . . . . . . . . . . . . 

\begin{theorem}[Balzano--Weierstrass] Every .... \end{theorem} 

\begin{axiom} The natural numbers form ........... \end{axiom} 

The optional extra title also appears in bold face within parentheses ( ) 

following the running number. 

Occasionally a structure is not numbered on its own but together with 

another structure. This can be included in the definition with another 

optional argument 

\newtheorem{struct type}[num like]{struct name} 

where num like is the name of an existing theorem structure that shares 

the same counter. Thus by defining \newtheorem{subthrm}[theorem] 

{Sub-Theorem}, the two structures theorem and subthrm will be numbered 

as a single series: Theorem 1, Sub-Theorem 2, Sub-Theorem 3, 

Theorem 4, and so on. 

For more powerful theorem tools, see the A M S amsthm package (Section 

12.3.1) and the theorem package in the tools collection (Section B.5.4). 

4.6 Tabulator stops 

4.6.1 Basics 

On a typewriter it is possible to set tabulator stops at various positions 

within a line; then by pressing the tab key the print head or carriage jumps 

to the next tab location. 

A similar possibility exists in L AT E X with the tabbing environment: 

\begin{tabbing} lines \end{tabbing} 

One can think of the set tab stops as being numbered from left to right. 

At the beginning of the tabbing environment, no tabs are set, except for 

the left border, which is called the zeroth tab stop. The stops can be set 

at any spot within a line with the command \=, and a line is terminated 

by the \\ command: 

Here is the \=first tab stop, followed by\= the second\\ 

sets the first tab stop after the blank following the word the, and the 

second immediately after the word by. 

After the tab stops have been set in this way, one can jump to each of 

the stops in the subsequent lines, starting from the left margin, with the 

command \>. A new line is started with the usual \\ command.


Example: 

Type Quality Color Price 

Paper med. white low 

Leather good brown high 

Card bad gray med. 

4.6.2 Sample line 

\begin{tabbing} 

Type\qquad\= Quality\quad\= 

Color\quad\= Price\\[0.8ex] 

Paper \> med. \> white \> low\\ 

Leather \> good \> brown \> high\\ 

Card \> bad \> gray \> med. 

\end{tabbing} 

It is often advantageous or even necessary to set the tab stops in a sample 

line that is not actually printed. It could contain, for example, the widest 

entries in the various columns that appear later, or the smallest intercolumn 

spacing between stops. The sample line may also contain \hspace 

commands to force the distance between stops to be a predetermined 

amount. 

To suppress the printing of the sample line, it is ended with the 

command \kill instead of the \\ terminator. 

\hspace*{3cm}\=sample column \=\hspace{4cm}\= \kill 

In addition to the left border, the above statement sets three tab stops: 

Left border 

3cm 

1st tab stop 

❄ 

✲sample 

column 

2nd tab stop 

❄ 

3rd tab stop 

❄ 

4cm ✲ 

An \hspace command at the beginning of a sample line must be of the 

*-form, otherwise the inserted spacing will be deleted at the line margin. 

4.6.3 Tab stops and the left margin 

The left border of each line of the tabbing environment is at first identical 

with the left margin of the enclosing environment, and is designated the 

zeroth stop. By activating the ‘tab key’ \> at the start of a line, one sets 

the following text beginning at the first tab stop. However, the command 

\+ has the same effect, putting the left border permanently at the first 

stop, for all subsequent lines. With \+\+ at the beginning or end of a line, 

all the next lines will start two stops further along. There can be as many 

\+ commands in all as there are tab stops set on the line. 

The command \- has the opposite effect: it shifts the left border for 

the following lines one stop to the left. It is not possible to set this border 

to be to the left of the zeroth stop. 

The effect of the \+ commands may be overridden for a single line by 

putting \< at the start for each tab to be removed. This line then starts so 

many tabs to the left of the present border. With the next \\ command, 

the new line begins at the current left border determined by the total 

number of \+ and \- commands.

4.6.4 Further tabbing commands 

4.6. Tabulator stops 83 

Tab stops can be reset or added in every line. The command \= will add 

a stop if there have been sufficient \> commands to have jumped to the 

last stop, otherwise it will reset the next stop. 

For example: 

Old column 1 Old column 2 

Left column Middle col Extra col 

New col 1 New col 2 Old col 3 

Column 1 Column 2 Column 3 

Old column 1 \= Old column 2\\ 

Left column \> Middle col 

\= Extra col\\ 

New col 1 \= New col 2 \> 

Old col 3\\ 

Column 1\> Column 2 \> Column 3 

\end{tabbing} 

Occasionally it is desirable to be able to reset the tab stops and then 

to reuse the original ones later. The command \pushtabs accomplishes 

this by storing the current tabs and removing them from the active line. 

All the tab stops can then be set once again. The stored stops can be 

reactivated with the command \poptabs. The \pushtabs command may 

be given as many times as needed, but there must be the same number of 

\poptabs commands within any one tabbing environment. 

It is possible to position text on a tab stop with left text \’ right text, 

where left text goes just before the current tab (or left border) with a 

bit of spacing, while right text starts exactly at the stop. The amount 

of spacing between the left text and the tab stop is determined by the 

tabbing parameter \tabbingsep. This may be changed by the user with 

the \setlength command as usual. 

Text may be right justified up against the right border of a line with 

the command \‘ text. There must not be any more \> or \= commands 

in the remainder of the line. 

The commands \=, \‘, and \’ function as accent commands outside 

of the tabbing environment (Section 2.5.7). If these accents are actually 

needed within tabbing, they must be produced with \a=, \a‘, and \a’ 

instead. For example, to produce ó, ò, or ō inside a tabbing environment, 

one must give \a’o, \a‘o, or \a=o. The command \- also has another 

meaning outside of the tabbing environment (suggested word division) 

but since lines are not broken automatically within this environment, 

there is no need for an alternative form. 

Here is an example illustrating all the tabbing commands:


Apples: consumed by: people 

horses 

and sheep 

reasonably juicy 

Grapefruits: a delicacy 

(see also: melons 

pumpkins) 

Horses feed on apples 

4.6.5 Remarks on tabbing 

\begin{tabbing} 

Grapefruits: \= \kill 

Apples: \> consumed by: \= people\+\+\\ 

horses \\ 

and \’ sheep\-\\ 

reasonably juicy\-\\ 

Grapefruits: \> a delicacy\\ 

\pushtabs 

(see also: \= melons\\ 

\> pumpkins)\\ 

\poptabs 

Horses \> feed on \> apples 

\end{tabbing} 

T E X treats the tabbing environment like a normal paragraph, breaking 

a page if necessary between two lines within the environment. However, 

the commands \newpage and \clearpage are not allowed within it, and 

the command \pagebreak is simply ignored. If the user wishes to force 

a page break within the tabbing environment, there is a trick that he or 

she may employ: specify a very large interline spacing at the end of the 

line where the break should occur (for example, \\[10cm]). This forces 

the break and the spacing disappears at the start of the new page. 

Each line of text is effectively within a { } pair, so that any size or font 

declarations remain in force only for that one line. The text need not be 

put explicitly inside a pair of curly braces. 

It is not possible to nest tabbing environments within one another. 

Beware: the tab jump command \> always moves to the next logical 

tab stop. This could actually be a move backwards if the previous text 

is longer than the space available between the last two stops. This is in 

contrast to the way the tabulator works on a typewriter. 

There is no automatic line breaking within the tabbing environment. 

Each line continues until terminated by a \\ command. The text could 

extend beyond the right margin of the page. The user must take care that 

this does not happen. 

The commands \hfill, \hrulefill, and \dotfill have no effect 

inside a tabbing environment, since no stretching takes place here. 

Exercise 4.9: Generate the following table with the tabbing environment. 

Project: Total Requirements = $900 000.00 

of which 2003 = $450 000.00 

2004 = $350 000.00 

2005 = $100 000.00 

2003 approved: $350 000.00 Deficiency: $100 000.00 

2004 $300 000.00 $150 000.00 

2005 $250 000.00 Surplus: $150 000.00

4.7. Boxes 85 

tentative 2004 = $100 000.00 for deficiency 2003 

2005 = $ 50 000.00 2004 

+ $100 000.00 excess for 2003 in 2004 

Commitments 2003 = $100 000.00 

2004 = $150 000.00 signed: H. André 

Hint: the first line in the tabbing environment should read 

Project: \=Total Requirements\= = \$900\,000.00 \+\\ 

What is the effect of the \+ command at the end of this line? How do you arrange, 

using these tab stops, for the years 2003, 2004, and 2005 in the second to fourth 

lines all to be positioned before the second tab stop? Which command should 

be at the end of the second line just before the \\terminator? 

Lines 1–4 and 8–12 all use the same set of tab stops, even though there are 

additional stops set in the eighth line. With \$1\=00\,000.00 one can align the 

entry \$\>50\,000.00 in the ninth line to match the decimal places of the lines 

above. 

Lines 5–7 have their own tab stops. Use the save and recall feature to store the 

preset tab stops and to bring them back. The left border of lines 5–7 correspond 

to the first stop of the first group. What command is at the end of the fourth line 

to ensure that the left border is reset to one stop earlier? How is the left border 

of the second-to-last line reset? 

The last line contains ‘signed: H. André’ right justified. With what tabbing 

command was this produced? Watch out for the accent é in this entry within the 

tabbing environment! 

4.7 Boxes 

A box is a piece of text that T E X treats as a unit, like a single character. A 

box (along with the text within it) can be moved left, right, up, or down. 

Since the box is a unit, T E X cannot break it up again, even it was originally 

made up of smaller individual boxes. It is, however, possible to put those 

smaller boxes together as one pleases when constructing the overall box. 

This is exactly what T E X does internally when it carries out the formatting: 

the individual characters are packed in character boxes, which 

are put together into line boxes horizontally with rubber lengths inserted 

between the words. The line boxes are stacked vertically into paragraph 

boxes, again with rubber lengths separating them. These then go into the 

page body box, which with the head and foot boxes constitutes the page 

box. 

L AT E X offers the user a choice of three box types: LR boxes, paragraph 

boxes, and rule boxes. The LR (left–right) box contains material that is 

ordered horizontally from left to right in a single line. A paragraph box 

will have its contents made into vertically stacked lines. A rule box is 

a rectangle filled solidly with black, usually for drawing horizontal and 

vertical lines.


4.7.1 LR boxes 

To create LR boxes containing single line text the commands 

\mbox{text} and \makebox[width][pos]{text} 

\fbox{text} and \framebox[width][pos]{text} 

are available. The two commands at the left produce an LR box with a 

width exactly equal to that of the text given between the braces { }. The 

\fbox command is the same as \mbox except that the text is also framed. 

With the two commands at the right, the width is predetermined by 

the optional length argument width. The other optional argument pos 

specifies how the text is positioned within the box. With no value given, 

the text is centered. Otherwise pos may be 

l to left justify the text, 

r to right justify it, 

s to stretch it to fill up the full width. 

Thus \makebox[3.5cm]{centered text} creates a box of width 3.5 cm 

in which the text is centered, as centered text , filled with white 

space, while with \framebox[3.5cm][r]{right justified} the text is 

right justified inside a framed box of width 3.5 cm: right justified . 

One may also give 

\framebox[3.5cm][s]{stretched\dotfill text} 

to fill up the box, as stretched . . . . . . . text , in which case some rubber 

length (Section 2.4.2) or other filler (page 30) must be added where the 

stretching is to occur. 

If the text has a natural width that is larger than that specified in width, 

it will stick out of the box on the left, right, or both sides, depending on 

the choice of pos. For example, 

\framebox[2mm]{centered} produces centered . 

The above application may appear rather silly for \framebox, but it 

can indeed be very useful for \makebox. A width specification of 0 pt for 

\makebox can generate a centered, left, or right justified positioning of 

text in diagrams made with the picture environment (see Chapter 13 for 

examples). It may also be used to cause two pieces of text to overlap, as 

\makebox[0pt][l]{/}S prints a slash through an S, as /S. 

Note: Length specifications must always contain a dimensional unit, even 

when they are zero. Thus 0pt must be given for the width, not 0. 

It is also possible to specify the width of an LR box relative to its 

natural dimensions (those produced by the simple \mbox command): 

\width is the natural width of the box, 

\height is the distance from baseline to top, 

\depth is the distance from baseline to bottom, 

\totalheight is \height plus \depth.

4.7. Boxes 87 

To make a framed box such that the width is six times the total height, 

containing centered text, 

\framebox[6\totalheight]{Text} Text 

Note: These special length parameters only have meaning within the 

width specification of an LR box, or within the height specification of a 

paragraph box, as shown below. In any other context, they will produce 

an error message. 

If a set piece of text is to appear in several places within the document, 

it can be stored by first giving the command 

\newsavebox{\boxname} 

to create a box with the name \boxname. This name must conform to 

L AT E X command name syntax (letters only) with an initial \. The name 

must not conflict with any existing L AT E X command names. After such a 

box has been initiated, the commands 

\sbox{\boxname}{text} or 

\savebox{\boxname}[width][pos]{text} 

will store the contents text for future use. The optional arguments width 

and pos have the same meanings as for \makebox and \framebox. Now 

with the command 

\usebox{\boxname} 

the stored contents are inserted into the document text wherever desired, 

as a single unit. 

The contents of an LR box may also be stored with the environment 

\begin{lrbox}{boxname} 

text 

\end{lrbox} 

This is equivalent to \sbox{\boxname}{text}. Its advantage is that it 

allows text within a user-defined environment (Section 8.4) to be stored 

for future use with \usebox. 

4.7.2 Vertical shifting of LR boxes 


\raisebox{lift}[height][depth]{text} 

produces an \mbox with contents text, raised above the current baseline 

by an amount lift. The optional arguments tell L AT E X to treat the box as 

though its extension above the baseline were height and that below were 

depth. Without these arguments, the box has its natural size determined 

by text and lift. Note that lift, height, and depth are lengths (Section 2.4.1). 

If lift is negative, the box is lowered below the baseline. 

For example:


Baseline \raisebox{1ex}{high} and \raisebox{-1ex}{low} 

and back again 

produces: Baseline high and low and back again. 

The values for height and depth can be totally different from the actual 

ones of the text. Their effect is to determine how far away the previous and 

next lines of text should be from the current line, based on the heights 

and depths of all the boxes (characters are also boxes) in the line. By 

raising a box but specifying height to be the regular character size, the 

raised box will overprint the line above, and similarly for depth when a 

box is lowered. 

4.7.3 Parboxes and minipages 

Whole paragraphs can be put into separate vertical boxes (or parboxes in 

the L AT E X jargon) with the command 

\parbox[pos]{width}{text} 

or with the environment 

\begin{minipage}[pos]{width} text \end{minipage} 

Both produce a vertical box of width width, in which the lines of text are 

stacked on top of each other as in normal paragraph mode. 

The optional positioning argument pos can take on the values 

b to align the bottom edge of the box with the current baseline, 

t to align the top line of text with the current baseline. 

Without any positioning argument, the parbox is centered vertically on 

the baseline of the external line of text. 

The positioning argument is only meaningful when the \parbox command 

or the minipage environment occurs within a line of text, for 

otherwise the current line and its baseline have no meaning. If the parbox 

is immediately preceded by a blank line, it begins a new paragraph. In 

this case, the vertical positioning of the parbox is made with reference 

to the following elements of the paragraph. These could be further parboxes. 

If the paragraph consists of only a single parbox or minipage, the 

positioning argument is meaningless and has no effect. 

Examples: 

\parbox{3.5cm}{\sloppy This is a 3.5 cm wide parbox. It is 

vertically centered on the} 

\hfill CURRENT LINE \hfill 

\parbox{5.5cm}{Narrow pages are hard to format. They usually 

produce many warning messages on the monitor. The command 

\texttt{\symbol{92}sloppy} can stop this.}

This is a 3.5 cm wide 

parbox. It is vertically 

centered on the 

CURRENT LINE 

4.7. Boxes 89 

Narrow pages are hard to format. 

They usually produce many warning 

messages on the monitor. The command 

\sloppy can stop this. 

\begin{minipage}[b]{4.3cm} 

The minipage environment creates a vertical box like the parbox 

command. The bottom line of this minipage is aligned with the 

\end{minipage}\hfill 

\parbox{3.0cm}{middle of this narrow parbox, which in turn is 

aligned with} 

\hfill 

\begin{minipage}[t]{3.8cm} 

the top line of the right hand minipage. It is recommended that 

the user experiment with the positioning arguments to get used 

to their effects. 

\end{minipage} 

The minipage environment 

creates a vertical box like the 

parbox command. The bottom 

line of this minipage is 

aligned with the 

middle of this narrow 

parbox, which in 

turn is aligned with 

the top line of the right 

hand minipage. It is recommended 

that the user 

experiment with the positioning 

arguments to get 

used to their effects. 

In Section 4.7.7 we demonstrate how parboxes can be vertically stacked 

in any desired manner relative to one another. 

The \parbox command produces a vertical box containing the text 

just like the minipage environment. However, the latter is more general. 

The text in a \parbox may not contain any of the centering, list, or other 

environments described in Sections 4.2–4.5. These may, on the other 

hand, appear within a minipage environment. That is, a minipage can 

include centered or indented text, as well as lists and tabbings. 

4.7.4 Problems with vertical placement 

Vertical positioning of minipages and parboxes can often lead to unexpected 

! results, which can be explained by showing more graphically how a box is treated 

by L AT E X. Suppose we want to place two parboxes of different heights side by 

side, aligned on their first lines, and the two together set on the current line of 

text at the bottom. The ‘obvious’ way of doing this is 

\begin{minipage}[b]{..} 

\parbox[t]{..}{..} \hfill \parbox[t]{..}{..} 

\end{minipage}


which does not work, for it produces instead the following results: 

Current line ••The boxes are made• 

•and by marking•• 

of text. 

visible here by fram- the baselines (the 

ing them, 

vertical alignment 

points) with black 

dots. 

The reason for this is that each parbox or minipage is treated externally as a 

single character with its own height and depth above and below the baseline. As 

far as the outer minipage is concerned, it contains only two ‘characters’ on the 

same line, and that line is both the top and bottom one. Thus the bottom line of 

the outer minipage is indeed aligned with the line of text, but that bottom line is 

simultaneously the top line. The solution is to add a dummy second line to the 

outer box, as 

\parbox[t]{..}{..} \hfill \parbox[t]{..}{..} \\ \mbox{} 

The dummy line may not be entirely empty, hence the \mbox. 

•The boxes are made• 

•and by marking• 

visible here by fram- the baselines (the 

ing them, 

vertical alignment 

points) with black 

dots. 

Current line • • • of text. 

A similar problem occurs if two boxes are to be aligned with their bottom 

lines, and the pair aligned at the top with the current line of text. Here there are 

two possibilities to add a dummy first line. 

\mbox{} \\ aligns with the very top, or 

\mbox{} \\[-\baselineskip] aligns with the first text line. 

An example of the first case is shown on page 176. Dummy lines are also needed 

for the solution of Exercise 4.10 on page 93. 

4.7.5 Paragraph boxes of specific height 

The complete syntax of the \parbox command and minipage environment 

includes two more optional arguments: 

\parbox[pos][height][inner pos]{width}{text} 

\begin{minipage}[pos][height][inner pos]{width} 

text 


In both cases, height is a length specifying the height of the box; the parameters 

\height, \width, \depth, and \totalheight may be employed 

within the height argument in the same way as in the width argument of 

\makebox and \framebox (page 86).

4.7. Boxes 91 

The optional argument inner pos states how the text is to be positioned 

internally, something that is only meaningful if height has been given. Its 

possible values are: 

t to push the text to the top of the box, 

b to shove it to the bottom, 

c to center it vertically, 

s to stretch it to fill up the whole box. 

In the last case, rubber lengths (Section 2.4.2) should be present where 

the vertical stretching is to take place. 

Note the difference between the external positioning argument pos and 

the internal one inner pos: the former states how the box is to be aligned 

with the surrounding text, while the latter determines how the contents 

are placed within the box itself. 

Example: 

\begin{minipage}[t][2cm][t]{3cm} 

This is a minipage of height 2˜cm with the text 

at the top. 

\end{minipage}\hrulefill 

\parbox[t][2cm][c]{3cm}{In this parbox, the text 

is centered on the same height.}\hrulefill 

\begin{minipage}[t][2cm][b]{3cm} 

In this third paragraph box, the text is at the bottom. 


This is a minipage 

of height 2 cm 

with the text at the 

top. 

In this parbox, the 

text is centered on 

the same height. 

In this third paragraph 

box, the text 

is at the bottom. 

The \hrulefill commands between the boxes show where the baselines 

are. All three boxes are the same size and differ only in their values of 

inner pos. 

4.7.6 Rule boxes 

A rule box is a basically a filled-in black rectangle. The syntax for the 

general command is: 

\rule[lift]{width}{height} 

which produces a solid rectangle of width width and height height, raised 

above the baseline by an amount lift. Thus \rule{8mm}{3mm} generates 

. Without the optional argument lift, the rectangle is set on the 

baseline of the current line of text.


The parameters lift, width, and height are all lengths (Section 2.4.1). If 

lift has a negative value, the rectangle is set below the baseline. 

It is also possible to have a rule box of zero width. This creates an 

invisible line with the given height. Such a construction is called a strut 

and is used to force a horizontal box to have a desired height or depth 

that is different from that of its contents. For this purpose, \vspace is 

inappropriate because it adds additional vertical space to that which is 

already there. 

For example: \fbox{Text} produces Text . In order to print Text , 

one has to tell T E X that the box contents extend above and below the 

baseline by the desired amounts. This was done with \fbox{\rule[-2mm] 

{0cm}{6mm}Text}. What this says is that the text to be framed consists 

of ‘an invisible bar beginning 2 mm below the baseline, 6 mm long, 

followed by the word Text’. The vertical bar indeed remains unseen, but 

it determines the upper and lower edges of the frame. 

4.7.7 Nested boxes 

The box commands described above may be nested to any desired level. 

Including an LR box within a parbox or a minipage causes no obvious 

conceptual difficulties. The opposite, a parbox within an LR box, is also 

possible, and is easy to visualize if one keeps in mind that every box is a 

unit, treated by L AT E X as a single character of the corresponding size. 

A parbox inside an \fbox command has the effect that the 

entire parbox is framed. The present structure was made with 

\fbox{\fbox{\parbox{10cm}{A parbox...}}} 

This is a parbox of width 10 cm inside a framebox inside a 

second framebox, which thus produces the double framing 

effect. 

Enclosing a parbox inside a \raisebox allows vertical 

displacements of any desired amount. The two boxes 

here both have positioning [b], but the one at the right 

has been produced with: 

\raisebox{1cm}{\begin{minipage}[b]{2.5cm} 

a b c d e ... x y z\\ 

\underline{baseline} 

\end{minipage} } 

a b c d e f g h i 

j k l m n o p q r 

s t u v w x y z 

baseline 

which displaces it upwards by 1 cm. baseline 

A very useful structure is one in which minipage environments are 

positioned relative to one another inside an enclosing minipage. The 

positioning argument of the outside minipage can be used to align its

4.7. Boxes 93 

contents as a unit with the neighboring text or boxes. An example of this 

is given in Exercise 4.10. 

Finally, vertical boxes such as \parbox commands and minipage environments 

may be saved as the text in an \sbox or \savebox command, 

to be recalled later with \usebox, as described in Section 4.7.1. 

4.7.8 Box style parameters 

There are two style parameters for the frame boxes \fbox and \framebox that 

! may be reset by the user: 

\fboxrule determines the thickness of the frame lines, 

\fboxsep sets the amount of spacing between the frame and enclosed text. 

New values are assigned to these length parameters in the usual L AT E X manner with 

the command \setlength: the line thickness for all the following \framebox 

and \fbox commands is set to 0.5 mm with \setlength{\fboxrule}{0.5mm}. 

The scope of these settings also obeys the usual rule: if they are found 

in the preamble then they apply to the entire document; if they are within an 

environment then they are valid only until the end of that environment. 

These parameters do not influence the \framebox command that is employed 

within the picture environment (Section 13.1.4) and which has different syntax 

and functionality from those of the normal \framebox command. 

Exercise 4.10: How can the following nested structure be generated? (Note: font 

size is \footnotesize) 

The first line of this 3.5 cm 

wide minipage or parbox is 

aligned with the first line 

of the neighboring minipage 

or parbox. 

This 4.5 cm wide minipage or parbox 

is positioned so that its top 

line is at the same level as that of 

the box on the left, while its bottom 

line is even with that of the 

box on the right. The naïve notion 

that this arrangement may be 

achieved with the positioning arguments 

set to t, t, and b is incorrect. 

Why? What would this 

selection really produce? 

The true solution 

involves the nesting 

of two of the three 

structures in an enclosing 

minipage, 

which is then separately 

aligned with the 

third one. 

Note: there are two variants for the solution, depending on whether the left and 

middle structures are first enclosed in a minipage, or the middle and right ones. 

Try to work out both solutions. Incidentally, the third minipage is 3 cm wide. 

Note: the problems of correctly aligning two side-by-side boxes as a pair on a 

line of text (Section 4.7.4) arises here once more. It will be necessary to add a 

dummy line to get the vertical alignment correct. 

Exercise 4.11: Produce the framed structure shown below and store it with 

the command \sbox{\warning}{structure}. You will first have to create a 

box named \warning with the \newsavebox{\warning} command. Print this 

warning at various places in your exercise file by giving \usebox{\warning}.


Vertical placement of minipages and parboxes can lead to 

surprising results which may be corrected by the use of dummy 

lines. 

Note: the parbox width is 10 cm. There should be no difficulty producing the 

framed structure if one follows the previous example for the double framed box. 

Watch out when writing \sbox{\warning}{structure} that you have the correct 

number of closing braces at the end. 

Next, change the values for the line thickness (\fboxrule) and frame spacing 

(\fboxsep) and print your results once more. 

4.7.9 Further framed boxes 

Package: The fancybox package, by Timothy van Zandt, allows additional framed 

fancybox boxes of various styles. These all make use of the length \fboxsep to 

set the distance between frame and text, the same as for \fbox and 

\framebox. Depending on the box type, additional new lengths are also 

applicable. They may be changed with \setlength to modify the box 

appearance. These new framed boxes are: 

\shadowbox{text} 

The width of the shadow is given by the length 

\shadowsize, default 4 pt. Multiline text must be placed 

in a minipage environment, the same as for \fbox. 

\doublebox{text} 

The width of the inner frame is 0.75\fboxrule, that of the 

outer frame 1.5\fboxrule, and the spacing between the 

frames is 1.5\fboxrule plus 0.5 pt. 

✬ 

\ovalbox{text} 

The thickness of the frame is that of \thinlines (Section 

13.1.4), the diameter of the corners is set with the 

command \cornersize{frac}, to frac times the smaller 

of the box width or height, or with \cornersize*{size} to 

the length size. The default is frac=.5. 

✫ 

✬ 

\Ovalbox{text} 

The frame thickness is set by \thicklines, but otherwise 

is the same as \ovalbox. 

✫ 

✩ 

✪ 

✩ 

✪

4.8. Tables 95 

This package also allows all pages to be boxed as part of the page style. 

This is done by issuing \fancypage{cmds1}{cmds2}, where cmds1 and 

cmds2 are commands setting box parameters, terminated by one of the 

box commands \fbox, \shadowbox, and so on. The first set, cmds1, form 

a box with the head and footlines outside, while the second set draw a 

box including them. Normally one would only specify one set, leaving the 

other blank, such as: 

\fancypage{\setlength{\fboxsep}{5pt} 

\setlength{\shadowsize}{3pt}\shadowbox}{} 

for a shadow box on each page excluding head and footlines. There is also 

the command \thisfancypage{cmds1}{cmds2} to box just the current 

page. 

4.8 Tables 

With the box elements and tabbing environment from the previous sections 

it would be possible to produce all sorts of framed and unframed 

tables. However, L AT E X offers the user far more convenient ways to build 

such complicated structures. 

4.8.1 Constructing tables 

The environments tabular, tabular*, and array are the basic tools 

with which tables and matrices can be constructed. The syntax for these 

environments is 

\begin{array}[pos]{cols} rows \end{array} 

\begin{tabular}[pos]{cols} rows \end{tabular} 

\begin{tabular*}{width}[pos]{cols} rows \end{tabular*} 

The array environment can only be applied in mathematical mode (see 

Chapter 5). It is described here only because its syntax and the meaning of 

its arguments are exactly the same as those of the tabular environment. 

All three environments actually create a minipage. The meaning of the 

arguments is as follows: 

pos Vertical positioning argument (see also the explanation of this argument 

for parboxes in Section 4.7.3). It can take on the values 

t the top line of the table is aligned with the baseline of the 

current external line of text; 

b the bottom line of the table is aligned with the external baseline; 

with no positioning argument given, the table is centered on 

the external baseline.


width This argument applies only to the tabular* environment and determines 

its overall width. In this case, the cols argument must 

contain the @-expression (see below) @{\extracolsep{\fill}} 

somewhere after the first entry. For the other two environments, 

the total width is fixed by the textual content. 

cols The column formatting argument. There must be an entry for 

every column, as well as possible extra entries for the left and right 

borders of the table or for the intercolumn spacings. 

The possible column formatting symbols are 

l the column contents are left justified; 

r the column contents are right justified; 

c the column contents are centered; 

p{wth} the text in this column is set into lines of width wth, and 

the top line is aligned with the other columns. In fact, the text is 

set in a parbox with the command \parbox[t]{wth}{column 

text}; 

*{num}{cols} the column format contained in cols is reproduced 

num times, so that *{5}{|c}| is the same as |c|c|c|c|c|. 

The available formatting symbols for the left and right borders and 

for the intercolumn spacing are 

| draws a vertical line; 

|| draws two vertical lines next to each other; 

@{text} this entry is referred to as an @-expression, and inserts text 

in every line of the table between the two columns where it 

appears. 

An @-expression removes the intercolumn spacing that is automatically 

put between each pair of columns. If white space 

is needed between the inserted text and the next column, this 

must be explicitly included with \hspace{ } within the text 

of the @-expression. If the intercolumn spacing between two 

particular columns is to be something other than the standard, 

this may be easily achieved by placing @{\hspace{wth}} between 

the appropriate columns in the formatting argument. 

This replaces the standard intercolumn spacing with the width 

wth. 

An \extracolsep{wth} within an @-expression will put extra 

spacing of amount wth between all the following columns, 

until countermanded by another \extracolsep command. In 

contrast to the standard spacing, this additional spacing is not 

removed by later @-expressions. In the tabular* environment, 

there must be a command @{\extracolsep\fill} somewhere 

in the column format so that all the subsequent intercolumn 

spacings can stretch out to fill the predefined table width.


If the left or right borders of the table do not consist of a 

vertical line, spacing is added there of an amount equal to half 

the normal intercolumn spacing. If this spacing is not wanted, 

it may be suppressed by including an empty @-expression @{} 

at the beginning or end of the column format. 

rows contain the actual entries in the table, each horizontal row being 

terminated with a \\ command. These rows consist of a sequence 

of column entries separated from each other by the & symbol. Thus 

each row in the table contains the same number of column entries 

as in the column definition cols. Some entries may be empty. The 

individual column entries are treated by L AT E X as though they were 

enclosed in braces { }, so that any changes in type style or size are 

restricted to that one column. 

\hline This command may only appear before the first row or 

immediately after a \\ row termination. It draws a horizontal 

line the full width of the table below the row that was just 

ended, or at the top of the table if it comes at the beginning. 

Two \hline commands together draw two horizontal lines 

with a little space between them. 

\cline{m − n} This command draws a horizontal line from the 

left side of column m to the right side of column n. Like 

\hline, it may only be given just after a \\ row termination 

and there may be more than one after another. The command 

\cline{1-3} \cline{5-7} draws two horizontal lines from 

column 1 to 3 and from column 5 to 7, below the row that was 

just ended. In each case, the full column widths are underlined. 

\multicolumn{num}{col}{text} This command combines the following 

num columns into a single column with their total width 

including intercolumn spacings. The argument col contains exactly 

one of the positioning symbols l, r, or c, with possible 

@-expressions and vertical lines |. A value of 1 may be given 

for num when the positioning argument is to be changed for 

that column in one particular row. 

In this context, a ‘column’ starts with a positioning symbol 

l, r, or c, and includes everything up to but excluding the next 

one. The first column also includes everything before the first 

positioning symbol. Thus |c@{}rl| contains three columns: 

the first is |c@{}, the second r, and the third l|. 

The \multicolumn command may only come at the start of 

a row or right after a column separation symbol &. 

\vline This command draws a vertical line with the height of the 

row at the location where it appears. In this way, vertical lines 

that do not extend the whole height of the table may be inserted 

within a column.


If a p-type column contains \raggedright or \centering, the \\ forces a 

! new line within the column entry and not the end of the whole row. If this occurs 

in the last column, then \\ cannot be used to terminate the row; instead one 

must use \tabularnewline to end such a row. 

Since a table is a vertical box of the same sort as parbox and minipage, 

it may be positioned horizontally with other boxes or text (see examples 

in Section 4.7.3). In particular, the table must be enclosed within 

\begin{center} table \end{center} 

in order to center it on the page. 

4.8.2 Table style parameters 

There are a number of style parameters used in generating tables which L AT E X sets 

! to standard values. These may be altered by the user, either globally within the 

preamble or locally inside an environment. They should not be changed within 

the tabular environment itself. 

\tabcolsep is half the width of the spacing that is inserted between columns 

in the tabular and tabular* environments; 

\arraycolsep is the corresponding half intercolumn spacing for the array 

environment; 

\arrayrulewidth is the thickness of the vertical and horizontal lines within a 

table; 

\doublerulesep is the separation between the lines of a double rule. 

Changes in these parameters can be made with the \setlength command as 

usual. For example, to make the line thickness to be 0.5 mm, give \setlength 

{\arrayrulewidth}{0.5mm}. Furthermore, the parameter 

\arraystretch can be used to change the distance between the rows of a table. 

This is a multiplying factor, with a standard value of 1. A value of 1.5 

means that the inter-row spacing is increased by 50%. A new value is set 

by redefining the parameter with the command 

4.8.3 Table examples 

\renewcommand{\arraystretch}{factor} 

Creating tables is much easier in practice than it would seem from the 

above list of formatting possibilities. This is best illustrated with a few 

examples. 

The simplest table consists of a row of columns in which the text 

entries are either centered or justified to one side. The column widths, 

the spacing between the columns, and thus the entire width of the table 

are automatically calculated.


Position Club Games W T L Goals Points 

1 Amesville Rockets 33 19 13 1 66:31 51:15 

2 Borden Comets 33 18 9 6 65:37 45:21 

3 Clarkson Chargers 33 17 7 9 70:44 41:25 

4 Daysdon Bombers 33 14 10 9 66:50 38:28 

5 Edgartown Devils 33 16 6 11 63:53 38:28 

6 Freeburg Fighters 33 15 7 11 64:47 37:29 

7 Gadsby Tigers 33 15 7 11 52:37 37:29 

8 Harrisville Hotshots 33 12 11 10 62:58 35:31 

9 Idleton Shovers 33 13 9 11 49:51 35:31 

10 Jamestown Hornets 33 11 11 11 48:47 33:33 

11 Kingston Cowboys 33 13 6 14 54:45 32:34 

12 Lonsdale Stompers 33 12 8 13 50:57 32:34 

13 Marsdon Heroes 33 9 13 11 50:42 31:35 

14 Norburg Flames 33 10 8 15 50:68 28:38 

15 Ollison Champions 33 8 9 16 42:49 25:41 

16 Petersville Lancers 33 6 8 19 31:77 20:46 

17 Quincy Giants 33 7 5 21 40:89 19:47 

18 Ralston Regulars 33 3 11 19 37:74 17:49 

The above table is made up of eight columns, the first of which is 

right justified, the second left justified, the third centered, the next three 

right justified again, and the last two centered. The column formatting 

argument in the tabular environment thus appears as 

{rlcrrrcc} 

The text to produce this table is 

\begin{tabular}{rlcrrrcc} 

Position & Club & Games & W & T & L & Goals & Points\\[0.5ex] 

1 & Amesville Rockets & 33 & 19 & 13 & 1 & 66:31 & 51:15 \\ 

2 & Borden Comets & 33 & 18 & 9 & 6 & 65:37 & 45:21 \\ 

... & ..... & .. & .. & .. & .. & ... & ... \\ 

17 & Quincy Giants & 33 & 7 & 5 & 21 & 40:89 & 19:47 \\ 

18 & Ralston Regulars & 33 & 3 & 11 & 19 & 37:74 & 17:49 

\end{tabular} 

In each row, the individual columns are separated from one another by 

the symbol & and the row itself is terminated with the \\ command. The 

[0.5ex] at the end of the first row adds extra vertical spacing between 

the first two rows. The last row does not need the termination symbol 

since it is ended automatically by the \end{tabular} command. 

The columns may be separated by vertical rules by including the 

symbol | in the column formatting argument. Changing the first line to 

\begin{tabular}{r|l||c|rrr|c|c}


results in 



2 Borden Comets 33 18 9 6 65:37 45:21 

. 

. 

. 

17 Quincy Giants 33 7 5 21 40:89 19:47 


The same symbol | before the first or after the last column format 

generates a vertical line on the outside edge of the table. Two symbols || 

produce a double vertical line. Horizontal lines over the whole width of 

the table are created with the command \hline. They may only appear 

right after a row termination \\ or at the very beginning of the table. Two 

such commands \hline\hline draw a double horizontal line. 

\begin{tabular}{|r|l||c|rrr|c|c|} \hline 

Position & Club & Games & W & T & L & Goals & Points\\ 

\hline\hline 

1 & Amesville Rockets & 33 & 19 & 13 & 1 & 66:31 & 51:15 \\ 

\hline 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

18 & Ralston Regulars & 33 & 3 & 11 & 19 & 37:74 & 17:49 \\ 

\hline 

\end{tabular} 

The table now appears as 



2 Borden Comets 33 18 9 6 65:37 45:21 

. 

. 

. 

17 Quincy Giants 33 7 5 21 40:89 19:47 


In this case, the row termination \\ must be given for the last row too 

because of the presence of \hline at the end of the table. 

In this example, all rows contain the same entry in the third column, 

that is, 33. Such a common entry can be automatically inserted in the 

column format as an @-expression of the form @{text}, which places text 

between the neighboring columns. This could be accomplished for our 

example by changing the column format to 

{rl@{ 33 }rrrcc} or {|r|l||@{ 33 }|rrr|c|c|} 

so that the text ‘ 33 ’, blanks included, appears between the second and 

third columns in every row. This produces the same table with slightly 

different row entries: for example, the fourth row would now be given as

4.8. Tables 101 

4 & Daysdon Bombers & 14 & 10 & 9 & 66:50 & 38:28 \\ 

The column format now consists of only seven column definitions, rlrrrcc. 

! The previous third column c has been removed, and so each row contains one less 

column separation symbol &. The new third column, the number of games won, 

begins with the second & and is separated from the club name by the contents 

of @{ 33 }, which is entered automatically without any additional & symbol. 

The last two columns give the relations between goals and points won 

and lost as a centered entry of the form m:n. The colons ‘:’ are only 

coincidentally ordered exactly over one another since two-digit numbers 

appear in every case on both sides of the colon. If one entry had been 

9:101, the colon would have been shifted slightly to the left as the entire 

entry was centered. 

A vertical alignment of the ‘:’ independent of the number of digits 

can also be achieved using an @-expression of the form r@{:}l in the 

column format. This means that a colon is placed in every row between a 

right and a left justified column. The column formatting argument in the 

example now becomes 

{rl@{ 33 }rrrr@{:}lr@{:}l} or 

{|r|l||@{ 33 }|rrr|r@{:}l|r@{:}l|} 

and the row entry is 

4 & Daysdon Bombers & 14 & 10 & 9 & 66 & 50 & 38 & 28 \\ 

Each of the former c columns has been replaced by the two columns in 

! r@{:}l. An @-expression inserts its text between the neighboring columns, 

removing the intercolumn spacing that would normally be there. Thus the r 

column is justified flush right with the ‘:’ and the following l column flush left. 

The same method can be employed when a column consists of numbers with 

decimal points and a varying number of digits. 

The entries for the goal and point relationships are now made up of 

two columns positioned about the ‘:’ symbol. This causes no problems 

for entering the number of goals won and lost or for the number of plus 

and minus points, since each entry has its own column. The column 

headings, however, are the words ‘Goals’ and ‘Points’, stretching over 

two columns each and without the colon. This is accomplished with the 

\multicolumn command, which merges selected columns in a particular 

row and redefines the column format. The first row of the unframed 

soccer table is then 

Position& Club & W & T & L & \multicolumn{2}{c}{Goals} 

& \multicolumn{2}{c}{Points}\\[0.5ex] 

Here \multicolumn{2}{c}{Goals} means that the next two columns 

are to be combined into a centered column, containing the text ‘Goals’. 

For the framed table, the new formatting argument in the \multicolumn 

commands must be {c|} since the vertical line symbol | was also removed 

when the old columns were combined. In deciding what belongs to a given


column, use the rule that a column ‘owns’ everything up to but excluding 

the next r, l, or c. 

The table of final results for our soccer league 2002/03 is to have the 

following title: 

\begin{tabular}{|r|l||rrr|r@{:}l|r@{:}l||c|}\hline 

\multicolumn{10}{|c|}{\bfseries 1st Regional Soccer League --- 

Final Results 2002/03}\\ \hline 

&\itshape Club &\itshape W &\itshape T &\itshape L & 

\multicolumn{2}{c|}{\itshape Goals} 

& \multicolumn{2}{c||}{\itshape Points} 

& \itshape Remarks \\ \hline\hline 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

1st Regional Soccer League — Final Results 2002/03 

Club W T L Goals Points Remarks 

1 Amesville Rockets 19 13 1 66:31 51:15 League Champs 

2 Borden Comets 18 9 6 65:37 45:21 Trophy Winners 

3 Clarkson Chargers 17 7 9 70:44 41:25 Candidates 

4 Daysdon Bombers 14 10 9 66:50 38:28 for 

5 Edgartown Devils 16 6 11 63:53 38:28 National 

6 Freeburg Fighters 15 7 11 64:47 37:29 League 

7 Gadsby Tigers 15 7 11 52:37 37:29 

8 Harrisville Hotshots 12 11 10 62:58 35:31 

9 Idleton Shovers 13 9 11 49:51 35:31 

10 Jamestown Hornets 11 11 11 48:47 33:33 

11 Kingston Cowboys 13 6 14 54:45 32:34 Medium Teams 

12 Lonsdale Stompers 12 8 13 50:57 32:34 

13 Marsdon Heroes 9 13 11 50:42 31:35 

14 Norburg Flames 10 8 15 50:68 28:38 

15 Ollison Champions 8 9 16 42:49 25:41 

16 Petersville Lancers 6 8 19 31:77 20:46 Disbanding 

17 

18 

Quincy Giants 

Ralston Regulars 

7 

3 

5 

11 

21 

19 

40:89 

37:74 

19:47 

17:49 

Demoted 

The horizontal lines for positions 3–5, 7–14, and 17 were made with 

the command \cline{1-9} while all the others used \hline: 

11 & Kingston Cowboys & 13 & 6 & 14 & 54&45 & 32&34 & 

Medium Teams \\ \cline{1-9} 

The last two rows of the table deserve a comment. The remark ‘Demoted’ is 

! vertically placed in the middle of the two rows. This is accomplished by typing 

18 & Ralston Regulars & 3 & 11 & 19 & 37&74 & 17&49 

& \raisebox{1.5ex}[0pt]{Demoted}\\ \hline

4.8. Tables 103 

The \raisebox command lifts the text ‘Demoted’ by 1.5 ex. If the optional 

argument [0pt] had been left out, this lifting of the box would have increased the 

total height of the last row by 1.5 ex. This would have resulted in correspondingly 

more vertical spacing between the horizontal line of row 17 and the text of row 

18. This additional spacing is suppressed by the optional argument height = 

[0pt]. (See Section 4.7.2 for a description of the \raisebox command.) 

Occasionally one wants to increase the vertical spacing between horizontal 

lines and enclosed text. The soccer table would look better if the 

heading were thus: 

1st Regional Soccer League — Final Results 2002/03 

Club W T L Goals Points Remarks 

This is done by inserting an invisible vertical rule, a strut (Section 4.7.6), 

into the heading text: 

\multicolumn{10}{|c|}{\rule[-3mm]{0mm}{8mm}\bfseries 1st 

Regional Soccer League --- Final Results 2002/03}\\ \hline 

The included rule has a width of 0 mm, which makes it invisible, 

extends 3 mm below the baseline, and is 8 mm high. It thus stretches 

5 mm (8 − 3) above the baseline. It effectively pushes the horizontal lines 

away from the baseline in both directions. If a row consists of more than 

one column, it is sufficient to include a strut in only one of them since 

the size of the whole row is determined by the largest column. 

Exercise 4.12: Produce your own table for the final results of your favorite team 

sport in the same manner as for the soccer results above. Watch out that the 

colons ‘:’ are properly aligned for the goals and points relationships. 

Exercise 4.13: Generate the following timetable. 

Day Subj. 

Mon. UNIX 

Tues. L AT E X 

Wed. UNIX 

Fri. L AT E X 

6.15–7.15 pm 7.20–8.20 pm 8.30–9.30 pm 

Teacher Teacher Teacher 

Subj. 

Subj. 

Room 

Room 

Room 

Dr. Smith Ms. Clarke Mr. Mills 

Fortran Math. 

Comp. Ctr 

Hall A 

Hall A 

Miss Baker Ms. Clarke Mr. Mills 

Conf. Room Fortran 

Math. 

Conf. Room 

Hall A 

Dr. Smith Dr. Jones Dr. Jones 

C 

ComSci. 

Comp. Ctr 

Hall B 

Hall B 

Miss Baker Ms. Clarke 

C++ 

Conf. Room 

Conf. Room 

canceled


The entries ‘Day’ and ‘Subj.’ are raised in the same way as ‘Demoted’ was in 

the soccer table. To simplify its application, one can introduce a user-defined 

command with 

\newcommand{\rb}[1]{\raisebox{1.5ex}[0pt]{#1}} (see Section 8.3.2) 

so that \rb{entry} behaves the same as \raisebox{1.5ex}[0pt]{entry}. This 

can be used, for example, as \rb{ Mon.} or \rb{UNIX} to elevate the entries by 

the necessary amount. 

In all the above examples, the entries in the individual columns are 

each a single line. Some tables contain certain columns with several lines 

of text that are somewhat separated from the rest of the row: 

Model Description Price 

FBD 360 Desktop: XP3600+ Processor, 512 MB DDR-RAM, 

80 GB hard disk, 16x DVD drive, 32x CDRW drive, 

64 MB TV output, Windows XP, 15” monitor 

FBD 480 Desktop DeLuxe: same as FBD 360 but with 

XP4800+ Processor, 48x CDRW drive, 17” monitor 

FBT 240 Laptop: XP2400+, 512 MB RAM, 40 GB hard disk, 

56 kb modem, 32x speed DVD/CD-RW drive, 2x USB, 

15” display, Windows XP, Infrared interface 

999.00 

1399.00 

1299.00 

The above table is made up of three columns, the first left justified, 

the third right justified. The middle column contains several lines of text 

with a line width of 7.5 cm. This is generated with the column formatting 

symbol p{width}. The whole column formatting argument in this example 

is {lp{7.5cm}r}. 

\begin{tabular}{lp{7.5cm}r} 

\bfseries Model & Description & \bfseries Price \\[1ex] 

FBD 360 &\small{\bfseries Desktop}: XP3600+ Processor, 

512˜MB DDR-RAM, 80˜GB hard disk, 16x DVD drive, 

32x CDRW drive, 64˜MB TVõutput, 

Windows˜XP, 15’’ monitor & 999.00\\ 

. . . . . . . . . . . . 

15’’ display, Windows˜XP, Infrared interface& 1299.00 

\end{tabular} 

The text for the middle column is simply typed in, being broken up 

into lines of width 8.0 cm automatically. The column is separated from 

the others with the & symbol in the usual way. 

Warning: The line termination command \\ is ambiguous within a p 

column, for it can either start a whole new row, or if \raggedright or 

\centering have been given, it ends a line of text within that column 

entry. In this case, if this is the last column in the row, the only way to

4.8. Tables 105 

terminate the row is with \tabularnewline which always starts a new 

row. 

Exercise 4.14: Produce the following table. 

Course and Date Brief Description Prerequisites 

Introduction to LSEDIT 

March 14 – 16 

Introduction to LAT E X 

March 21 – 25 

Logging on — explanation of the VMS file system 

— explanation and intensive application 

of the VMS editor LSEDIT — user modifications 

Word processors and formatting programs — 

text and commands — environments — document 

and page styles — displayed text — math 

equations — simple user-defined structures 

none 

LSEDIT 

The final example describes a blank form produced as a framed table. 

The difficulty here is to set the heights and widths of the empty boxes, 

since these are normally determined automatically by the text entries. 

The example shows how this may be accomplished with the help of struts 

and \hspace commands. 

Budget Plan 2003–2004 

Project Nr. Name 

Year 2003 2004 2005 

€ US $ € US $ € US $ 

Investment 

Costs 

Operating 

Costs 

Industrial 

Contracts 

Signature Authorization 

\newsavebox{\k}\newsavebox{\kkk} 

\sbox{\k}{\framebox[4mm]{\rule{0mm}{3mm}}} 

\sbox{\kkk}{\usebox{\k}\usebox{\k}\usebox{\k}} 

\begin{tabular} {|l|c|c|c|}\hline 

\multicolumn{4}{|c|}{\rule[-0.3cm]{0mm}{0.8cm}\bfseries 

Budget Plan 2003--2004}\\ 

\hline\hline 

\rule[-0.4cm]{0mm}{1cm}Project 

& \multicolumn{3}{l|}{Nr. \usebox{\kkk}\hspace{0.5cm} 

\vline\hspace{0.5cm}Name\usebox{\kkk}\usebox{\kkk}\usebox{\kkk} 

\usebox{\kk}}\\ \hline


\multicolumn{1}{|r|}{Year} & 2003 & 2004 & 2005 \\ 

\cline{2-4} 

& \euro\ \vline\ US \$ & \euro\ \vline\ US \$ 

& \euro\ \vline\ US \$ \\ \hline 

Investment & \hspace{2.5cm}& \hspace{2.5cm}& \hspace{2.5cm} \\ 

Costs & & & \\ \hline 

Operating & & & \\ 

Costs & & & \\ \hline 

Industrial& & & \\ 

Contracts & & & \\ \hline 

\multicolumn{4}{|l|}{\rule[-1.0cm]{0mm}{1.3cm}Signature 

\hspace{5cm}\vlineÃuthorization} \\ \hline 

\end{tabular} 

The first three lines are only indirectly related to the table construc- 

tion. They arrange for three empty boxes to be drawn when the 

command \usebox{\kkk} is given (see Section 4.7.1). 

Except for the command \hspace{2.5cm} to set the column widths 

of the last three columns and the command \vline to draw a vertical line 

within a column, this example contains nothing new that was not in the 

previous examples. It is only necessary to give a brief explanation of the 

last row in the table: 

The command \multicolumn{4}{|l|} merges all four table columns 

into one, in which the text is set flush with the left margin. This text 

consists first of a strut \rule[-12mm]{0mm}{15mm} that says the height 

of the last row begins 12 mm below the baseline and is a total of 15 mm 

high. Then, beginning at the left margin, comes the word Signature, 

followed 5 cm later by \vline, a vertical line. The word Authorization 

is separated from the vertical line by a blank space (˜). 

The above examples clearly illustrate how the column widths and row 

heights are automatically determined for tables. These sizes, however, 

may be influenced by struts and \hspace commands. In addition, the 

commands described in Section 4.8.2 permit the intercolumn and interrow 

spacings as well as line thickness to be altered. For example, 

\setlength{\tabcolsep}{5mm} 

inserts 5 mm of spacing before and after every column; that is, it produces 

an intercolumn spacing of 10 mm. Section 4.8.2 gives more information 

about the use of these table style parameters. 

4.8.4 Extension packages for tables 

As powerful as the tabular environment is, it does have limitations. For 

this reason, there are a number of tools packages (Section B.5.4) that add 

additional features for constructing tables.

4.8. Tables 107 

Package: The array package extends the normal functionality of the tabular 

array and array environments by adding several column formatting arguments, 

and by allowing the user to be able to define his or her own such arguments. 

m{wth} produces a column of width wth which is aligned vertically in the 

middle. (The standard p{wth} aligns the text with the top line.) 

b{wth} is like p and m but aligns the text on the bottom line. 

>{decl} inserts decl before the next column; thus >{\bfseries} sets the 

entire column in bold face without having to type \bfseries in each 

row. 

{$}c{$}c


Package: The delarray package loads the array package and redefines the 

delarray array environment so that it may be enclosed in braces that are automatically 

adjusted in size, as with the \left and \right commands of 

Section 5.4.1. The braces surround the column specifier. For example, 

 

a b 

\begin{array}[{cc}] a & b \\ c & d \end{array} ⇒ 

c d 

Package: 

longtable 

The longtable package produces tables extending over several pages. 

It does not require the array package, but does recognize its extra features 

if loaded. The longtable environment takes the same column formatting 

argument as tabular and array, but has additional row entries at the 

start to determine: 

• those rows that appear at the start of the table, terminated by 

\endfirsthead; this often includes the main \caption; 

• those at the top of every continuation page, terminated by \endhead; 

these normally include an additional \caption and the column 

headers; 

• those at the bottom of each page, terminated by \endfoot; 

• and those rows at the end of the table, terminated by \endlastfoot. 

An example of a long table is: 

\begin{longtable}{|l|c|r|} 

\caption[Short title]{Demonstration of a long table}\\ 

\hline 

Left & Center & Right \\ 

\hline \endfirsthead 

\caption[]{\emph{continued}}\\ 

\hline 

Left & Center & Right \\ 

\hline \endhead 

\hline 

\multicolumn{3}{r}{\emph{continued on next page}} 

\endfoot 

\hline\endlastfoot 

Twenty-two & fifty & A hundred and eighty \\ 

22 & 50 & 180 \\ 

. . . . . . 

\end{longtable} 

The \caption command normally may only appear within table and 

figure environments (Section 7.4) but may also be used within longtable 

which never goes into a table environment; in a continued row, \caption 

must have an empty optional argument [] to prevent multiple entries in 

the list of tables. 

A \newpage command forces a new page within the long table. 

Up to four L AT E X runs may be needed to get the column widths right.

Primary Energy Consumption 

4.8. Tables 109 

Energy Source 1975 1980 1986 

Total Consumption 

(in million tons of BCU a ) 347.7 390.2 385.0 

of which (percentages) 

petroleum 52.1 47.6 43.2 

bituminous coal 19.1 19.8 20.0 

brown coal 9.9 10.0 8.6 

natural gas 14.2 16.5 15.1 

nuclear energy 2.0 3.7 10.1 

other b 2.7 2.3 3.0 

a BCU = Bituminous Coal Unit (1 ton BCU corresponds to the heating equivalent of 1 ton 

of bituminous coal = 8140 kwh) 

b Wind, water, solar energy, etc. 

Source: Energy Balance Study Group, Essen 1987. 

4.8.5 Floating tables 

In Chapter 7 we explain how figures and tables can be made to float, that 

is, to be moved to the top or bottom of a page, or to a separate page 

altogether. The reason for doing this is that the author does not know 

in advance where the page breaks will occur, and so does not know if 

there is enough room on the current page for such a large object. The 

float mechanism allows the material to be saved and then placed later at 

an appropriate location. It also allows for table and figure captions and 

automatic numbering. 

Table material is made to float with the environment: 

\begin{table} head text table foot text \end{table} 

where table stands for the entire table as defined in a tabular environment, 

head text for whatever text appears above the table, and foot text 

for that below. Widths, spacing, and positioning of the texts relative to 

the table are all matters for the user to arrange. 

Independently of the enclosed text, everything that appears between 

\begin{table} and \end{table} is normally placed at the start of the 

current page. If a table already occupies the top of the page, an attempt 

is made to place it at the page bottom, if there is enough space for it. 

Otherwise, it will be placed on the next page, where further tables may be 

accumulated. The surrounding text is printed as though the table were 

not there. For further details about floats in general, including automatic 

sequential numbering, see Chapter 7. 

The table at the top of this page was generated within the text at this 

location with the following (excluding the footnotes, which are described 

later in Section 4.10.4):


\begin{table} {\bfseries Primary Energy Consumption}\\[1ex] 

\begin{tabular*}{118mm}{@{}ll...rr@{}} 

. . . . . . . . . . . . . . . . . . . . 

\end{tabular*}\\[0.5ex] 

\emph{Source:} Energy Balance Study Group, . . . 

\end{table} 

There are a number of formatting parameters that may be used in 

connection with the table environment, which are described together 

with those for figures in Section 7.3. 

Exercise 4.15: Complete the above text for the table on the previous page (without 

the footnotes). Pay attention to the following questions (check the explanations 

for the @-expressions in Section 4.8.1): 

1. What is the effect of the @{} entries at the beginning and end of the 

formatting definition? 

2. The tabular* environment generates a table with a given width, here 

118 mm. What would be the effect of @{extracolsep{\fill}} at the 

beginning of the formatting definition? 

3. Where in the formatting definition should @{extracolsep{\fill}} and 

the countermanding @{\hspace{1em}}@{\extracolsep{1em}} appear in 

order to format the table as it is printed here? How would the table appear 

if only @{\extracolsep{1em}} were given as countermand? 

4.9 Printing literal text 

Occasionally it is necessary to print text exactly as it is typed, with all 

special characters, blanks, and line breaks appearing literally, unformatted, 

and in a typewriter font. Lines of computer code or samples of L AT E X 

input text are examples of such literal text. This is accomplished with the 

environments 

\begin{verbatim} text \end{verbatim} 

\begin{verbatim*} text \end{verbatim*} 

A new line is inserted before and after these environments. 

With the *-form, blanks are printed with the symbol to make them 

visible. 

As an example, on page 115 some input text is printed to demonstrate 

the use of footnotes in forbidden modes. This was done with 

\begin{verbatim} 

\addtocounter{footnote}{-1}\footnotetext{Small insects} 

\stepcounter{footnote}\footnotetext{Large mammals} 

\end{verbatim} 

Literal text may also be printed within a line using the commands 

\verb and \verb*, as for example

4.9. Printing literal text 111 

\verb=\emph{words of text}= \emph{words of text} 

\verb*=\emph{words of text}= \emph{words of text} 

where the first character after \verb or \verb* (here =) is the delimiter, 

such that all text up to the next occurrence of that character is printed 

literally. This character may not appear in the literal text, obviously. 

In contrast to the behavior in the verbatim environment, the literal 

text must be all on one line in the input text, otherwise an error message 

is printed. This is to indicate that you just might have forgotten to repeat 

the delimiting character. 

Important: neither the verbatim environment nor the \verb command 

may be used in an argument of any other command! 

Exercise 4.16: Reproduce some input lines from this book as literal text. 

4.9.1 Extension packages for literal text 

Package: The standard package alltt (Section B.5.3, page 392) provides an alltt 

alltt environment that also prints its contents literally in a typewriter font, 

except that the characters \ { } retain their normal meaning. Thus LAT EX commands can be included within the literal text. For example, 

\begin{alltt} 

Underlining \underline{typewriter} 

text is also possible. 

Note that dollar ($) and 

percent (%) signs are 

treated \emph{literally}. 

\end{alltt} 

Underlining typewriter 

text is also possible. 

Note that dollar () and 

percent (%) signs are 

treated literally. 

Package: The standard package shortvrb (Section B.5.3, page 393) offers a 

shortvrb shorthand for the \verb command. After issuing \MakeShortVerb{\|}, 

one can print short literal text with |text|. The counter command 

\DeleteShortVerb{\|} then restores the original meaning to |. Any 

character may be temporarily turned into a literal switch this way. 

Package: One problem with the verbatim environment is that the entire litverbatim 

eral text is input and stored before processing, something that can lead 

to memory overflows. The verbatim package in the tools collection 

(Section B.5.4, page 396) re-implements the environment to avoid this 

problem. A minor drawback is that there must not be any other text on 

the same line as the \end{verbatim}. 

The verbatim package offers two other extra features. It provides 

a comment environment that simply ignores its contents, as though 

each line started with a % sign (Section 4.11). And it adds a command 

\verbatiminput{filename} to input the specified file as literal text. This 

is useful for listing actual computer programs rather than copying them 

into the LAT EX file.


4.9.2 Email and Internet addresses 

Package: Email and Internet addresses present some special problems that are 

url solved with the url package by Donald Arseneau. These addresses are 

best listed in a typewriter font, often contain special symbols, and should 

never be hyphenated, since the hyphen could be interpreted as part of 

the address. The obvious solution would be to use the \verb command, 

which fulfills all these conditions. However, if the address does not fit 

on the current line, it will stick out into the right margin. Using \texttt 

instead will allow the words in the address to be broken over two lines, 

but at the price of inserting an ambiguous hyphen. 

Inserting the address with the \url command allows it to be broken at 

non-letters between words, without a hyphen. Moreover, its argument is 

treated literally, so all special characters are printed as given, just as with 

\verb. In fact, the argument of the \url command may either be enclosed 

in curly braces as usual, or delimited by some arbitrary character, just as 

with \verb. 

An Internet address is given as 

An Internet address may be given as 

http://address.edu/home/ 

\url{http://address.edu/home/page/} 

page/ or an email address 

or an email address might be given as 

might be given as fred.smith@ 

\url=fred.smith@general.services.gov= 

general.services.gov 

The \url command is to be preferred for another reason: it conforms 

to logical markup indicating the purpose of its argument. In fact, the 

hyperref package of Section 10.2.4 will even turn the argument into an 

active link, something it does not do for \verb arguments. 

The printed appearance of the address can be controlled by specifying 

\urlstyle{style}, where style is one of tt (default), rm, sf, or same, for 

typewriter, Roman, sans serif, or unchanged font, respectively. 

A fixed address can be predefined with, e.g., 

\urldef{\myurl}\url{myname@mydomain.uk} 

Now \myurl is used to print myname@mydomain.uk. 

There is also a \path command for giving directory/folder names, with 

the same syntax as \url. As logical markup, it has a different significance; 

for example, the hyperref package does not turn it into a link. 

For further information on advanced uses of this package, see the 

comments at the end of the file url.sty. 

4.10 Footnotes and marginal notes 

4.10.1 Standard footnotes 

Footnotes are generated with the command 

\footnote{footnote text}

4.10. Footnotes and marginal notes 113 

which comes immediately after the word requiring an explanation in a 

footnote. The text footnote text appears as a footnote in a smaller typeface 

at the bottom of the page. The first line of the footnote is indented and 

is given the same footnote marker as that inserted in the main text. The 

first footnote on a page is separated from the rest of the page text by 

means of a short horizontal line. 

The standard footnote marker is a small, raised number 1 , which is 

sequentially numbered. This footnote is produced with: 

... raised number\footnote{The usual method of marking 

footnotes in a typewritten ... same page.}, which is ... 

The footnote numbering is incremented throughout the whole document 

for the article class, whereas it is reset to 1 for each new chapter 

in the report and book classes. 

The \footnote command may only be given within the normal paragraph 

mode, and not within math or LR modes. In practice, this means it 

may not appear within an LR box (Section 4.7.1) or a parbox (Section 4.7.3). 

However, it may be used within a minipage environment, in which case 

the footnote text is printed beneath the minipage and not at the bottom 

of the actual page. 2 

The \footnote command must immediately follow the word that is to 

receive the note, without any intervening blanks or spacing. A footnote at 

the end of a sentence can be given after the period, as in the last example 

above: 

... of the actual page.\footnote{With nested ... wrong place.} 

4.10.2 Non-standard footnotes 

If the user wishes the footnote numbering to be reset to 1 for each 

\section command with the article class, this may be achieved with 

\setcounter{footnote}{0} 

just before or after a \section command. 

The internal footnote counter has the name footnote. Each call to 

\footnote increments this counter by one and prints the new value in 

Arabic numbering as the footnote marker. A different style of marker can 

be implemented with the command 

\renewcommand{\thefootnote}{\number style{footnote}} 

1 The usual method of marking footnotes in a typewritten manuscript with *, **, etc., 

could also be done; however, since the page breaks are not known at the time of typing the 

text, there would be a problem of avoiding duplication of the symbols on the same page. 

2 With nested minipages, the footnote comes after the next \end{minipage} command, 

which could be at the wrong place.


where number style is one of the counter print commands described in 

Section 4.3.5: \arabic, \roman, \Roman, \alph, or \Alph. However, 

for the counter footnote, there is an additional counter print command 

available, \fnsymbol, which prints the counter values 1–9 as one of nine 

symbols: 

* † ‡ § ** †† ‡‡ 

It is up to the user to see that the footnote counter is reset to zero 

sometime before the tenth \footnote call. 

An optional argument may be added to the \footnote command 

\footnote[num]{footnote text} 

where num is a positive integer that is used instead of the value of the 

footnote counter for the marker. In this case, the footnote counter is not 

incremented. For example ∗∗ , 

\renewcommand{\thefootnote}{\fnsymbol{footnote}} 

For example\footnote[7]{The 7th symbol ... marker.}, 

\renewcommand{\thefootnote}{\arabic{footnote}} 

where the last line is necessary to restore the footnote marker style to 

its standard form. Otherwise, all future footnotes would be marked with 

symbols and not with numbers. 

4.10.3 Footnotes in forbidden modes 

A footnote marker can be inserted in the text with the command 

\footnotemark[num] 

even where the \footnote command is normally not allowed, that is, 

in LR boxes, tables, and math mode. The marker is either the optional 

argument num or, if it is omitted, the incremented value of the footnote 

counter. The footnote itself is not generated. This must be done external 

to the forbidden mode with the command 

\footnotetext[num]{footnote text} 

If the optional argument has been used for the footnote marker, the same 

num must be given as the option for the text command. Similarly, if no 

option was used for the marker, none may appear with the text. The 

footnote will be generated with the value of num or with that of the 

footnote counter. 

This counter is incremented by a call to \footnotemark without an 

optional argument. The corresponding \footnotetext command, on the 

other hand, does not alter the counter. 

∗∗ The 7th symbol appears as the footnote marker.


If there are a number of \footnotemark commands without optional 

arguments appearing before the next \footnotetext command, it is 

necessary to adjust the counter with the command 

\addtocounter{footnote}{dif } 

where dif is a negative number saying how many times the counter must 

be set back. Then before every \footnotetext command, the counter 

must be incremented by one. This can be done either with the command 

\addtocounter, with dif =1, or with the command 

\stepcounter{footnote} 

which adds 1 to the given counter. 

For example: mosquitoes 3 and elephants 4 

For example: \fbox{mosquitoes\footnotemark\ and 

elephants\footnotemark} 

generates the footnote markers 3 and 4 . Now the counter has the value 4. 

In order for the first \footnotetext outside the framed box to operate 

with the correct counter value, it must first be decremented by one. The 

two footnote texts are made with 

\addtocounter{footnote}{-1}\footnotetext{Small insects} 

\stepcounter{footnote}\footnotetext{Large mammals} 

immediately following the \fbox{} command. The footnote counter now 

has the same value as it did on leaving the \fbox. 

4.10.4 Footnotes in minipages 

As mentioned in Section 4.10.1, footnote commands are allowed inside 

the minipage environment. However, the footnote appears underneath 

the minipage, not below the main page. 

Footnote commands within a minipage 

a have a different marker style. 

The footnote comes after the next 

\end{minipage} command. b Minipage 

footnotes have a counter separate from 

that of the main page, called mpfootnote, 

counting independently of footnote. 

a The marker is a raised lower-case letter. 

b Watch out for nested minipages. 

\begin{minipage}{6cm} 

Footnote commands within 

a minipage\footnote{The 

marker is a raised 

lower-case letter.} have 

a different... 


Footnotes within a tabular environment can normally only be generated 

! with the commands described above: \footnotemark within the table and 

3 Small insects 

4 Large mammals


\footnotetext outside the environment. However, if the tabular environment 

is inside a minipage, normal \footnote commands may also be used inside the 

table. The footnote appears below the table where the minipage comes to an end. 

Exercise 4.17: Produce a number of footnotes in your standard exercise file by 

inserting them where you think fit and by selecting some appropriate footnote 

text. 

Exercise 4.18: Redefine the command \thefootnote so that the footnote markers 

become the symbols illustrated in Section 4.10.2. Add the redefinition to the 

preamble of your standard exercise file. 

Exercise 4.19: Complete Exercise 4.15 so that the footnotes a and b appear as in 

the table on page 109. 

4.10.5 Marginal notes 

This 

is a 

marginal 

note 

=⇒ 

Notes in the page margin are produced with the command 

\marginpar{note text} 

which puts the text note text into the margin beginning at the level of the 

line where the command is given. The marginal note appearing here was 

generated with 

... The marginal note \marginpar{This\\ is a\\ margin-\\al note} 

appearing here ... 

The text is normally enclosed in a parbox of width 1.9 cm (0.75 in). 

Such a narrow box causes great difficulties with line breaking, which is 

why the lines are manually broken with the \\ command in the above 

example. Such a box is far more appropriate for marginal notes in the 

form of a single symbol, such as the arrow shown here. 

Another common use for the marginal note is to draw attention to 

certain text passages by marking them with a vertical bar in the margin. 

This is often done to indicate changes in a text, for comparison with 

earlier versions after updated single sheets have been redistributed. The 

example marking this paragraph was made by including 

\marginpar{\rule[-17.5mm]{1mm}{20mm}} 

in the first line. 

The width of the marginal note can be changed with a style parameter 

described in the next section. The user must watch out that the total page 

width does not become too big for the printer. 

By default, marginal notes appear in the right-hand margin of the page, 

or in the outer margin when the twoside option has been selected. ‘Outer’ 

means the right margin for odd pages, and left for even ones. With the 

twocolumn option, they are placed in the outside margins: left for the left 

column and right for the right one.


This leads to a problem for marginal markings such as the arrow 

illustrated on the previous page. On this page, it must point in the 

opposite direction. In fact, its direction depends on which side of the ⇐= 

page it is to appear, and that in turns depends on the page number or 

column. Since these are not known at the time of writing (and may even 

change with later revisions) it is necessary to have another solution. This 

is provided by the extended syntax of the \marginpar command 

\marginpar[left text]{right text} 

This form of the command contains two versions of the marginal text, 

left text to go into the left margin, and right text for the right margin, 

depending on which one is selected. Both the arrows on this and the 

previous page were generated with the same command 

\marginpar[\hfill$\Longrightarrow$]{$\Longleftarrow$} 

(The mathematical arrow commands are explained in Section 5.3.5.) 

Without the \hfill command in the above \marginpar example, the arrow 

! in the left margin appears as it does at the side of this paragraph, too far over to 

the left. The reason for this is that the \marginpar command sets its contents 

flush with the left edge of the narrow margin box, made visible here with a frame. 

This left edge is aligned with the main text only when the note is put on the right 

=⇒ side; however, in the left margin, it is displaced from the main text. The \hfill 

command has the effect of setting the contents flush with the right edge of the 

margin box, which is then properly aligned with the main text. 

A similar device was used to make the first marginal note in this section. The 

actual command given was 

=⇒ 

\marginpar[\flushright This\\ is a\\ margin-\\al note] 

{This\\ is a\\ margin-\\al note} 

Here \flushright (Section 4.2.2) is equivalent to putting an \hfill on each line. 

The standard positioning of the marginal notes can be switched with 

the command \reversemarginpar. Once this command has been given, 

marginal notes will appear in the left margin, or in the ‘inner’ margin for 

the twoside option. The command \normalmarginpar restores normal 

behavior. These commands have no effect with the twocolumn option. 

If the note appears at the bottom of a page, it will extend downwards 

below the last line of regular text. For this reason, and because of the 

difficulties with line breaking for narrow columns, marginal notes should 

be kept short, limited to a few words or a symbol. 

4.10.6 Style parameters for footnotes and marginal notes 

There are two footnote style parameters that may be changed as needed, either 

! in the preamble or locally within an environment. 

\footnotesep 

The vertical spacing between two footnotes. This is a length that can 

be changed with the \setlength command.


\footnoterule 

The command that draws a horizontal line between the page text and 

the footnotes. It should not add any net vertical spacing. It may be 

changed, for example, by 

\renewcommand{\footnoterule} 

{\rule{wth}{hght}\vspace{-hght}} 

A value of 0 cm for the hght produces an invisible line of zero thickness. 

The following style parameters may be changed to redefine how marginal 

notes appear: 

\marginparwidth 

determines the width of the margin box; 

\marginparsep 

sets the separation between the margin box and the main text; 

\marginparpush 

is the smallest vertical distance between two marginal notes. 

These parameters are all lengths and are assigned new values as usual with the 

\setlength command. 

4.11 Comments within text 

All computer languages provide a means of inserting comments into the 

code. These are explanatory notes, documentation, history of development, 

or alternative text or code that has been temporarily deactivated. 

Comment lines are completely ignored during processing. They are only 

intended for human readers inspecting the source text. 

In L AT E X, the comment character is the percent sign %. When this 

character appears in the text, it and the rest of the line are ignored. If a 

comment is several lines long, each line must be prefixed with %. 

As for other single character commands, the percent sign itself is 

printed with the command \%, as explained in Section 2.5.4. 

The comment character % is also useful for experimenting with text 

or definitions of user commands or formatting parameters, to try alternatives 

without deleting the old versions. By ‘commenting out’ selective 

lines, one can play around with variations without losing them. 

Large sections of text may be more effectively commented out with the 

comment environment from the verbatim package (page 111). 

Finally, the % character has an important role to play in suppressing 

implied blanks at the end of a line. This is especially desirable in user definitions 

where unexpected blanks can creep in between otherwise invisible 

declarations with arguments. See Section 8.5.2. 

Exercise 4.20: Comment out the changes from Exercise 4.18 in your preamble. 

These commands may be reactivated later by removing the % character.

5 

Mathematical Formulas 

Mathematics is the soul of T E X. It was because the setting of mathematical 

formulas is so complicated in normal printing, not to mention on a 

typewriter, that Donald Knuth invented his text formatting system. On 

the other hand, the soul of L AT E X is logical markup. Nevertheless, all the 

power of T E X’s math setting is also available in L AT E X, offering an unbeatable 

combination. 

In this chapter, we confine ourselves to the elements of mathematical 

typesetting available in standard L AT E X. The simplifications and additional 

elements provided by A M S-L AT E X are reserved for Chapter 12. 

Mathematical formulas are produced by typing special descriptive 

text. This means that L AT E X must be informed that the following text is 

to be interpreted as a mathematical formula, and it must also be told 

when the math text has come to an end and normal text recommences. 

The processing of math text is carried out by switching to math mode. 

Mathematical environments serve this purpose. 

5.1 Mathematical environments 

Mathematical formulas may occur within a line of text, as (a + b) 2 = 

a 2 + 2ab + b 2 , or separated from the main text as 

∞ 

0 

f (x) dx ≈ 

n 

wie xif (xi) 

These two types are distinguished by referring to them as text and displayed 

formulas respectively. 

Text formulas, or equations, are generated with the environment 

i=1 

\begin{math} formula text \end{math} 

Since text formulas are often very short, sometimes consisting of only a 

single character, a shorthand version is available as $ formula text $. 

119

120 Chapter 5. Mathematical Formulas 

However, most authors prefer the very short form $formula text $ which 

is actually the T E X method. All three are essentially the same, and there 

is no reason not to use the $ sign. 

The contents of the formula, formula text, consist of math constructs, 

which are described in the following sections. 

Displayed formulas, or equations, are produced in the environments 

\begin{displaymath} formula text \end{displaymath} 

\begin{equation} formula text \end{equation} 

The difference between these two is that the equation environment automatically 

adds a sequential equation number. The displaymath environment 

may be given with the shorthand forms \[ . . . \] or $$. . . $$. 

By default displayed formulas are centered horizontally with the equation 

number, if it is present, set flush with the right margin. By selecting 

the document class option fleqn (Section 3.1.1), the formulas are set left 

justified with an adjustable indentation. This option remains valid for the 

entire document whereas the amount of indentation may be changed at 

will with \setlength{\mathindent}{indent}, where indent is a length 

specification. Moreover, the document class option leqno sets the equation 

numbers flush with the left margin throughout the whole document. 

Finally, multiline formulas can be created with the environments 

\begin{eqnarray} formula text \end{eqnarray} 

\begin{eqnarray*} formula text \end{eqnarray*} 

where the standard form adds a sequential equation number for each line 

and the *-form is without equation numbers. 

5.2 Main elements of math mode 

5.2.1 Constants and variables 

Numbers that appear within formulas are called constants, whereas simple 

variables are represented by single letters. The universal practice 

in mathematical typesetting is to put constants in Roman typeface and 

variables in italics. L AT E X adheres to this rule automatically in math mode. 

Blanks are totally ignored and are included in the input text simply to improve 

the appearance for the writer. Spacing between constants, variables, 

and operators like +, −, = are set automatically by L AT E X. For example 

$z=2a+3y$, $ z = 2 a + 3 y $ both produce z = 2a + 3y. 

Mathematical symbols that are available on the keyboard are 

+ - = < > / : ! ’ | [ ] ( ) 

all of which may be used directly in formulas. The curly braces { } serve 

the purpose of logically combining parts of the formula and therefore

5.2. Main elements of math mode 121 

cannot act as printable characters. To include braces in a formula, the 

same commands \{ and \} are used as in normal text. 

M(s) < M(t) < |M| = m $M(s)


5.2.3 Fractions 

Short fractions, especially within a text formula, are best represented 

using the slash character /, as in $(n+m)/2$ for (n + m)/2. For more 

complicated fractions, the command 

\frac{numerator}{denominator} 

is employed to write the numerator on top of the denominator with a 

horizontal fraction line of the right width between them. 

1 

x + y 

\[ \frac{1}{x+y} \] 

a 2 − b 2 

5.2.4 Roots 

a + b = a − b \[ \frac{aˆ2 - bˆ2}{a+b} = a-b \] 

Fractions may be nested to any depth within one another. 

a b 

x−y + x+y 

1 + a−b 

a+b 

\[ \frac{\frac{a}{x-y} + \frac{b}{x+y}} 

{1 + \frac{a-b}{a+b}} \] 

L AT E X sets fractions within fractions in a smaller typeface. Section 5.5.2 

describes how the automatic type sizes may be overridden if L AT E X’s choice 

is unsuitable. 

Roots are printed with the command 

\sqrt[n]{arg} 

as in the example: $\sqrt[3]{8} = 2$ produces 3√ 8 = 2. If the optional 

argument n is omitted, the square root is generated: $\sqrt{a} yields 

√ 

a. 

The size and length of the root signare automatically fitted to arg: 

$\sqrt{xˆ2 + yˆ2 + 2xy} = x+y$ x2 + y2 + 2xy = x + y, or 

 

n 

x n − y n 

1 + u 2n 

\[ \sqrt[n]{\frac{xˆn - yˆn}{1 + uˆ{2n}}} \] 

Roots may be nested inside one another to any depth: 

 

3 

−q + 

 

q2 + p3 \[ \sqrt[3]{-q + \sqrt{qˆ2 + pˆ3}} \]

5.2.5 Sums and integrals 

5.2. Main elements of math mode 123 

Summation and integral signs are made with the two commands \sum 

and \int, which may appear in two different sizes depending on whether 

they occur in a text or displayed formula. 

Sums and integrals very often possess upper and lower limits. These 

are printed with the exponent and index commands ˆ and _. The positioning 

of the limits also depends on whether the formula is in text or 

displayed. 

In a text formula \sum_{i=1}ˆn and \int_aˆb produce n i=1 and b 

a , 

whereas in a displayed formula they appear as at the left below: 

n 

i=1 

b 

a 

Some authors prefer the limits for the integral to be 

placed above and below the integral sign, the same 

as for summation. This is achieved with the command 

\limits immediately following the integral 

sign: \int\limits_{x=0}ˆ{x=1} 

The rest of the formula text coming before and after the sum and 

integral signs is correctly aligned with them. 

2 

n 

i=1 

ai 

b 

a 

fi(x)gi(x) dx 

x=1 

x=0 

\[ 2\sum_{i=1}ˆn a_i \intˆb_a 

f_i(x)g_i(x)\,\mathrm{d}x \] 

Two points must be made regarding integrals such as y dx and f (z) dz. 

! First, there should be a little extra spacing between the differential operators 

dx and dz and the integrands preceding them. This is achieved with the small 

spacing command \, mentioned in Section 2.7.1, not with a blank which is 

ignored in math mode. Second, the differential operator should be written 

upright, not italic, as explained in Section 5.4.10. This is accomplished by setting 

it within a \mathrm command (Section 5.4.2). Thus \int y\,\mathrm{d}x and 

\int f{z}\,\mathrm{d}z produce the desired results shown above, whereas 

\int y dx and \int f(z) dz yield ydx and f (z)dz. 

5.2.6 Continuation dots—ellipsis 

Formulas occasionally contain a row of dots . . . , meaning and so on. 

Simply typing a period three times in a row produces an undesirable 

result: ..., that is, the dots are too close together. Therefore L AT E X provides 

several commands 

\ldots . . . low dots \cdots · · · center dots 

\vdots 

. vertical dots \ddots 

. .. 

diagonal dots 

to space the dots correctly. The difference between the first two commands 

is best illustrated by the examples a0, a1, . . . , an and a0 + a1 + 

· · · + an, which are produced with $a_0,a_1,\ldots,a_n$ for the first 

and $a_0 + a_1 + \cdots + a_n$ for the second.


The command \ldots is also available in normal text mode, whereas 

the other three are only allowed in math mode. In text mode, the command 

\dots may be used in place of \ldots with the same effect. 

Exercise 5.2: Generate the following output: 

The reduced cubic equation y 3 + 3py + 2q = 0 has one real and two complex 

solutions when D = q 2 + p 3 > 0. These are given by Cardan’s formula as 

where 

u + v 

y1 = u + v, y2 = − 

2 

i u + v 

+ 3(u − v), y3 = − 

2 

2 

u = 3 

 

 

−q + q2 + p3 , v = 3 

 

 

−q − q2 + p3 i 

− 3(u − v) 

2 

Note: the spacings between the parts of the displayed equations are made with 

the spacing commands \quad and \qquad. 

Exercise 5.3: Select the option fleqn in the document class command and include 

the specification \setlength{\mathindent}{2cm} in the preamble. Redo 

the three y equations above, each as a separate displayed formula, using the 

equation environment instead of displaymath or \[. . . \] brackets. 

Exercise 5.4: Create the following text: 

Each of the measurements x1 < x2 < · · · < xr occurs p1, p2, . . . , pr times. The 

mean value and standard deviation are then 

 

x = 1 

n 

r 

 

 

pixi, s = 1 

n 

i=1 

where n = p1 + p2 + · · · + pr . 

r 

pi(xi − x) 2 

Exercise 5.5: Although this equation looks very complicated, it should not 

present any great difficulties: 

(ax + b) 3 

x 

dx = 2 (ax + b) 3 

3 

i=1 

 

+ 2b ax + b + b 2 

 

The same applies to 8 

−1 (dx/ 3√ x) = 3 

2 (82/3 + 1 2/3 ) = 15/2. 

5.3 Mathematical symbols 

dx 

x √ ax + b 

There is a very wide range of symbols used in mathematical text, of which 

only a few are directly available from the keyboard. L AT E X provides many of 

the mathematical symbols that are commonly used. They are called with 

the symbol name prefixed with the \ character. The names themselves 

are derived from their mathematical meanings.

5.3.1 Greek letters 

5.3. Mathematical symbols 125 

Lower case letters 

α \alpha θ \theta o o τ \tau 

β \beta ϑ \vartheta π \pi υ \upsilon 

γ \gamma ι \iota ϖ \varpi φ \phi 

δ \delta κ \kappa ρ \rho ϕ \varphi 

ɛ \epsilon λ \lambda ϱ \varrho χ \chi 

ε \varepsilon µ \mu σ \sigma ψ \psi 

ζ \zeta ν \nu ς \varsigma ω \omega 

η \eta ξ \xi 

Upper case letters 

Γ \Gamma Λ \Lambda Σ \Sigma Ψ \Psi 

∆ \Delta Ξ \Xi Υ \Upsilon Ω \Omega 

Θ \Theta Π \Pi Φ \Phi 

The Greek letters are made simply by putting the command character \ 

before the name of the letter. Upper case (capital) letters are distinguished 

by capitalizing the first letter of the name. Greek letters that do not appear 

in the above list are identical with some corresponding Latin letter. For 

example, upper case ρ is the same as Latin P and so needs no special 

symbol. 

L AT E X normally sets the upper case Greek letters in Roman (upright) 

type within a mathematical formula. If they need to be in italics, this 

can be brought about with the math alphabet command \mathnormal: 

$\mathnormal{\Gamma\Pi\Phi}$ appears as Γ ΠΦ. 

Greek letters may only be used in math mode. If they are needed in 

normal text, the command must be enclosed in $. . . $ signs. 

5.3.2 Calligraphic letters 

The following 26 calligraphic letters may also be used in math formulas: 

A, B, C, D, E, F, G, H , I, J, K, L, M, N , O, P, Q, R, S, T, U, V, W, X, Y, Z 

These are called with the math alphabet command \mathcal: 

$\mathcal{A, B, C,...,Z}$ 

5.3.3 Binary operators 

Two mathematical quantities combined with one another to make a new 

quantity are said to be joined by a binary operation. The symbols that are 

available for use as binary operators are


Package: 

latexsym 

amsfonts 

± \pm ∩ \cap ◦ \circ ○ \bigcirc 

∓ \mp ∪ \cup • \bullet \Box 

× \times ⊎ \uplus ⋄ \diamond ♦ \Diamond 

÷ \div ⊓ \sqcap ⊳ \lhd △ \bigtriangleup 

· \cdot ⊔ \sqcup ⊲ \rhd ▽ \bigtriangledown 

∗ \ast ∨ \vee ⊴ \unlhd ⊳ \triangleleft 

⋆ \star ∧ \wedge ⊵ \unrhd ⊲ \triangleright 

† \dagger ⊕ \oplus ⊘ \oslash \ \setminus 

‡ \ddagger ⊖ \ominus ⊙ \odot ≀ \wr 

∐ \amalg ⊗ \otimes 

The underlined symbol names in the above and following tables are 

only available if one of the packages latexsym (Section B.5.3) or amsfonts 

(Section 12.4.4) has been loaded. 

5.3.4 Relations and their negations 

When two mathematical quantities are compared, they are connected 

by a relation. The different types of relational symbols for the various 

comparisons are 

≤ \le 

≪ \ll 

\leq ≥ \ge 

≫ \gg 

\geq ≠ \neq 

. 

= \doteq 

∼ \sim 

\simeq 

⊂ \subset ⊃ \supset ≈ \approx ≍ \asymp 

⊆ \subseteq ⊇ \supseteq ≅ \cong ⌣ \smile 

⊏ \sqsubset ⊐ \sqsupset ≡ \equiv ⌢ \frown 

⊑ \sqsubseteq ⊒ \sqsupseteq ∝ \propto ⋈ \bowtie 

∈ \in ∋ \ni ≺ \prec ≻ \succ 

⊢ \vdash ⊣ \dashv \preceq \succeq 

⊨ \models ⊥ \perp \parallel \| | \mid | 

A number of the above symbols may be called by more than one name. 

For example, ≤ may be produced with either \le or \leq. 

The opposite, or negated, meaning of the relation is indicated in mathematics 

with a slash / through the symbol: = and = mean equals and 

not equals. One may put a slash through any of the above symbols by 

prefixing its name with \not. Thus \not\in yields ∈. The same is true 

for the keyboard characters: \not=, \not>, and \not< produce =, >, and 


< \not< > \not> = \not= 

≤ \not\le ≥ \not\ge ≡ \not\equiv 

≺ \not\prec ≻ \not\succ ∼ \not\sim 

\not\preceq \not\succeq \not\simeq 

⊂ \not\subset ⊃ \not\supset ≈ \not\approx 

⊆ \not\subseteq ⊇ \not\supseteq ≅ \not\cong 

⊑ \not\sqsubseteq ⊒ \not\sqsupseteq ≍ \not\asymp 

∈ \not\in ∉ \notin 

5.3.5 Arrows and pointers 

Mathematical manuscripts often contain arrow symbols, also called pointers. 

The following arrow symbols are available: 

← \leftarrow \gets ←− \longleftarrow ↑ \uparrow 

⇐ \Leftarrow ⇐= \Longleftarrow ⇑ \Uparrow 

→ \rightarrow \to −→ \longrightarrow ↓ \downarrow 

⇒ \Rightarrow =⇒ \Longrightarrow ⇓ \Downarrow 

↔ \leftrightarrow ←→ \longleftrightarrow \updownarrow 

⇔ \Leftrightarrow ⇐⇒ \Longleftrightarrow \Updownarrow 

↦→ \mapsto ↦−→ \longmapsto ↗ \nearrow 

←↪ \hookleftarrow ↩→ \hookrightarrow ↘ \searrow 

↼ \leftharpoonup ⇀ \rightharpoonup ↙ \swarrow 

↽ \leftharpoondown ⇁ \rightharpoondown ↖ \nwarrow 

⇌ \rightleftharpoons ⇝ \leadsto 

Here again the symbols → and ← may also be referred to under the names 

\to and \gets. Furthermore, the command \Longleftrightarrow may 

be substituted by \iff, although the latter ( ⇐⇒ ) has a little more spacing 

on either side than the former (⇐⇒). 

5.3.6 Various other symbols 

The above lists by no means exhaust the complete repertoire of mathematical 

symbols. However, the following are additional characters that 

standard L AT E X does make available. (Even more symbols are possible with 

the A M S symbol fonts and amssymb package, Section 12.4.4.) 

ℵ \aleph ′ \prime ∀ \forall \Box 

ℏ \hbar ∅ \emptyset ∃ \exists ♦ \Diamond 

ı \imath ∇ \nabla ¬ \neg △ \triangle 

j \jmath √ \surd ♭ \flat ♣ \clubsuit 

ℓ \ell ∂ \partial ♮ \natural ♦ \diamondsuit 

℘ \wp ⊤ \top ♯ \sharp ♥ \heartsuit 

ℜ \Re ⊥ \bot \| ♠ \spadesuit 

ℑ \Im ⊢ \vdash ∠ \angle ⋈ \Join 

\mho ⊣ \dashv \ \backslash ∞ \infty


5.3.7 Symbols with two sizes 

The following symbols are printed in different sizes depending on whether 

they appear in text or displayed formulas: 

 

\sum 

\bigcap 

\bigodot 

 

 

 

\int 

\bigcup 

\bigotimes 

 

 

 

\oint 

\bigsqcup \bigoplus 

 

\prod 

\bigvee 

\biguplus 

 

\coprod \bigwedge 

The symbols \int and \sum have already been introduced in Section 5.2.5. 

There it was shown how these symbols may take on upper and lower limits; 

in the same way, all the above symbols may also be assigned upper and 

lower limits using the shifting commands ˆ and _. The positioning of 

the limits varies for some symbols depending on whether they occur in 

text or displayed formulas. As indicated in Section 5.2.5, the command 

\limits forces the limits to be written above and below the symbol where 

they would otherwise be placed beside it. Similarly the complementary 

command \nolimits sets them beside the symbol when the standard 

positioning is above and below. 

∞ 

0 

n 

ν=0 

∞ 

0 

n ν=0 

5.3.8 Function names 

\[ \ointˆ\infty_0 \oint\limitsˆ\infty_0 \] 

\[ \prodˆn_{\nu=0} \prod\nolimitsˆn_{\nu=0} \] 

The universal standard for mathematical formulas is to set variable names 

in italics but the names of functions in Roman. If one were simply to write 

the function names sin or log in math mode, L AT E X would interpret these 

as variables s i n and l o g and write them as sin and log. In order 

to tell L AT E X that a function name is wanted, it is necessary to enter a 

command consisting of the backslash \ plus the function name. The 

following names are recognized by L AT E X: 

\arccos \cosh \det \inf \limsup \Pr \tan 

\arcsin \cot \dim \ker \ln \sec \tanh 

\arctan \coth \exp \lg \log \sin 

\arg \csc \gcd \lim \max \sinh 

\cos \deg \hom \liminf \min \sup


Some of these functions may also appear with limits attached to them. 

This is easily achieved by means of the subscript command after the name 

of the function: \lim_{x\to\infty} yields 

and 

limx→∞ in text formulas 

lim 

x→∞ 

in displayed formulas 

The following function names may accept a limit with the lowering 

(index) command _: 

\det \gcd \inf \lim \liminf \limsup \max \min 

\Pr \sup 

Finally, there are the function commands \bmod and \pmod{arg}, both 

of which produce the function mod in one of two forms: 

$ a \bmod b $ ⇒ a mod b 

$ y \pmod{a+b} $ ⇒ y (mod a + b). 

With A M S-L AT E X (Section 12.2.5) it is possible to define additional 

function names. 

5.3.9 Mathematical accents 

The following mathematical accents are available within math mode: 

â \hat{a} ă \breve{a} à \grave{a} ā \bar{a} 

ǎ \check{a} á \acute{a} ã \tilde{a} a \vec{a} 

˙a \dot{a} ä \ddot{a} ˚a \mathring{a} 

The letters i and j should be printed without their dots when they 

are given an accent. To accomplish this, type the symbols \imath and 

\jmath instead of the letters, as in 

$\vec{\imath} + \tilde{\jmath}$: ı + ˜j 

There are wider versions of \hat and \tilde available with the names 

\widehat and \widetilde. In this way, these accents may be placed over 

parts of a formula: 

1 − x = −y $\widehat{1-x}=\widehat{-y}$ 

xyz $\widetilde{xyz}$ 

Exercise 5.6: The union of two sets A and B is the set of all elements that 

are in at least one of the two sets, and is designated as A ∪ B. This operation 

is commutative A ∪ B = B ∪ A and associative (A ∪ B) ∪ C = A ∪ (B ∪ C). If 

A ⊆ B, then A ∪ B = B. It then follows that A ∪ A = A, A ∪ {∅} = A and 

J ∪ A = J. 

Exercise 5.7: Applying l’Hôpital’s rule, one has 

ln sin πx 

lim 

= lim 

x→0 ln sin x x→0 

cos πx 

π sin πx 

cos x 

sin x 

π tan x π/ cos 

= lim = lim 

x→0 tan πx x→0 

2 x 

π/ cos2 cos 

= lim 

πx x→0 

2 πx 

cos2 = 1 

x


Exercise 5.8: The gamma function Γ (x) is defined as 

Γ (x) ≡ lim 

n−1 

n→∞ 

ν=0 

n! n x−1 

x + ν 

n! n 

= lim 

n→∞ 

x−1 

x(x + 1)(x + 2) · · · (x + n − 1) ≡ 

∞ 

e 

0 

−t t x−1 dt 

The integral definition is valid only for x > 0 (2nd Euler integral). 

Exercise 5.9: Remove the option fleqn from the document class command in 

Exercise 5.3 and redo the output. 

Exercise 5.10: 

αx = xα, αβx = βαx, (α + β)x = αx + βx, α(x + y) = αx + α y. 

x y = y x but x × y = − y × x, x y = 0 for x ⊥ y, x × y = 0, for x y. 

Exercise 5.11: Reproduce Equations 5.1 and 5.2 from the next section. 

5.4 Additional elements 

The math elements described in the previous sections already permit the 

construction of very complex formulas, such as 

√ 

1 + x − 1 ( 

lim 

= lim 

x→0 x 

x→0 

√ 1 + x − 1)( √ 1 + x + 1) 

x( √ 1 

= lim √ = 

1 + x + 1) 

x→0 1 + x + 1 1 

(5.1) 

2 

I(z)=sin( π 

2 z2 ∞ 

) 

∂2U ∂x2 + ∂2U ∂y 2 = 0 =⇒ UM = 1 

 

 

1 ∂U 1 

ds − 

4π r ∂n 4π 

Σ 

Σ 

n=0 

(−1) n π 2n 

1 · 3 · · · (4n + 1) z4n+1 − cos( π 

2 z2 ) 

n=0 

∂ 1 

r 

U ds (5.2) 

∂n 

∞ (−1) nπ 2n+1 

1 · 3 · · · (4n + 3) z4n+3 

(5.3) 

By reading the formulas from left to right there should be no difficulty 

in reconstructing the text that produced them. For example, the last 

equation is generated with 

\begin{equation} 

I(z) = \sin( \frac{\pi}{2} zˆ2 ) \sum_{n=0}ˆ\infty 

\frac{ (-1)ˆn \piˆ{2n} }{1 \cdot 3 \cdots (4n+1) } zˆ{4n+1} 

-\cos( \frac{\pi}{2} zˆ2 ) \sum_{n=0}ˆ\infty 

\frac{ (-1)ˆn \piˆ{2n+1} }{ 1 \cdot 3 \cdots (4n+3) } zˆ{4n+3} 

\end{equation} 

The above examples were made using the equation environment instead 

of the displaymath environment or its abbreviated form \[. . . \], 

which has the effect of adding the equation numbers automatically. In the 

document classes book and report, equations are sequentially numbered 

within the chapter, the number being preceded by the chapter number 

and set within parentheses ( ), as illustrated above. For document class

5.4. Additional elements 131 

article, the equations are numbered sequentially throughout the entire 

document. 

By default the equation number appears right justified and vertically 

centered with the equation. If there is not enough room for it on the 

same line, it is printed right justified below the equation. If the document 

class option leqno has been selected, the equation numbers are set left 

justified for the entire document. 

The automatic numbering of equations means that the author may 

not know at the time of writing just what the equation number is. The 

L AT E X cross-reference system described in Section 9.2.1 has already been 

explained for referring to section numbers (Section 3.3.3) and may also 

be used for equation numbers. By including a command \label{name} 

within the equation environment, one can print the unknown equation 

number in the text with the command \ref{name}, where name is a 

keyword consisting of any combination of letters, numbers, or symbols. 

Examining Equation 5.3 more closely, one notices that the two parentheses 

pairs ( ) in cos() and sin() could be somewhat larger. Furthermore, 

this equation just fills the line width, and if it were any longer, it would 

have to be broken at some appropriate spot and the parts positioned in 

a meaningful way relative to one another. None of the math elements 

described so far can accomplish these requirements. 

Even something so simple as including some normal text within a 

formula has not yet been mentioned. The rest of this section addresses 

these problems. 

Finally, there are times when one is not happy with the sizes that TEX has chosen, as for example in the last integral of Equation 5.2 where ∂ 1 

r 

would be more desirable than ∂ 1 

r . This and other formatting aids, such 

as adjusting horizontal spacing between parts of formulas, are dealt with 

in Section 5.5. 

5.4.1 Automatic sizing of bracket symbols 

Mathematics often contains bracketing symbols, usually in pairs that 

enclose part of the formula. When printed, these bracket symbols should 

be the same size as the included partial formula. L AT E X provides a pair of 

commands 

\left lbrack sub form \right rbrack 

to accomplish this. The command \left is placed immediately before 

the opening (left hand) bracket symbol lbrack while \right comes just 

before the closing (right hand) symbol rbrack.


 

x=1 + 

x=0 

\[ \left[ \int + \int \right]_{x=0}ˆ{x=1} \] 

The pair of brackets [ ] is adjusted to the size of the enclosed 

formula, as are the raised exponent and lowered 

index as well. 

The commands \left and \right must appear as a pair. For every 

\left command there must be a corresponding \right command somewhere 

afterwards. The pairs may be nested. The first \left is paired with 

the last \right; the following \left with the second last \right, and so 

on. There must be the same number of \right as \left commands in a 

nesting. 

The corresponding bracket symbols lbrack and rbrack may be perfectly 

arbitrary and do not need to be a logical pair. 

x + y + z = 

a 

b 

 

This set of brackets is admittedly unusual but 

permissible. 

\[ \vec{x} + \vec{y} + \vec{z} = 

\left( ... \right[ \] 

Sometimes a formula contains only a single opening or closing bracket 

without a corresponding counterpart. However, the \left. . . \right 

commands must still be given as a pair, but with a period ‘.’ as an invisible 

bracket symbol. 

⎧ 

⎪⎨ −1 : x < 0 

y = 

⎪⎩ 

0 

+1 

: 

: 

x = 0 

x > 0 

\[ y = \left\{ \begin{array} 

{r@{\quad:\quad}l} 

-1 & x0 

\end{array} \right. \] 

The array environment in the above example is described in Section 4.8.1 

and produces a table in math mode. 

The \left. . . \right commands may be applied to a total of 22 different 

symbols. These are 

( ( ) ) ⌊ \lfloor ⌋ \rfloor 

[ [ ] ] ⌈ \lceil ⌉ \rceil 

{ \{ } \} 〈 \langle 〉 \rangle 

| | \| ↑ \uparrow ⇑ \Uparrow 

/ / \ \backslash ↓ 

 

\downarrow 

\updownarrow 

⇓ 

 

\Downarrow 

\Updownarrow 

For example, \left|. . . \right| produces two vertical bars adjusted in 

height to contain the enclosed formula text. 

 

π 

Exercise 5.12: In Equation 5.3, generate cos 

2 z2 

 

π 

and sin 

2 z2 

 

instead of 

cos( π 

2 z2 ) and sin( π 

2 z2 ).

5.4.2 Ordinary text within a formula 


It is often necessary to include some normal text within a formula, for 

example single words such as and, or, if, and so on. In this case one 

must switch to LR mode (Section 4.7.1) while staying in math mode. This 

is carried out with the command \mbox{normal text} given inside the 

formula, together with horizontal spacing commands such as \quad or 

\hspace. For example: 

Xn = Xk if and only if Yn = Yk and Zn = Zk 

\[ X_n = X_k \qquad\mbox{if and only if}\qquad 

Y_n = Y_k \quad\mbox{and}\quad Z_n = Z_k \] 

In order to set a longer piece of text beside a displayed formula, as 

in some of the above examples, it is more appropriate to put both the 

formula and the text in their own parboxes or minipages, placed side by 

side with the proper vertical positioning. 

On the other hand, if letters from text fonts are required as mathematical 

symbols, they should be entered with the math alphabet commands: 

\mathrm \mathtt \mathbf 

\mathsf \mathit \mathcal 

We have already met \mathcal in Section 5.3.2 on calligraphic letters. All 

these commands function the same way: they set their argument in the 

corresponding font. 

B 0 (x) T i 

j 

$\mathbf{B}ˆ0(x)$ \quad $\mathsf{T}î_j$ 

The command \mathnormal in Section 5.3.1 also belongs to this group. 

The difference between it and \mathit is that it sets its argument in the 

regular math italic font, while the latter uses the normal text italic. The 

letters are the same, but the spacing is different. 

$\mathnormal{differ} \ne \mathit{differ}$ 

dif f er = differ 

All the math alphabet commands set their text in math mode, which 

means that spaces are ignored as usual. This is not the case for text 

placed in an \mbox. 

5.4.3 Matrices and arrays 

a11 a12 · · · a1n 

. 

. 

. .. 

. 

an1 an2 · · · ann 

Structures like the one at the left are the 

basis for matrices, determinants, system of 

equations, and so on. They will all be referred 

to here as arrays.


Arrays are produced by means of the array environment, whose syntax 

and construction are described in Section 4.8.1 on tables. The array 

environment generates a table in math mode, that is, the column entries 

are interpreted as formula text. For example: 

a11x1 + a12x2 + · · · + a1nxn = b1 

a22x1 + a22x2 + · · · + a2nxn = b2 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

an1x1 + an2x2 + · · · + annxn = bn 

\[ \begin{array}{*{3}{c@{\:+\:}}c@{\;=\;}c} 

a_{11}x_1 & a_{12}x_2 & \cdots & a_{1n}x_n & b_1 \\ 

a_{22}x_1 & a_{22}x_2 & \cdots & a_{2n}x_n & b_2 \\ 

\multicolumn{5}{c}{\dotfill} \\ 

a_{n1}x_1 & a_{n2}x_2 & \cdots & a_{nn}x_n & b_n 

\end{array} \] 

As a reminder of the table construction elements (Section 4.8.1): @{t} in- 

! serts the contents of t between the adjacent columns. In the above example, 

this is \:+\: and \;+\;. The commands \: and \; have not yet been 

introduced but they produce small horizontal spacing in math mode (Section 

5.5.1). *{3}{c@{\:+\:}} is an abbreviation for three repetitions of the 

column definition c@{\:+\:}. c defines the column to be one of centered text. 

\multicolumn{5}{c} says that the next five columns are to be merged and replaced 

by one with centered text. The command \dotfill fills the column with 

dots. 

It is possible to nest array environments: 

⎛ 

 

 

⎜ 

⎜ 

⎜ 

⎝ 

x11 x12 

x21 x22 

y 

z 

 

⎞ 

 

 

 

⎟ 

⎠ 

\[ \left( \begin{array}{c} 

\left| \begin{array}{cc} 

x_{11} & x_{12} \\ x_{21} & x_{22} 

\end{array} \right| \\ 

y \\ z \end{array} \right) \] 

The outermost array consists of one column with centered text (c). 

The first entry in this column is also an array, with two centered columns. 

This array is surrounded left and right by vertical lines with adjusted 

sizes. 

The array environment is structurally the same as a vertical box. This 

means that it is treated as a single character within the surrounding environment, 

so that it may be coupled with other symbols and construction 

elements. 

 

 

 

 

(1,2,...,n) 

p1


\[ \sum_{p_1


is given by the expression 

 

 

 

 

 

 

± 

l1 m1 

l2 m2 

x1 − x2 y1 − y2 z1 − z2 

l1 m1 n1 

l2 m2 n2 

 

2 

 

 

 

+ 

 

m1 n1 

m2 n2 

 

2 

 

 

 

+ 

 

n1 l1 

n2 l2 

If the numerator is zero, the two lines meet somewhere. 

Note: we do not recommend using {|cc|} in the formatting argument of 

the three determinants in the denominator under the root sign. Here the 

\left|. . . \right| pair should be applied. Try out both possibilities for yourself 

and compare the results. 

Exercise 5.15: Laurent expansion: using cn = 1 

−n−1 (ζ −a) f (ζ) dζ, for every 

2πi 

function f (z) the following representation is valid (n = 0, ±1, ±2, . . . ) 

+∞ 

f (x) = cn(z − a) 

n=−∞ 

n ⎧ 

⎪⎨ c0 + c1(z − a) + c2(z − a) 

= 

⎪⎩ 

2 + · · · + cn(z − a) n + · · · 

+ c−1(z − a) −1 + c−2(z − a) −2 + · · · 

+ c−n(z − a) −n + · · · 

Tip: the right-hand side of the equation can be created with an array environment 

consisting of only one column. What is its formatting argument? 

5.4.4 Lines above and below formulas 

The commands 

\overline{sub form} and \underline{sub form} 

can be used to draw lines over or under a formula or sub-formula. They 

may be nested to any level: 

a 2 + xy + z 

\[ \overline{\overline{a}ˆ2 + \underline{xy} 

+ \overline{\overline{z}}} \] 

The command \underline may also be employed in normal text mode 

to underline text, whereas \overline is allowed only in math mode. 

Exactly analogous to these are the two commands 

\overbrace{sub form} and \underbrace{sub form} 

which put horizontal curly braces above or below the sub-formula. 

 

a + b 

 

+ c 

+d 

\overbrace{a + \underbrace{b+c} + d} 

In displayed formulas, these commands may have exponents or indices 

attached to them. The (raised) exponent is set above the overbrace while 

the (lowered) index is placed below the underbrace. 

 

 

 

 

 

2

123 

 

a + b + · · · + y +z 

 

αβγ 


\[ \underbrace{a + \overbrace{b + \cdots + 

y}ˆ{123} + z}_{\alpha\beta\gamma} \] 

Exercise 5.16: The total number of permutations of n elements taken m at a 

time (symbol P m n ) is 

5.4.5 Stacked symbols 

P m m−1 

n! 

n = (n − i) = n(n − 1)(n − 2) . . . (n − m + 1) = 

(n − m)! 

i=0 

total of m factors 


\stackrel{upper sym}{lower sym} 

places the symbol upper sym centered on top of lower sym, whereby the 

symbol on top is set in a smaller typeface. 

x def 

= (x1, . . . xn) $ \vec{x} \stackrel{\mathrm{def}}{=} (x_1,...$ 

A α′ 

−→ B β′ 

←− C $ A \stackrel{\alpha’}{\longrightarrow} B ... $ 

By making use of math font size commands (Section 5.5.2) it is possible 

to construct new symbols with this command. For example, some authors 

prefer to have the \le symbol appear as < = instead of ≤. This is achieved 

by combining < and = with $\stackrel{\textstyle


⎛ 

 

⎝ 

j≥0 

k≥0 

ajkz k 

⎞ 

 

⎠ = 

n≥0 

z n 

⎛ 

⎜ 

⎝ 

 

k0,k1,...≥0 

k0+k1+···=0 

a0k0 

⎞ 

⎟ 

a1k1 . . . ⎟ 

⎠ 

\[ \prod_{j\ge0}\left( \sum_{k\ge0} a_{jk}zˆk \right) = 

\sum_{n\ge0} zˆn \left(\sum_{k_0,k_1\ldots\ge0 \atop 

k_0+k_1+\cdots=0} a_{0k_0} a_{1k_1}\ldots \right) \] 

Similar structures can be generated with the L AT E X environments 

\begin{array}{c} upper line \\ lower line \end{array} (atop) 

\left(\begin{array}{c} upper \\ lower \end{array}\right) (choose) 

The difference between these array structures and those of the T E X 

commands is that the former are always printed in the size and style of 

normal text formulas, whereas the latter will have varying sizes depending 

on where they appear within the formula. 

For comparison 

∆p1p2···pn−k 

p1p2···pn−k 

∆ p1p2 · · · pn−k 

p1p2 · · · pn−k 

The index array is produced using \atop 

The index array is produced using \array 

The above TEX commands may also 

 

be 

 

employed 

 

to 

 

produce small 

1 0 a b c 

matrices within text formulas, such as 0 1 or l m n . Here the first 

matrix was typed in with 

${1\,0\choose0\,1}$ 

and the second with 

$\left({a\atop l}{b\atop m}{c\atop n}\right)$ 

The syntax of these Plain T E X commands is radically different from that 

! normally used by L AT E X. See page 189 for a way to correct this. 

5.4.7 Multiline equations 

A multiline equation is one that is developed over several lines in which 

the relation symbols (for example, = or ≤) in each line are all vertically 

aligned with each other. For this purpose, the environments 

\begin{eqnarray} line 1\\ ... \\ line n end{eqnarray} 

\begin{eqnarray*} line 1\\ ... \\ line n end{eqnarray*} 

are used to set several lines of formulas or equations in displayed math 

mode. The individual lines of the equation or formula are separated from 

one another by \\. Each entry line has the form

left formula & mid formula & right formula \\ 


When printed, all the left formulas appear right justified in a left column, 

the right formulas left justified in a right column, and the mid formulas 

centered in between. The column separation character & designates 

the various parts of the formula. Normally the mid formula is a single 

math character, the relation operator mentioned above. The individual 

lines thus have the same behavior as they would in a \begin{array} 

{rcl}. . . \end{array} environment. 

The difference between the array and eqnarray environments is that in the 

! latter the lines are set as displayed formulas. This means that for those symbols 

listed in Section 5.3.7 the larger form will be selected, and that the numerator 

and denominator of fractions will be in normal size. On the other hand, for the 

array environment the column entries will be set as text formulas, the smaller 

form of these symbols will be chosen and the parts of the fraction will appear in 

a smaller type size. 

The standard form of the eqnarray environment adds an automatic 

sequential equation number, which is missing in the *-form. To suppress 

the equation number for a single line in the standard form, add the 

command \nonumber just before the line termination \\. 

The equation numbers may be referred to in the text with the command 

\ref{name} once the keyword name has been assigned to the equation 

number with the \label{name} command somewhere in the line of the 

equation. See Section 9.2.1 for more details. 

Examples: 

(x + y)(x − y) = x 2 − xy + xy − y 2 

= x 2 − y 2 

(x + y) 2 = x 2 + 2xy + y 2 

\begin{eqnarray} 

(x+y)(x-y) & = & xˆ2-xy+xy-yˆ2 \nonumber\\ 

& = & xˆ2 - yˆ2 \\ 

(x+y)ˆ2 & = & xˆ2 + 2xy + yˆ2 

\end{eqnarray} 

xnu1 + · · · + xn+t−1ut = xnu1 + (axn + c)u2 + · · · 

 

+ a t−1 xn + c(a t−2 

+ · · · + 1) 

ut 

(5.4) 

(5.5) 

= (u1 + au2 + · · · + a t−1 ut)xn + h(u1, . . . , ut) 

\begin{eqnarray*} 

x_nu_1 + \cdots + x_{n+t-1}u_t & = & x_nu_1 + (ax_n + c)u_2 + 

\cdots\\ 

& & + \left(aˆ{t-1}x_n + c(aˆ{t-2} + \cdots+1)\right)u_t\\ 

& = & (u_1 + au_2 + \cdots + aˆ{t-1}u_t)x_n + h(u_1,\ldots,u_t) 

\end{eqnarray*}


The second example requires some explanation. In the second line 

there is a \left(. . . \right) pair for the automatic sizing of the (). Such 

a pair may only appear within a single line; that is, it may not be broken 

by the \\ line terminator! If automatic bracket sizing is to occur in a 

multiline equation, it may only be found within a single line. 

If bracket pairs must appear on different lines, one can try using the 

construction \left(· · · \right. \\ \left.· · · \right). In the first line, 

the \left( is paired with the invisible bracket \right., while the second 

begins with the invisible \left. which is paired with the closing \right). 

However, this will only work satisfactorily if both parts of the equation 

have roughly the same height so that the two automatic sizings yield 

much the same results. Section 5.5.3 describes how to select bracket 

sizes manually in the event that the automatic method fails. 

The + sign at the beginning of the second line also requires a remark. 

The signs + and − have two meanings in mathematics: between two math 

quantities they act as a coupling (binary operator), but coming before a 

math symbol they serve as a sign designation (positive or negative). L AT E X 

stresses this difference by inserting different spacing in the two cases (for 

example, compare +b with a + b). 

y = a + b + c + d 

+e + f + g 

+ h + i + j 

If a long formula is broken into several lines 

and one of them begins with + or −, L AT E X 

regards it as a sign designation and moves it 

closer to the next character. 

The solution is to introduce an invisible character of zero width at the 

beginning of such a line. This may be the empty structure {}. Compare 

the effects of && +e+f+g and &&{}+h+i+j in the above equation. 

Since the extra spacing is always inserted between + and an opening 

parenthesis (, it was not necessary to give the {} at the beginning of the 

second line of the eqnarray* example on the previous page. 

Occasionally it is better to break long multiline equations as follows: 

that is, the second and subsequent lines are 

w + x + y + z = 

not aligned with the equals sign but are left 

a + b + c + d + e + f + justified with a certain indentation from the 

g + h + i + j + k + l beginning of the first line. 


\lefteqn{w+x+y+z = } \\ 

& & a+b+c+d+e+f+ \\ 

& & g+h+i+j+k+l 

\end{eqnarray*} 

The command \lefteqn{w+x+y+z =} \\ 

in the first line has the effect that the contents 

of the argument are indeed printed 

out, but that L AT E X considers them to have 

zero width. The left-hand column then 

contains only intercolumn spacing, which 

produces the indentation for the rest of the lines. 

The indentation depth may be altered by inserting \hspace{depth} 

between the \lefteqn{. . . } and the \\ line termination. A positive value 

for depth increases the indentation, a negative value decreases it.


Exercise 5.17: The following equations are to be broken as shown: 

arcsin x = − arcsin(−x) = π 

 

− arccos x = arccos 1 − x 

2 2 

 

x 

= arctan √ 

1 − x2 = 

√ 

1 − x2 arccot 

x 

 

∂f (x, y) ∂f (x, y) 

f (x + h, y + k) = f (x, y) + 

h + k 

∂x 

∂y 

+ 1 

 

2 ∂ f (x, y) 

2 ∂x2 h 2 + 2 ∂2f (x, y) 

∂x∂y kh + ∂2f (x, y) 

∂y 2 

k 2 

 

+ 1 

1 

{· · · } + · · · + {· · · } + Rn 

6 n! 

Note on possible error messages: 

Long formulas with many sets of logical brackets, especially 

deeply nested ones, will almost always contain errors at first. The 

cause is often brackets that have been incorrectly ordered or overlooked. 

If L AT E X produces error messages during formula processing, which 

the beginner is not able to interpret correctly (error messages are 

described in detail in Appendix C), he or she should check the bracket 

pairing very carefully in the formula text. Some text editors can do 

the search for the matching bracket, which greatly simplifies the task. 

If the error is not found in this way, one can try pressing the return 

key at the error message in order to proceed with the processing. The 

resulting printout can indicate where things may have gone wrong. 

(5.6) 

(5.7) 

Exercise 5.18: The eqnarray environment inserts additional spacing where the 

column separation character & appears. This is undesirable when an equation is 

to be broken and aligned on + or − within a long summation. For example: 

The inverse function of the polynomial expansion y = f (x) = ax + bx 2 + cx 3 + 

dx 4 + ex 5 + f x 6 + · · · (a = 0) begins with the elements 

x = ϕ(y) = 1 

y 

a 

− 

b 

a3 y 2 + 1 

a5 (2b2 − ac)y 3 

+ 

1 

a7 (5abc − z2d − f b 3 )y 4 

+ 

+ 

1 

a9 (6a2bd + 3a 2 c 2 + 14b 4 − a 3 e − 21ab 2 c)y 5 

1 

a11 (7a3be + 7a 3 cd + 84ab 3 c − a 4 f − 

28a 2 b 2 d − 28a 2 bc 2 − 43b 5 )y 6 +· · · 

Select a value for the declaration \arraycolsep (Section 4.8.2) such that the 

distance between the + and − and the break points are as near as possible to 

those in the rest of the formula.


5.4.8 Framed or side-by-side formulas 

Displayed formulas or equations may be put into vertical boxes of appropriate 

! width, that is, in a \parbox command or minipage environment. Within the vertical 

box, the formulas are horizontally centered or left justified with indentation 

\mathindent according to the selected document class option. 

Vertical boxes may be positioned relative to one another just like single 

characters (Sections 4.7.3 and 4.7.7). In this way the user may place displayed 

formulas or equations side by side. 

α = f (z) (5.8) 

β = f (z 2 ) (5.9) 

γ = f (z 3 ) (5.10) 

x = α 2 − β 2 

y = 2αβ 

The left-hand set of equations 

is set in a \parbox of width 

4 cm, the right-hand set in one 

of width 2.5 cm, while this text 

is inside a minipage of width 

4.5 cm. 

\parbox{4cm}{\begin{eqnarray} \alpha &=& f(z)...\end{eqnarray}} 

\hfill \parbox{2.5cm}{\begin{eqnarray*} 

x &=& \alphaˆ2 - \betaˆ2\\ y &=& 2\alpha\beta \end{eqnarray*}} 

\hfill \begin{minipage}{4.5cm} The left-hand ... \end{minipage} 

Vertical boxes can also be useful when equation numbers are placed in an 

! unconventional manner. The eqnarray environment generates an equation number 

for every line, which may be suppressed with \nonumber. To add a vertically 

centered equation number to a set of equations, for example, 

P(x) = a0 + a1x + a2x 2 + · · · + anx n 

P(−x) = a0 − a1x + a2x 2 − · · · + (−1) n anx n (5.11) 

the following text may be given: 

\parbox{10cm}{\begin{eqnarray*} ... \end{eqnarray*}} \hfill 

\parbox{1cm}{\begin{eqnarray}\end{eqnarray}} 

The actual set of equations is produced here in the eqnarray* environment, 

within a vertical box of width 10 cm, followed by an empty eqnarray environment 

in a box of width 1 cm that generates the equation number. Both boxes are 

vertically aligned along their center lines. 

Emphasizing formulas by framing requires no new construction elements. It 

! is sufficient to put them into an \fbox (Section 4.7.7). Text formulas a + b 

are simply framed with \fbox{$a+b$}. For displayed formulas, \displaystyle 

(Section 5.5.2) must be issued, else they are set as text formulas. 

is produced with 

∞ 

0 

f (x) dx ≈ 

n 

wie xif (xi) 

\[\fbox{$\displaystyle \intˆ\infty_0 f(x)\,\mathrm{d}x ..$}\] 

An alternative method to frame displayed equations is with the A M S-L AT E X \boxed 

command, page 270. 

i=1


5.4.9 Chemical formulas and bold face in math formulas 

In mathematics it is sometimes necessary to set individual characters or 

parts of the formula in bold face. This can be achieved simply with the 

math alphabet command \mathbf that we met in Section 5.4.2: 

$\mathbf{Sˆ{-1}TS = dg(\omega_1,\ldots,\omega_n) = \Lambda}$ 

produces S −1 TS = dg(ω1, . . . , ωn) = Λ. 

In this example, the entire formula has been set as the argument of 

\mathbf so that everything should be set in bold face. In fact, only 

numbers, lower and upper case Latin letters, and upper case Greek letters 

are set in bold Roman with \mathbf. Lower case Greek letters and other 

math symbols appear in the normal math font. 

If only part of the formula is to be set in bold face, that part must be 

given as the argument of the \mathbf command. 

$\mathbf{2\sqrt{x}/y} = z$ 2 √ x/y = z 

The math font style command \boldmath will set all characters in 

bold face, with the following exceptions: 

• raised and lowered symbols (exponents and indices) 

• the characters + : ; ! ? ( ) [ ] 

• symbols that exist in two sizes (Section 5.3.7) 

The \boldmath declaration may not appear in math mode. It must 

be called before switching to math mode or within a parbox or minipage. 

The countercommand \unboldmath resets the math fonts back to the 

normal ones. 

 

\boldmath \[ \oint\limits_C V 

V dτ = ∇ × V dσ \,\mathrm{d}\tau =\oint\limits_\Sigma 

\nabla\times V\,d\sigma \] \unboldmath 

C 

Σ 

If \boldmath has been turned on outside the math mode, it may be 

temporarily turned off inside with \mbox{\unboldmath$. . . $}. 

\boldmath$ P = \mbox{\unboldmath$m$}b$\unboldmath 

yields: P = mb. Similarly, \boldmath can be temporarily turned on 

within math mode with the structure \mbox{\boldmath$. . . $}: Wr = 

M dϕ = r 2 mω 2 /2 

$ W_r = \int\mbox{\boldmath$M\,\mathrm{d}\varphi$} =..$ 

An alternative method of printing single symbols in bold face is provided 

by the bm package described on page 394. 

Chemical formulas are normally set in Roman type, not in italics as 

for mathematical formulas. This may be brought about by setting the 

formula as the argument of the font command \mathrm: 

$\mathrm{Fe_2ˆ{2+}Cr_2O_4}$ Fe 2+ 

2 Cr2O4


5.4.10 International typesetting standards 

The International Standards Organization (ISO) has established the recognized 

! conventions for typesetting mathematics, the essential elements of which are 

presented in an article by Beccari (1997), along with a description of how they 

may be realized with L AT E X. Some of these rules have already been mentioned and 

demonstrated in the examples. Here are the major points. 

1. Simple variables are represented by italic letters, as a b c x y z. 

2. Vectors are written in bold face italic, as B v ω. 

3. Tensors of 2nd order and matrices may appear in a sans serif font, as 

M D I. 

4. The special numbers e, i, π, as well as the differential operator d, are to be 

written in an upright font to emphasize that they are not variables. 

5. A measurement consisting of a number plus a dimension is an indivisible 

unit, with a smaller than normal space between them, as 5.3 km and 62 kg. 

The dimension is in an upright font. 

Point 1 is fulfilled automatically by L AT E X. Point 5 is easily achieved by inserting 

the small space \, command between the number and dimension, as 5.3\,km 

and 62\,kg. Using the protected space ˜ instead is very common practice among 

L AT E X users, which ensures that the two parts are not split, but with regular, not 

small spacing: 5.3 km and 62 kg. 

Point 2 is not satisfied with the \vec command, which produces B. Nor does 

\mathbf help, for this yields B, in an upright font. The best solution is to use 

the \boldsymbol command from the A M S package amsbsy (Section 12.2.1) or 

the \bm command from the tools package bm (Section B.5.4). Otherwise one must 

resort to defining 

\renewcommand{\vec}[1]{\mbox{\boldmath$#1$}} 

for a revised \vec command. 

Similarly point 3 can be met with the math alphabet command \mathsf. 

However, Beccari does point out that even tensor variables should probably be 

italic. This is more difficult, but it can be accomplished. See example 5 on 

page 191. 

Point 4 is the one that is most often violated, especially for the differential d. 

We have demonstrated it in the examples in this book and have shown how it may 

be achieved with the \mathrm math alphabet command. However, to simplify the 

application, it is recommended to create some user-defined commands, such as 

\newcommand{\me}{\mathrm{e}} for math e 

\newcommand{\mi}{\mathrm{i}} for math i 

\newcommand{\dif}{\mathrm{d}} for differential operator d 

An upright π is not so easy since this is not provided in the usual math fonts. 

With these new commands, the equation on page 119 is more conveniently 

set with 

\[ \intˆ{\infty}_0 f(x)\,\dif x \approx 

\sumˆn_{i=1}w_i \meˆ{x_i} f(x_i) \]

5.5 Fine-tuning mathematics 

5.5.1 Horizontal spacing 

5.5. Fine-tuning mathematics 145 

Even though T E X has a thorough knowledge of the rules of mathematical 

typesetting, it cannot hope to understand the mathematical meaning. For 

example, y dx normally means the joining of the variable y with the 

differential operator dx, and this joining is designated by a small space 

between the two. However, T E X removes the blank in the entry y dx and 

prints ydx, the product of three variables y, d, and x. At this point, L AT E X 

needs some fine-tuning assistance. (And the d should really be upright, 

the \dif or \mathrm{d} of the previous section.) 

Small amounts of horizontal spacing can be added in math mode with 

the commands 

\, small space = 3/18 of a quad 

\: medium space = 4/18 of a quad 

\; large space = 5/18 of a quad 

\! negative space = −3/18 of a quad 

In the following examples, the third column contains the results without 

the additional horizontal spacing command. 

√ √ 

$\sqrt{2}\,x$ 

2 x 

2x 

 

$\sqrt{\,\log x}$ 

log x 

log x 

$O\left(1/\sqrt{n}\,\right)$ O (1/ √ n ) O (1/ √ n) 

$[\,0,1)$ [ 0, 1) [0, 1) 

$log n\,(\log\log n)ˆ2$ log n (log log n) 2 log n(log log n) 2 

$xˆ2\!/2$ x 2 /2 x 2 /2 

$n/\!\log n$ n/log n n/ log n 

$\Gamma_{\!2}+\Deltaˆ{\!2}$ Γ2 + ∆ 2 Γ2 + ∆ 2 

$R_i{}ˆj{}_{\!kl}$ Ri j kl Ri j kl 

$\int_0ˆx\!\int_0ˆy 

x 

0 

y 

0 

dF(u, v) 

\mathrm{d}F(u,v)$ 

 

\[ \int\!\!\!\int_D 

dx dy 

\mathrm{d}x\,\mathrm{d}y \] 

D 

x 

0 

y 

0 

dF(u, v) 

 

dxdy 

D 

Note: In the third last example R_i{}ˆj is so constructed that an invisible 

character of zero width comes after the index of Ri, and it is this dummy 

character that receives the following exponent. The result is Ri j , instead 

of R j 

i which is produced by R_iˆj. 

There are no hard and fast rules for applying the math spacing commands. 

Some candidates are the differential operator, small root signs 

in text formulas followed by a variable, the dividing sign /, and multiple 

integral signs. The above examples illustrate many suitable situations.


5.5.2 Selecting font size in formulas 

It is possible to alter the font sizes that T E X selects for the various parts 

of the formula. First we must explain what sizes are available in math 

mode and what T E X’s selection rules are. 

In math mode there are four font sizes that may be chosen, their actual 

sizes being relative to the basic font size of the document class: 

\displaystyle D Normal size for displayed formulas 

\textstyle T Normal size for text formulas 

\scriptstyle S Normal size for first sub-, superscript 

\scriptscriptstyle SS Normal size for later sub-, superscripts 

From now on we shall make use of the symbolic abbreviations D, T , S, 

and SS. When math mode is switched on, the active font size becomes D 

for displayed and T for text formulas. Their only difference lies in those 

symbols that appear in two sizes, plus the corresponding style of those 

superscripts and subscripts (Section 5.3.7). The larger symbols belong to 

D, the smaller to T . 

Starting from these base sizes, various math elements will be set in 

other sizes. Once another size has been chosen for an element, it remains 

the active size within that element. 

The table below shows the selection 

rules: 

Active Fractions Super-, 

size upper lower subscripts 

D T T S 

T S S S 

S SS SS SS 

SS SS SS SS 

If the active size is D, size T is selected 

for both parts of the fraction. That is, 

for both numerator and denominator 

T becomes the active size. If they in 

turn contain further fractions, these 

will be set in S. If the active size is 

D or T , superscripts and subscripts 

(exponents and indices) appear in S; 

within them S is the active size and 

any fractions or shiftings inside them 

will be set in SS. 

The T E X elements { \atop } and { \choose } are treated as fractions. 

The active font size inside an array environment is T . 

The smallest available math font size is SS. Once it has been reached, 

no further reduction is possible, so that all subsequent superscripts and 

subscripts appear in SS as well. 

From the table one easily sees that 

\[ a_0 + \frac{1}{a_1 + \frac{1}{a_2 

+ \frac{1}{a_3 + \frac{1}{a_4}}}} \] 

must appear as shown at the right. 

a0 + 

a1 + 

1 

1 

a2+ 1 

a 3 + 1 

a 4 

Such a math structure, called a continued fraction, is normally printed as

a0 + 

a1 + 

1 

a2 + 

1 

1 

a3 + 1 

a4 


\[ a_0 + \frac{1}{\displaystyle a_1 

+ \frac{1}{\displaystyle a_2 

+ \frac{1}{\displaystyle a_3 

+ \frac{1}{a_4}}}} \] 

By explicitly giving the font size within each element, one makes that 

size active rather than relying on the internally selected size. In this 

example, D is chosen for each denominator. Thus the next fraction acts 

as though it were outermost. The \displaystyle command may be 

omitted in the last fraction. (Can you see why?) 

The selection table allows one to predict precisely which math font size 

will be applied at any part of the formula so that an explicit specification 

may be made if necessary. The effects of such a choice may also be 

calculated for the following math elements. 

In the examples below, the right-hand column shows how the formulas 

would appear if the math font size is not explicitly specified. 

a b 

+ 

x − y x + y 

a − b 

1 + 

a + b 

⎛ 

⎜ 

⎝ 

xi−xj 

− 

e ni +nj 

ab e + f 

cd 

0 

g 

 

− h 

 

 

ij 

 

 

 

kl 

 

\[ \frac{\displaystyle\frac{a}{x-y} 

+\frac{b}{x+y}} 

{\displaystyle 1+\frac{a-b}{a+b}} \] 

a b 

x−y + x+y 

1 + a−b 

a+b 

\[ eˆ{\textstyle - 

\frac{x_i-x_j}{nî+nˆj}} \] e − x i −x j 

n i +n j 

⎞ 

⎟ 

⎠ 

\[ \left(\begin{array}{cc} 

\displaystyle{ab\choose cd} & 

\displaystyle\frac{e+f}{g-h}\\ 

0 & \displaystyle \left| 

{ij\atop kl} \right| 

\end{array}\right) \] 

⎛ 

⎝ 

 

ab 

cd 

0 

e+f 

g−h 

 

ij 

kl 

⎞ 

⎠ 

 

 

If an explicit size specification is to be given frequently within a document—say 

within every entry in an array environment—then considerable 

writing effort may be avoided by adding to the preamble 

\newcommand{\D}{\displaystyle}\newcommand{\T}{\textstyle}... 

In this way the size command may be given simply by typing \D or \T, 

etc. 

The math font size selection rules given above are in fact a simplification. For 

! those readers who wish to know the exact details, we now present the complete 

description.


For each of the four math font sizes D, 

T, S, SS there is a modified version D ′ , 

T ′ , S ′ , SS ′ . The difference is that with 

D, T, S, SS the superscripts (exponents) 

are somewhat higher than they are with 

D ′ , T ′ , S ′ , SS ′ . Compare the positions 

of the exponent 2 above and below the 

Active Fractions Super- Subsize 

upper lower scripts scripts 

D T T ′ S S ′ 

D ′ T ′ T ′ S ′ S ′ 

T S S ′ S S ′ 

T ′ S ′ S ′ S ′ S ′ 

S, SS SS SS ′ SS SS ′ 

S ′ , SS ′ SS ′ SS ′ SS ′ SS ′ 

line: x2 

. Otherwise the primed and 

x2 unprimed font sizes are identical. The 

true selection rules are given in the table 

at the right. 

When the font size is explicitly stated within a numerator or superscript, 

the unprimed version is selected; and in the denominators and subscripts, the 

primed fonts. 

5.5.3 Manual sizing of bracket symbols 

The \left. . . \right commands preceding one of the 22 bracket symbols 

listed in Section 5.4.1 adjust the symbol size automatically according to 

the height of the enclosed formula text. However, it is possible to select 

a size explicitly by placing one of the T E X commands \big, \Big, \bigg, 

or \Bigg before the bracket symbol. 

none 

\big 

( ) [ ]{ } ⌊ ⌋ ⌈ ⌉ 〈 〉 / \ | ↑ ⇑ ↓ ⇓ ↕ ⇕ 

⏐ 

⏐ 

\Big 

⏐ 

⏐ ⏐ ⏐ 

⏐ 

\bigg 

\Bigg 

⏐ 

⏐ ⏐ ⏐ 

⏐ ⏐ ⏐ 

⏐ 

⏐ 

⏐ ⏐ ⏐ 

⏐ ⏐ ⏐ 

⏐ ⏐ ⏐ 

⏐ 

In contrast to the \left. . . \right pairs, the explicit bracket size 

commands do not have to be contained within one line of a multiline 

formula. The opening and closing brackets may appear on different lines. 

This applies also to the commands described in the next paragraph. 

5.5.4 Math style parameters 

The mathematical style parameters listed below are set to standard values by 

! L AT E X. The user may alter them at any time with the \setlength command in the 

usual manner.


\arraycolsep 

Half the width of the intercolumn spacing for the array environment 

(see also Section 4.8.2). 

\jot Extra vertical spacing that appears between the rows of multiline equations 

in the eqnarray and eqnarray* environments. 

\mathindent 

The amount of indentation for math formulas when the document class 

option fleqn has been selected. 

\abovedisplayskip 

The vertical spacing above displayed formulas when the left side of the 

formula is closer to the left margin than the end of the preceding line 

of text. Such a formula is designated long. 

\belowdisplayskip 

The vertical spacing inserted below a long displayed formula. 

\abovedisplayshortskip 

Vertical spacing added above a short displayed formula. This is one in 

which the left edge is to the right of the end of the preceding line of 

text. 

\belowdisplayshortskip 

Vertical spacing inserted after a short displayed formula. 

\topsep 

The above four spacings are not used with the document class option 

fleqn where instead \topsep is inserted above and below displayed 

formulas (see Section 4.4.2). 

All the above parameters, except \jot, should be rubber lengths (Section 2.4.2). 

5.5.5 Some further advice 

Sometimes authors desire horizontal and vertical alignments of their 

formulas that just cannot be achieved using the means described so far. 

They should consider placing their formulas inside horizontal or vertical 

boxes which may then be positioned wherever they please. 

Similarly the array environment together with explicit size declarations 

and the table construction and style elements from Sections 4.8.1 

and 4.8.2 should be able to accomplish just about any horizontal and 

vertical alignment. 

Exercise 5.19: Generate this continued fraction. 

Note: in contrast to the example on page 146, the 

1 here in the numerator appears left justified. 

Hint: do you remember the command \hfill? 

a0 + 1 

a1 + 1 

a2 + 1 

a3 + 1 

Exercise 5.20: Produce the following set of equations with the array environment. 

sin 2α = 2 sin α cos α, cos 2α = cos 2 α − sin 2 α 

sin 3α = 3 sin α − 4 sin 3 α cos 3α = 3 cos 3 α − 3 cos α 

sin 4α = 8 cos 3 α sin α − 4 cos α sin α cos 4α = 8 cos 4 α − 8 cos 2 α + 1 

a4


Hint: recall that @{...} expressions in the formatting field of the array environment 

can be used to insert horizontal spacing and/or mathematical text between 

the columns (see the first example in Section 5.4.3). 

Exercise 5.21: Create the following output with the array environment. 

Equations for the tangential plane and surface normal 

Equation 

for the Tangential plane Surface normal 

surface 

F(x, y, z) = 0 

∂F 

∂F 

(X − x) + (Y − y) 

∂x ∂y 

+ ∂F 

(Z − z) = 0 

∂z 

z = f (x, y) Z − z = p(X − x) + q(Y − y) 

x = x(u, v) 

y = y(u, v) 

z = z(u, v) 

r = r(u, v) 

X − x Y − y Z − z 

∂x 

∂u 

∂x 

∂v 

∂y 

∂u 

∂y 

∂v 

∂z 

∂u 

∂z 

∂v 

(R − r)(r1 × r2) = 0 

or (R − r)N = 0 

= 0 

X − x 

∂F 

∂x 

= Y − y 

∂F 

∂y 

= Z − z 

∂F 

∂z 

X − x Y − y 

= 

p q 

Z − z 

= 

−1 

 

 

 

 

 

 

X − x 

∂y ∂z 

∂u ∂u 

 

 

= 

 

 

 

 

Y − y 

 

 

 

 

 

 

∂z ∂x 

∂u ∂u 

 

 

 

 

 

 

∂y 

∂v 

∂z 

∂v 

∂z 

∂v 

∂x 

∂v 

= Z − z 

 

∂x ∂y 

 

∂u ∂u 

 

 

 

∂x 

∂v 

∂y 

∂v 

R = r + λ(r1 × r2) 

or R = r + λN 

In this Table x, y, z and r are the coordinates and the radius vector of a fixed 

point M on the curve; X, Y , Z and R are the coordinates and radius vector 

of a point on the tangential plane or surface normal with reference to M; 

furthermore p = ∂z 

∂x 

, q = ∂z 

∂y and r1 = ∂r/∂u, r2 = ∂r/∂v. 

Note: If you succeed in reproducing this mathematical table you should have no 

more problems with positioning formulas and their parts! 

Hints for the solution: 

1. Define abbreviations such as \D for \displaystyle and \bm for \boldmath 

and possibly even \ba and \ea for \begin{array} and \end{array}. 

2. Build the table in stages. Start with the head and only proceed further 

when it looks reasonable. Watch out that normal text within the array 

environment must be included inside an \mbox. 

3. Continue with the first mathematical row. Here the first entry in the second 

column is also an array environment that is aligned with the rest of the 

row by means of the [t] positioning argument. Do not forget to activate 

the size \D at the necessary places in the inner structures. Insert a possible 

strut (Section 4.7.6) to generate the right distance from the head. The 

distance to the next row can be adjusted with a length specification added 

to the row terminating command \\[..].

5.6. Beyond standard L AT E X 151 

4. After this row has been completed successfully, the next should pose no 

difficulties. 

5. In the third mathematical row, both the first column and the left side of the 

second column consist of array environments. The third column contains 

three fractions whose denominators may be produced either with array 

or with the T E X {...\atop...} command; the fractions themselves are 

placed in two rows with another array environment. 

6. The second and third columns of the last mathematical row again contain 

array environments. Some of the sub-formulas in this row appear in 

\boldmath. Recall that this command may only be invoked within text 

mode, that is, within an \mbox. 

7. In the last row, all three columns of the outer array environment are 

merged together and the text is set within a parbox of appropriate width: 

\multicolumn{3}{|c|}{\parbox{..}{......}} 

5.6 Beyond standard L AT E X 

The typesetting elements presented in this chapter should be able to 

handle many of the everyday problems of mathematical composition. The 

American Mathematical Society, A M S, which has supported T E X from its 

beginning, has developed a set of packages to assist authors in composing 

even more complicated formulas with L AT E X. Known as A M S-L AT E X, it 

includes many extra commands, further environments to supplement 

eqnarray, and many additional symbols. This system is presented in 

Chapter 12. 

In any event, typesetting mathematics is a very complicated issue. It 

was for this reason that Donald Knuth selected the $ sign as the switch to 

math mode: in the old days of movable lead type, setting mathematical 

formulas was an expensive business because of the extra workload it 

entailed. Even today, electronic typesetting is far more involved for math 

than for simple text.

6 

Graphics Inclusion and 

Color 

The two topics of this chapter have one thing in common: neither is 

handled by T E X at all, and thus neither by L AT E X. Outputting graphics 

files produced by other programs, or printing in color, are functions of 

the output device, that is, the printer, the monitor, the electronic output 

file, and these are controlled by the DVI driver program that converts the 

DVI output of L AT E X into the final output form. For each output type, a 

different set of commands is needed to switch color or to send an image. 

What T E X does have is a way of sending special driver-specific commands 

to the driver by placing them in the.dvi file; the command to do 

this is even named \special. L AT E X exploits this feature to pass driver 

commands to the.dvi file, but to do so, it needs to know which driver is 

going to be used for the further processing. 

With the advent of L AT E X 2ε, a standardized syntax was introduced, 

replacing the driver-specific commands that abounded with L AT E X 2.09 

and which caused much confusion. At last the importation of external 

images had become a basic part of L AT E X usage, and was no longer an 

exotic, magical art. Well . . . , nearly. 

It is still necessary to tell L AT E X which driver will be used afterwards, 

as an option to the graphics and color packages, but once that is done, 

all the driver-specific special commands are loaded from a corresponding 

.def files, which the user does not have to worry about. The top-level 

syntax remains the same for all types. 

It is also important to realize that not all image formats can be handled 

by every driver. PostScript drivers for example will only accept PostScript 

images, and then only as encapsulated PostScript. Image format conversion 

may be required. 

6.1 The graphics packages 

Strictly speaking, these packages, referred to as the graphics collection, 

are extensions to the basic L AT E X installation. We describe them here 

153

154 Chapter 6. Graphics Inclusion and Color 

because of their indispensability in producing professional documents 

and camera-ready copy. Provided by Sebastian Rahtz and David Carlisle 

and many other contributors, they can be expected to be part of every L AT E X 

installation. A basic manual is provided in a file named grfguide.tex, 

which may already be preprocessed to.dvi or some other form. Even more 

details on these packages and on advanced applications of PostScript to 

L AT E X are to be found in the book The L AT E X Graphics Companion (Goossens 

et al., 1997). 

The commands defined by these packages are the building blocks 

for other packages that either emulate the older driver-specific ones or 

provide a more comfortable syntax for these functions. As long as these 

other packages are based on graphics and color, they should be equally 

compatible with all the supported drivers. 

These features are not limited to PostScript drivers. As long as a 

driver can support the inclusion and manipulation (scaling, rotation) of 

graphics, and/or the use of color, a.def file can be written to enable it to 

make use of the standardized graphics and color commands. 

Driver names that may be used as options are 

dvipdf 

dvipdfm 

dvips 

dvipsone 

dviwin ◭⊲⇑ 

emtex ◭⊲⇑ 

oztex 

pctex32 

pctexhp ◭⊲⇑ 

pctexps 

pctexwin ◭⊲⇑ 

pdftex 

tcidvi ⇑ 

textures 

truetex ⇑ 

vtex 

(Limited functionality: ◭⊲ no color support; no scaling; ⇑ no rotation.) 

The xdvi previewer (and its Windows equivalent windvi) work with 

PostScript graphics; although there is also an xdvi option, it is only an 

alias for dvips. 

A driver option must be specified when loading the graphics and 

color packages, for example, as 

\usepackage[dvips]{graphics,color} 

However, it is possible to establish a default for the local configuration, 

as described in Section 6.1.7. It is conceivable that your installation has 

already done this, so it is worth doing an experiment to see if a default 

driver option already exists. 

6.1.1 Importing external graphics 

We wish to have a graphics file produced by some other program included 

in the document, possibly scaled to a desired size or rotated by 90 ◦ . One 

wants to do by computer what used to be done with scissors and glue. 

There are two packages available for importing and manipulating external 

graphics files: the basic graphics package and the more extended 

graphicx one. They both offer identical functionality, differing only in 

their syntax.

6.1. The graphics packages 155 

In addition to the driver names, there are some other options that may 

be invoked when the packages graphics or graphicx are loaded: 

draft does not import but places a framed box where the graphic 

would appear, with the file name printed inside; this speeds up 

the processing considerably when one is only working on the 

text; 

final counteracts draft; needed when the draft option has been 

issued globally in \documentclass; 

hidescale 

leaves blank space where scaled text should be; 

hiderotate 

leaves blank space where rotated text should be; this and the option 

hidescale are useful if the previewer cannot handle scaling 

or rotation; 

hiresbb 

look for bounding box values in %%HiResBoundingBox instead 

of the normal %%BoundingBox line. 

6.1.2 Importing with the graphics package 

Package: 

graphics 

The basic importing command with the graphics package is 

\includegraphics[llx,lly][urx,ury]{file name} 

where llx, lly are the coordinates of the lower left corner, and urx, ury 

those of the upper right corner of the bounding box containing the part 

of the picture that is to be included. In other words, they say where the 

scissors are to be applied. Units may be specified (like [3cm, 2in]) but 

if they are omitted, big points (bp, 72 per inch, 28.3464. . . per cm) are 

assumed. If only one optional argument is given, it is the upper right 

corner, and the lower left is assumed to be [0,0]. 

If no bounding box coordinates are given, the driver 

XII 

will obtain them some other way, depending on the 

IX III 

type of graphics file. For example, for the very com- 

VI 

mon encapsulated PostScript files with extension .eps, 

the bounding box information is extracted from the 

graphics file itself. The figure at the right is stored in such a file and is 

included simply with the command 

\includegraphics{clock} 

(It is not necessary, nor recommended, to include the file extension.eps 

in the file name designation; L AT E X will automatically test for all the extensions 

allowed by the selected driver option, something that permits 

greater flexibility in the source text.)


With \includegraphics*, the figure is clipped, so that any drawing 

outside the specified bounding box is suppressed. This is useful if only 

part of a figure is to be reproduced. It is also vital with some perverse 

figures that paint the whole page white! 

Scaling 

The graphics file in the above commands is transferred to the document in 

its original size. In order to rescale it, there are two commands available: 

\scalebox{h scale}[v scale]{text} 

which applies horizontal and vertical scale factors to the contents text; if 

v scale is omitted, it is the same as h scale; 

\resizebox{h length}{v length}{text} 

adjusts the figure to fit into the specified horizontal and vertical sizes; if 

either length is given as !, the one scale factor is used for both dimensions. 

A *-form allows v length to refer to the height + depth of the box, rather 

than just to the height. In both cases, the contents text may be an 

\includegraphics command, but it may also be any arbitrary text. 

Reflection 

The contents of a box may be reflected horizontally with 

Rotation 

\reflectbox{text} 

Rotation of a box about the left-hand end of its baseline is done with 

\rotatebox{angle}{text} 

where angle is in degrees, and the rotation is counterclockwise. 

IX 

XII 

figure to a height of 2 cm and then rotated it by 30◦ , 

using the commands 

To illustrate this, we have scaled the previous clock 

III 

VI 

\rotatebox{30}{\resizebox{!}{2cm}{% 

\includegraphics{clock}}} 

For demonstration purposes, we have added framed boxes around the 

figure before and after the rotation. It is the inner (tilted) box that is 

2 cm high, while the overall outer box is somewhat higher and broader 

due to its inclined contents. Without these frames the figure seems to 

have extraneous space above and to the left, something that can be very


puzzling even to experienced users. (The frames would not normally 

appear with the above commands.) 

Exercise 6.1: Copy the lines 

at the right to a file named 

demo.eps and then include it 

in a L AT E X document, with some 

normal text above and below it. 

The result should appear as: 

%!PS-Adobe-3.0 EPSF-3.0 

%%BoundingBox: 169 158 233 242 

220 200 moveto 

200 200 20 0 360 arc 

170 170 moveto 

230 220 lineto 





stroke 

showpage 

Note: This file can be copied from the enclosed CD in the directory 

\books\Kopka_and_Daly\, where it is also available in PDF format. 

6.1.3 Importing, scaling, rotating with the graphicx package 

Package: If one selects the graphicx rather than the graphics package, a different 

graphicx interface is available for both importing and rotation, one making use of 

keys and values: 

\includegraphics[key=value,. . . ]{file name} 

The keys are of two types: those that take a numerical value, and those 

that are flags with the values true or false. Simply giving the name of a 

flag without a value is equivalent to setting it to true. Possible keys and 

their values are: 

scale= number; enters the number by which the figure size should be 

magnified over its natural size; 

width= length; specifies the width to which the figure should be scaled; 

if height not given, it is scaled with the same factor as the width; 

height= length; specifies the height to which the figure should be scaled; 

if width is not given, it is scaled with the same factor as the height; 

totalheight= length; like height but specifies the height plus depth of 

the figure; should always be used in place of height if the figure 

has been rotated; 

keepaspectratio (=true|false); if both height and width are specified, 

this flag ensures that the original height/width ratio remains 

unchanged; the figure will not exceed either of the given dimensions;


angle= number; the angle by which the figure is to be rotated counterclockwise, 

in degrees; any height or width specifications coming 

before this key are also rotated, so that the height becomes the 

width, while the width becomes either the height (positive angle) or 

depth (negative angle); 

origin= loc; determines the point about which the rotation occurs; default 

is bl for bottom left corner; also possible are c for center, t for 

top, r for right, and B for baseline; any sensible combination, such 

as tr, is allowed; 

draft (=true|false); like the draft package option but applied to the 

one graphics file; the figure is not imported, but rather a framed box 

of the correct size is printed containing the name of the file; 

clip (=true|false); suppresses the printing of any graphic outside the 

bounding box; 

bb= llx lly urx ury; enters the coordinates of the bounding box manually, if 

they are missing or incorrect in the graphics file, or to be deliberately 

altered; the specifications are four lengths separated by blanks; units 

may be given, but if omitted, big points (bp) are assumed; this option 

should never be used with pdfT E X; 

viewport= llx lly urx ury; specifies the bounding box but relative to the 

lower left corner of the existing one; this is the preferred method for 

correcting the bounding box, or (with clip) to select only a portion 

of the whole figure; with pdfT E X, this option must be used rather 

than bb; 

trim= dllx dlly durx dury; reduces the existing bounding box by the 

amounts specified; 

hiresbb (=true|false); like the hiresbb package option but applied 

to the one graphics file; reads bounding box information from the 

%%HiResBoundingBox line in the graphics file. 

The keys are all optional; they are included as needed. Their order is 

not important other than that angle interchanges any previous height 

and width meanings. The sets of key/values are separated from each 

other by commas. It is often best to set width relative to the current line 

width \linewidth, say as width=0.8\linewidth. 

With the key/value syntax, the tilted, scaled graphic on page 156 is 

produced with 

\includegraphics[height=2cm,angle=30]{clock}


For compatibility with the graphics package, there also exists an 

\includegraphics* version that clips the imported figure; this is equivalent 

to including the key clip. 

With the graphicx package, the \rotatebox command is similarly 

redefined to accept the optional key origin. 

Exercise 6.2: Include the graphics file demo.eps from Exercise 6.1 scaled by a 

factor of 2 and rotated by 45 ◦ using the \includegraphics command with the 

graphicx package. Experiment with various keys, in particular with height and 

width together. 

6.1.4 Additional graphics packages 

Package: Sebastian Rahtz has provided a package epsfig that not only updates the 

epsfig earlier (LAT EX 2.09) version, but also re-implements Rokicki’s epsf package 

by means of the graphics commands. This is helpful for users who are 

accustomed to those syntaxes. For epsf, this is 

\epsfysize=y size or \epsfxsize=x size 

\epsf[llx lly urx ury]{file name} 

The epsfig package also defines an importing command that makes 

use of the regular keys and values to enter its parameters: 

\epsfig{file=file name,key=value,. . . } 

For compatibility with some older versions, there is a \psfig command 

which is synonymous with the above. 

The epsfig package is included in the bundle of graphics packages 

and drivers. 

Package: Another extra package in the graphics bundle is lscape, by David 

lscape Carlisle. This defines a landscape environment that prints its contents 

rotated 90◦ on a page for itself. Head and footlines remain as normal. 

This is intended primarily for inserting figures that are in landscape mode, 

that is, wider than they are high. 

Package: The rotating package by Sebastian Rahtz and Leonor Barroca tries 

rotating to make the interface for rotation somewhat simpler. It defines 

\begin{sideways} text \end{sideways} 

\begin{turn}{angle} text \end{turn} 

\begin{rotate}{angle} text \end{rotate} 

\turnbox{angle}{text} 

where sideways rotates text by 90 ◦ , turn by an arbitrary angle. The 

environment rotate and command turnbox are equivalent: they rotate 

but in a box of zero size, so that the contents overlap the surroundings. 

This package is not part of the graphics bundle and must be obtained 

separately.


6.1.5 Superimposing images 

❛ ❛ 

❍ 

❛ 

❍ 

❛ ❛ 

❍ 

❛ 

❍ 

❛ 

❍ 

❛ 

❍ 

❛ ❛ 

❍ 

❛ 

❍ 

❛ 

❍ 

❛ ❛ 

❍ 

❛ ❍ 

❛ ❍ 

❛ ❍ 

❛ ❛ 

❍ 

❛ ❍ 

❛ ❍ 

❛ ❛ 

❍ 

❍ 

❤❤❤❤ 

❵ 

❤ 

❵ 

❤ 

❵ 

❤❤ 

❵ 

❤ 

❵ 

❤ 

❵ 

❤❤ 

❵ 

❤ 

❵ 

❤ ❵ 

❤ ❵ 

❤❤ 

❵ 

❤ 

❵ 

❤ ❵ 

❤❤ 

❵ 

❤ 

❵ 

❤ ❵ ❵ ❵ 

✄ 

✄ 

✄✄✄✄✗ 

Blow-Up 

Figure 6.1: An embellished image file 

Sometimes it is desired to superimpose one image file over part of another, 

! say to add an inset in a corner. Or one may want to add text elements, or even 

lines and arrows pointing to some feature in the image. One might even want 

to make some area of the image into a link to another document. The picture 

environment described in Section 13.1 can be used to allow exact placement of 

such objects. 

For example, in the figure on this page, we embellish the T E XLive welcome 

image from Figure B.1 on page 383 by adding a blow-up of a small section, 

pointing a labeled arrow to it, and making the area around the text ‘T E XLive!’ into 

a link. We do this with the following input: 

\setlength{\unitlength}{0.01\linewidth} 

\setlength{\fboxsep}{0pt} 

\setlength{\fboxrule}{1.5pt} 

\begin{picture}(80,80) 

\thicklines 

\put(0,0){\includegraphics[width=.8\linewidth]{texlive}} 

\put(57,26){\framebox(14.2,14.6){}} 

\drawline(57,26)(-10,36.7)


\drawline(71.2,26)(20.5,36.7) 

\drawline(71.2,40.6)(20.5,67.2) 

\drawline(57,40.6)(-10,67.2) 

\put(-10,37){\fbox{\includegraphics[width=.3\linewidth, 

viewport=390 180 490 280,clip]{texlive}}} 

\put(-9,19){\vector(1,4){4}} 

\put(-9,18){\makebox(0,0)[t]{\large Blow-Up}} 

\put(43,2){\hyperlink{info}{\makebox(30,7){}}} 

\end{picture} 

Here the unit length for the picture placement has been set to 1/100 of the 

current line width \linewidth, so we can conveniently use 100 units for the full 

width. The picture environment reserves a space of 80×80 units; the image 

file texlive is placed at the lower left corner (location 0,0) and set to a width 

of 80% of \linewidth. We then draw a framed box around a portion of this 

image, and then add that portion as a second image (selected with viewport 

and clipped) as a blow-up. The \drawline commands (from the epic package) 

join the corners of the original and blown-up areas. The order here is important 

since later images overwrite the earlier ones. Then we add an arrow (\vector) 

and the text ‘Blow-up’ (with \makebox). Finally, the \hyperlink command from 

the hyperref package turns the 30×7 unit area around the text ‘T E XLive!’ into a 

active link to the target named info. 

One can also place objects outside of the reserved picture area, so arrows 

could be added pointing from the main text to some part of an image. This is 

more appropriate for the production of slides (Chapter 15) where one wants rigid 

control over the relative locations of graphics and text. 

6.1.6 How importation can go wrong 

In order to understand better what can go wrong when a graphics file is imported, 

! and to know what to do about it, it is important to realize how the interplay 

between L AT E X and the driver program functions. 

L AT E X has no idea what is in the graphics file; for it, the figure is simply a box 

of a given height, width, and depth, as indeed are all the characters that L AT E X 

processes. The information on the graphic’s natural size is somehow obtained, 

either by information in the graphics file itself, or through the optional entries 

in the \includegraphics command or equivalent. After scaling and rotating, 

L AT E X knows the final size that it must reserve in the output text for the figure. 

What is then written to the .dvi file is the name of the graphics file and 

information on how it should be transformed. Just how this information is 

coded depends on the graphics driver selected. When the printer driver program 

processes the .dvi file, it interprets these special instructions, reads in the 

specified graphics file, performs the transformations, and places the result where 

L AT E X has said it should go. The end result is that the area inside the designated 

bounding box coincides with the box that L AT E X has reserved for it. If the bounding 

box information is incorrect, the figure is obviously going to be misplaced.


PostScript output 

Graphics inclusion was originally developed for PostScript graphics adhering to 

the encapsulated specification that was designed to allow PostScript code to be 

included within another PostScript file. Tomas Rokicki and his dvips driver broke 

the ground in this field and thus established the standards. It is for that reason 

that we place so much emphasis on PostScript graphics here. Encapsulated 

PostScript files may not contain any commands that reset the whole graphics 

page, and furthermore, they must contain a comment line of the form 

%%BoundingBox: llx lly urx ury 

giving the coordinates of the lower left and upper right corners of the bounding 

box. The units are never specified, always being big points. 

PDF output 

Today PDF is an important alternative to PostScript, and the pdfT E X program 

(Section 10.2.3) is the most convenient way to produce it with L AT E X. However, 

pdfT E X is not only a replacement for the classical T E X program, producing identical 

output, it also combines the functionality of a DVI-to-PDF driver. Thus no 

intermediate .dvi file is generated at all, with the .pdf file being generated 

directly from the L AT E X source file in one operation. This has consequences for 

graphics inclusion with pdfT E X. First of all, the number of allowed graphics 

formats is increased: images may be PNG (.png files), JPEG (.jpg or .jpeg files), 

TIFF (.tif or.tiff files) or even a single image in a one-page PDF file. Ironically, 

PostScript images are absolutely out! Secondly, the concept of bounding box does 

not really apply to these formats, since they have a natural size anyway. The 

viewport option should therefore be used to make a cutout from the original 

rather than bb. 

An alternative method of obtaining PDF output is to employ the true DVIto-PDF 

driver, dvipdfm, written by Mark A. Wicks. In this case, a .dvi file is 

produced as usual (with the graphics option dvipdfm), and then the PDF output 

is generated with this conversion program. The same set of graphics formats is 

allowed as for pdfT E X, for much the same reasons. 

The third method would be to generate DVI, then PostScript, and then convert 

it to PDF. As far as the graphics are concerned, this is the same as PostScript. 

Having pointed out how the importation takes place, we can now discuss 

what can go wrong along this chain of processes. 

Problems with importation: pdfT E X 

The single most common error when importing a graphics file with pdfT E X 

is that it does not exist in one of the permitted formats. Many users who 

are used to using classical T E X with PostScript simply switch to pdfT E X 

and are astonished to get the error message: 

! LaTeX Error: Unknown graphics extension: .eps. 

As explained above, pdfT E X cannot handle PostScript files, so the graphics 

must be converted to, or regenerated in, another format.


The above error message will appear only if one has specified the full 

name of the graphics file including the extension .eps. If the extension 

is absent, which we recommend, the error message will report that the 

file cannot be found. This means that it cannot be found with any of the 

allowed extensions. It might very well be that it exists with the extension 

.eps, but since that is not allowed for pdfT E X, it will not be sought. 

To maintain flexibility, one can have the same graphics file available 

both as an .eps and as a .pdf file. Leaving the extension off in the 

\includegraphics command means one does not need to make any 

changes to the file name when switching between regular T E X and pdfT E X. 

Problems with importation: PostScript 

The most common problems encountered when importing encapsulated 

PostScript files are listed here. 

No bounding box. If the bounding box information is totally missing 

from the graphics file, L AT E X issues the error message 

! LaTeX Error: Cannot determine size of graphic 

in ... (no BoundingBox). 

The solution is to determine the bounding box coordinates somehow 

(see next point) and to include them, either in the \includegraphics 

command, or by editing the graphics file itself. However, if there 

really is no bounding box information in the file, it is unlikely to conform 

to the encapsulated standard and will cause other problems. 

The placement is incorrect. Both L AT E X and the driver process without 

error messages, but the figure is either displaced from the expected 

position, or is far too small. 

Most likely the bounding box information is incorrect. Many applications 

that produce PostScript files are too lazy to calculate the true 

bounding box, or they think they are generating a whole page with 

a figure somewhere in the middle. In either case, the bounding box 

corresponds to the full page even though the printed figure occupies 

only a portion of it. 

Find the true bounding box by one of the following methods: 

1. Print the figure, mark the lower left and upper right corners 

of the box containing the figure, and measure their distances 

from the left and bottom edges. Enter these distances in the 

\includegraphics command, or edit the PostScript file. In the 

latter case, convert to big points. 

Difficulties with this are that some encapsulated PostScript files 

cannot be printed on their own, and that the left and bottom


edges of the paper need not be the exact lines from which the 

printer really measures. 

2. Include the figure in a short L AT E X file such as 

\setlength{\fboxsep}{-\fboxrule} 

\fbox{\includegraphics{test.eps}} 

and print the output. The apparent bounding box appears as 

a framed box. Measure the true bounding box relative to the 

left and bottom edges of the apparent one, and enter the values 

in the \includegraphics command with the viewport= key. 

If necessary, scale the figure down to fit on the page, but then 

remember to increase the measurements by the same scale 

factor. 

3. Use the GhostView program to fix up the bounding box, either 

automatically or manually. This is the most convenient method 

if you have this utility for viewing and manipulating PostScript 

files. 

Immovable graphic. It does not shift, nor scale nor rotate, no matter what 

is specified. In this case, it violates the encapsulated PostScript rules 

and contains some global plotting commands. Graphics produced 

by word processing programs are notorious for this. Often the 

offending command is setpagedevice. 

There is little that can be done to correct this, other than trying to regenerate 

the graphics file with an option for encapsulated PostScript. 

Judicious editing can remove the troublesome lines, but this could 

result in the file becoming totally unreadable. 

6.1.7 Configuring graphics importation 

Although the graphics syntax has been standardized, and most of the driver- 

! specific coding hidden in the .def files, there are still a number of items that 

must be set up for any particular installation and operating system. These are 

most conveniently placed in the local configuration file graphics.cfg which is 

read in if it is present. 

Many installations come with a ready-made configuration file, so there may 

not be any need to make up your own. On the other hand, you may want to edit 

or replace it for you own purposes. 

Default driver 

The choice of driver option must always be given, but a local default can be 

specified in the configuration file. For example, if dvips is the standard graphics 

driver, you may add

\ExecuteOptions{dvips} 


to graphics.cfg and you do not need to give the option dvips when loading 

graphics or graphicx. You may still specify any other graphics driver option if 

you wish. 

Many supplied configuration files go further and determine whether pdfT E X 

is activated or not; if so, the default graphics option is set to pdftex, otherwise 

to dvips. 

The rest of the configuring commands in this section can be issued either in 

the graphics.cfg file, or in the document. 

Search path for graphics files 

One can specify the directories where L AT E X is to look for graphics files with 

\graphicspath{dir list} 

where dir list is a list of directory names, each enclosed in brackets { }, with no 

other separator. The syntax of the local operating system must be used. Without 

this command, L AT E X searches for graphics files in the same directories as for all 

other T E X files. Example, 

\graphicspath{{figs/}{eps/}} for Unix, DOS, Windows 

\graphicspath{{:figs:}{:eps:}} for Macintosh 

\graphicspath{{[.figs]}{[.eps]}} for VMS 

Note: The Unix syntax works even with DOS and Windows, since the normal 

backslash in the directory names must be replaced by a slash / to avoid looking 

like a command name. 

Default extensions 

A list of default extensions for the graphics files can be defined with 

\DeclareGraphicsExtensions{ext list} 

This means that only the root name of the file must be given and L AT E X will 

attempt to find it by attaching all the possible extensions. For PostScript drivers, 

the ext list is usually set to .eps,.ps. At our installation, we also include the 

non-standard extension.psc. Note that the above command does not add to the 

list of extensions but rewrites it anew; if you wish to add to the list, you must 

include all the allowed extensions in the one declaration. 

Graphics types 

Defining the extensions is only part of the task: one must also associate each 

extension with a graphics type so that L AT E X knows how to process it. PostScript 

recognizes only one type, eps, encapsulated PostScript, but there do exist other 

types such as bmp and pcx for other drivers. For the non-standard.psc extension 

above, we must also give 

\DeclareGraphicsRule{.psc}{eps}{}{} 

to inform L AT E X that this extension belongs to type eps. The other two (empty) 

arguments specify that the bounding box information is to be read from the file 

itself, and that no other program needs to be applied to the file.


Compressing graphics files 

Since PostScript files are often extremely large, it makes sense to try to compress 

them with either the zip or gzip programs. In such a case, the .eps file is 

replaced by a file with extension.zip, or.eps.gz, or.eps-gz. Two problems now 

arise: first, L AT E X cannot read such files to obtain the bounding box information, 

and secondly, the driver needs to unpack such a file to include it in the final 

output. This can be accomplished with, for example, 

\DeclareGraphicsRule{.eps.gz}{eps}{.eps.bb}{‘gunzip -c #1} 

which establishes the graphics type as eps, with the bounding box information in 

the file of the same name and extension.eps.bb, and that the operating system 

command gunzip -c must be applied to the file (represented as #1). The single 

quote ‘ is required to indicate a system command. The %%BoundingBox line of 

the original file must be copied and stored in the.eps.bb file. 

Such decompression rules are system dependent and thus need to be configured 

for the local installation. For example, under the VMS operating system, 

the gzip program produces files with extension .eps-gz and decompression 

is performed with gzip -d rather than with gunzip. The corresponding rule 

becomes 

6.2 Adding color 

\DeclareGraphicsRule{.eps-gz}{eps}{.bb}{‘gzip -d -c #1} 

The color package accepts the same driver options listed on page 154 

for the graphics package. In addition, it also recognizes the options: 

monochrome 

to convert all color commands to black and white, for previewers 

that cannot handle color; 

dvipsnames 

makes the named color model of dvips (Section 6.2.1) available 

to other drivers; 

nodvipsnames 

disables the named model for dvips, to save memory; 

usenames 

loads all the named colors as defined ones; again, see Section 6.2.1 

for details. 

A local configuration file color.cfg can be set up in the same way as for 

! the graphics package. The default driver option is specified in exactly the same 

way as in Section 6.1.7. 

Colors are specified either by a defined name, or by the form 

[model]{specs}

6.2. Adding color 167 

where model is one of rgb (red, green, blue), cmyk (cyan, magenta, yellow, 

black), gray, or named. The specs is a list of numbers from 0 to 1 giving the 

strengths of the components in the model. Thus [rgb]{1,0,0} defines 

red, [cmyk]{0,0,1,0} yellow. The gray model takes only one number. 

The named model accesses colors by internal names that were originally 

built into the dvips driver, but which may now be used by some other 

drivers too. This model is described in Section 6.2.1. 

A color can be defined with 

\definecolor{name}{model}{specs} 

and then the name may be used in all the following color commands. 

Certain colors are automatically predefined for all drivers: red, green, 

blue, yellow, cyan, magenta, black, white. 

In the following color commands, col spec is either the name of a 

defined color, like {blue}, or [model]{spec}, like [rgb]{0,1,0}. 

\pagecolor col spec sets the background color for the current and 

following pages; 

\color col spec is a declaration to switch to setting text in the given 

color; 

\textcolor col spec{text} sets the text of its argument in the given 

color; 

\colorbox col spec{text} sets its argument in a box with the given color 

as background; 

\fcolorbox col spec1 col spec2{text} like \colorbox, with a frame of 

col spec1 around a box of background color col spec2; the two specifications 

must either both be defined ones, or both use the same 

model, which is given only once; for example, \fcolorbox{red} 

{green}{Text} sets ‘Text’ in the current text color on a green background 

with a red frame; 

\normalcolor switches to the color that was active at the end of the 

preamble. Thus placing a \color command in the preamble can 

change the standard color for the whole document. This is the 

equivalent to \normalfont for font selection. 

Normally one would try to define all the colors needed as names for the 

col spec entries. This simplifies changing the color definition everywhere 

should fine-tuning be required after the initial printed results are seen. 

The same color definition can produce quite different effects on different 

printers. Even the display on the monitor is no reliable guide as to how 

the output will appear on paper.


6.2.1 The named color model 

One very useful color model is called named and is based on the 68 

predefined internal colors of the dvips PostScript driver. Sample names 

are BurntOrange or DarkOrchid. This model can be activated for other 

drivers with the option dvipsnames, in which case one can define colors 

as, for example 

\definecolor{titlecol}{named}{DarkOrchid} 

The color titlecol can then be used as col spec in the various color 


The named colors can be defined with their own names if one invokes 

the option usenames, which effectively declares 

\definecolor{BurntOrange}{named}{BurntOrange} 

and so on, for all 68 colors. 

It is possible to generate a palette of the named colors by processing 

the following short L AT E X file and sending the output to the desired printer. 

\documentclass[12pt,a4paper]{article} 

\usepackage[dvipsnames]{color} 

\usepackage{multicol} 

\pagestyle{empty} 

\setlength{\oddsidemargin}{0pt} 

\setlength{\textwidth}{16cm} 

\setlength{\textheight}{22cm} 

\setlength{\parindent}{0pt} 

\setlength{\parskip}{0pt} 


\renewcommand*{\DefineNamedColor}[4]{% 

\textcolor[named]{#2}{\rule{7mm}{7mm}}\quad 

\texttt{#2}\strut\\} 

\begin{center}\Large Named colors in \texttt{dvipsnam.def} 

\end{center} 

\begin{multicols}{3} 

\input{dvipsnam.def} 

\end{multicols} 


Remember, each printer can reproduce the colors differently, so it is 

important to test this table with every color printer that might be used. 

Exercise 6.3: Copy the above lines to a file named palette.tex or copy it from 

the enclosed CD from \books\Kopka_and_Daly\. Process it and view the output 

on a color monitor or send it to a color printer.

7 

Floating tables and 

figures 

Whether a figure is imported with \includegraphics (Chapter 6) or 

produced with the picture environment of Chapter 13, it is inserted in 

the text where the drawing or importing commands are issued, coming 

between the previous and following texts. This can present the same 

difficulties as for tables, described in Section 4.8.5: if the figure is so 

high that it no longer fits on the current page, that page is prematurely 

terminated and the figure is placed at the top of the next page, with too 

much empty space left on the original page. 

In Section 4.8.5 we mention how this problem is solved for tables. Here 

we now give a complete description of the float procedures that apply to 

both figures and tables. 

7.1 Float placement 

L AT E X does make it possible to float figures and tables, together with their 

headlines and captions, to an appropriate location without interrupting 

the text. This is invoked with the environments 

\begin{figure}[where] figure \end{figure} 

\begin{figure*}[where] figure \end{figure*} 

\begin{table}[where] table \end{table} 

\begin{table*}[where] table \end{table*} 

The *-forms apply only to the two-column page format and insert the 

figure or table across both columns instead of the normal single column. 

They function exactly the same as the standard forms when the page 

format is single column. 

In the above syntax, figure and table are the texts for the contents of the 

float, either a picture or tabular environment, or an \includegraphics 

command, together with a possible \caption command, as described 

later in Section 7.4. 

169

170 Chapter 7. Floating tables and figures 

The argument where specifies the allowed locations for the figure or 

table. There are several possibilities so that where consists of from zero 

to four letters with the following meanings: 

h here: the float may appear at that point in the text where the 

environment is typed in; this is not permitted for the *-form; see 

also the here package and the H parameter, page 178; 

t top: the float may appear at the top of the current page, provided 

there is enough room for both it and the previous text; if this is not 

the case, it is added at the top of the next page; the subsequent text 

continues on the current page until the next normal page break; (for 

two-column format, read column in place of page); 

b bottom: the float may be placed at the bottom of the page; the 

subsequent text continues until the room left on the current page 

is just enough for the float; if there is already insufficient room, the 

float will be put at the bottom of the next page; this is not permitted 

for the *-form; 

p page: the float may be put on a special page (or column) containing 

only figures and/or tables; 

! used together with any combination of the other letters, suspends 

all the spacing and number restrictions described in Section 7.3. 

The argument values may be combined to allow several possibilities. If 

none is given, L AT E X assumes the standard combination tbp. 

The placement arguments permit certain possibilities for locating the 

float, but the actual insertion takes place at the earliest possible point in 

accordance with the following rules: 

• no float appears on a page prior to that where it is defined; 

• figures and tables are output in the order in which they were defined 

in the text, so that no float appears before a previously defined float 

of the same type; figures and tables may, however, be mixed in their 

output sequence; in two-column format, the double column *-floats 

may also appear out of sequence; 

• floats will be located only at one of the allowed positions given by 

the placement argument where; without an argument, the standard 

combination tbp is used; 

• unless the ! is included in where, the positioning obeys the limitations 

of the style parameters described in Section 7.3; 

• for the combination ht, the argument h takes priority; the float is 

inserted at the point of its definition even if there is enough room 

for it at the top of the page.

Package: 

afterpage 

7.3. Style parameters for floats 171 

When any of the commands \clearpage, \cleardoublepage, or 

\end{document} is given, all floats that have not yet been output will 

be printed on a separate page or column regardless of their placement 

arguments. 

Sometimes LAT EX gets stuck on a float which holds up the entire queue 

until the end of the document. One way to clear the queue is to issue 

\clearpage right after the troublesome float. However, that would insert 

a new page at that point, something that may not be desired. The 

package afterpage in the tools collection (Section B.5.4) provides the 

command \afterpage which executes its argument at the end of the 

current page. Thus \afterpage{\clearpage} solves this problem by 

delaying the \clearpage command until the actual page break. 

7.2 Postponing floats 

Occasionally one wants to prevent floats from appearing on a certain 

page, for example at the top of a title page. (L AT E X automatically corrects 

that case.) However, there are other situations where a float should be 

suppressed temporarily. One might want it at the top of a page, but not 

before the start of the section that refers to it. The command 

\suppressfloats[loc] 

sees to it that for the current page only no further floats of the specified 

placement loc should appear. If the optional loc is omitted, all floats are 

suppressed; otherwise loc may be either t or b, but not both. 

Note that \suppressfloats does not suppress all floats for the current 

page, but only further ones that come between the issuing of this 

command and the end of the page. Thus it is still possible for floats from 

a previous section to appear on the page. 

Package: Alternatively, the package flafter (Section B.5.3) may be loaded to 

flafter ensure that all floats appear only after their position in the text. 

The command \suppressfloats and the location parameter ! are 

attempts to give the author more control over the sometimes capricious 

actions of float placement. 

7.3 Style parameters for floats 

There are a number of style parameters that influence the placement of floats, 

! which may be altered by the user as desired. 

topnumber 

The maximum number of floats that may appear at the top of a page. 

bottomnumber 

The maximum number of floats that may appear at the bottom of a 

page.


totalnumber 

The maximum number of floats that may appear on any page regardless 

of position. 

dbltopnumber 

The same as topnumber but for floats that extend over both columns 

in two-column page format. 

The above parameters are all counters and may be reset to new values with 

the command \setcounter{ctr}{num}, where ctr is the name of the counter 

and num the new value that it is to take on. 

\topfraction 

A decimal number that specifies what fraction of the page may be used 

for floats at the top. 

\bottomfraction 

A decimal number that specifies what fraction of the page may be used 

for floats at the bottom. 

\textfraction 

The fraction of a page that must be filled with text. This is a minimum, 

so that the fraction available for floats, whether top or bottom, can 

never be more than 1−\textfraction. 

\floatpagefraction 

The smallest fraction of a float page that is to be filled with floats before 

a new page is called. 

\dbltopfraction 

The same as \topfraction but for double column floats in two-column 

page format. 

\dblfloatpagefraction 

The same as \floatpagefraction but for double column floats in 

two-column page format. 

These style parameters are altered with \renewcommand{cmd}{frac} where 

cmd stands for the parameter name and frac for the new decimal value, which 

in every case must be less than 1. 

\floatsep 

The vertical spacing between floats appearing either at the top or at the 

bottom of a page. 

\textfloatsep 

The vertical spacing between floats and text, for both top and bottom 

floats. 

\intextsep 

The vertical spacing above and below a float that appears in the middle 

of a text page with the h placement argument. 

\dblfloatsep 

The same as \floatsep but for double column floats in two-column 

page format. 

\dbltextfloatsep 

The same as \textfloatsep but for double column floats in twocolumn 

page format.

7.4. Float captions 173 

This group of style declarations are rubber lengths that may be changed with 

the \setlength command (Section 2.4.2). 

\topfigrule 

A command that is executed after a float at the top of a page. It may be 

used to add a rule to separate the float from the main text. Whatever it 

adds must have zero height. 

\botfigrule 

Similar to \topfigrule, but is executed before a float that appears at 

the bottom of a page. 

\dblfigrule 

Similar to \topfigrule, but for double column floats. 

These three commands normally do nothing, but they may be redefined if 

necessary. For example, to add a rule of thickness 0.4 pt below a top float, 

\renewcommand{\topfigrule}{\vspace*{-3pt} 

\rule{\columnwidth}{0.4pt}\vspace{2.6pt} } 

Because of the negative argument in \vspace*, the total vertical spacing is zero, 

as required. 

All the one-column style parameters also function within the two-column 

page format, but they apply only to floats that fill up one column. 

If the style parameters are set to new values within the preamble, they apply 

from the first page onwards. However, if they are changed within a document, 

they do not take effect until the next page. 

7.4 Float captions 

A figure caption or table title is produced with the command 

\caption[short title]{caption text} 

inside the figure or table environment. The caption text is the text that 

is printed with the float and may be fairly long. The short title is optional 

and is the text that appears in the list of figures or tables (Section 3.4.4). 

If it is missing, it is set equal to caption text. The short title should be 

given if the caption text is longer than about 300 characters, or if it is 

more than one line long. 

In the table environment, the \caption command generates a title of 

the form ‘Table n: caption text’, and in the figure environment ‘Figure n: 

caption text’, where n is a sequential number that is automatically incremented. 

In document class article, the figures and tables are numbered 

from 1 through to the end of the document. For the report and book 

classes, they are numbered within each chapter in the form c.n, where c 

is the current chapter number and n is the sequential number reset to 1 at 

the start of each chapter. Figures and tables are numbered independently 

of one another.


Table 7.1: Computer Center Budget for 2004 

Nr. Item 51505 52201 53998 Total 

1.1 Maintenance 130 000 15 000 145 000 

1.2 Network costs 5 000 23 000 28 000 

1.3 Repairs 25 000 6 000 31 000 

1.4 Expendables 68 000 68 000 

1. Total 160 000 74 000 38 000 272 000 

The \caption command may be omitted if numbering is unwanted, 

since any text included in the float environment will accompany the 

contents. The advantages of \caption over simple text are the automatic 

numbering and entries in the lists of figures and tables. However, a 

manual entry in these lists may be made as described in Section 3.4.4 

using 

\addcontentsline and \addtocontents 

A title, or headline, is produced with \caption when the command 

comes at the beginning of the material in the float environment: the 

number and text are printed above the table or figure. A caption is added 

below the object if the command comes after all the other float commands. 

In other words, the \caption is just another item within the float and 

whether its text appears at the top (as title) or below (as caption) depends 

on how the user places it. 

The caption text will be centered if it is shorter than one line, otherwise 

it is set as a normal paragraph. The total width may be adjusted to that of 

the table or figure by placing the command inside a parbox or minipage. 

For example, 

\parbox{width}{\caption{caption text}} 

The following demonstrations contain further examples of how text, 

table, and picture commands may be combined in floats. 

7.5 Float examples 

The first two tables are produced with the following texts: 

\begin{table} \caption{Computer Center Budget for 2004} 

\begin{tabular}{|l|l||r|r|r|r|} ... ... \end{tabular} 

\end{table} 

(This text was typed in before the last paragraph of the previous section. The 

second table is entered here in the current text.)

7.5. Float examples 175 

Table 7.2: Estimates for 2004. A continuation of the previous budget is 

no longer practical since, with the installation of the new computing system 

in 2003, the operating conditions have been completely overhauled. 

Nr. Item 51505 52201 53998 Total 

1.1 Maintenance 240 000 240 000 

1.2 Line costs 12 000 8 000 36 000 56 000 

1.3 Training 50 000 50 000 

1.4 Expansion 80 000 3 000 83 000 

1.5 Expendables 42 000 42 000 

1. Total 332 000 53 000 86 000 471 000 

\begin{table} 

\caption{\textbf{Estimates for 2004} \emph{A continuation...}} 

\begin{tabular}{|l|l||r|r|r|r|} ... ... ... \end{tabular} 

\end{table} 

Because the placement argument is missing from the table environments, 

the standard values tbp are used. The first was placed at the 

top on the facing page because it was typed in early enough (during the 

last section) that there was still room for it there. Then the command 

\suppressfloats (Section 7.2) was issued to prevent any more floats 

appearing on the same page. The second table is therefore forced to float 

to the top of this page. 

Narrow figures or tables may be set beside each other, as shown in 

the following example (which appears at the bottom of this page). (The 

picture environment is explained in Chapter 13.) 

\begin{figure}[b] 

\setlength{\unitlength}{1cm} 


\begin{picture}(5.0,2.5) ... ... \end{picture}\par 

\caption{Left} 


✻ 

Space for Left Figure 

2.5 cm 

✛ 

❄ 

5.0 cm ✲ 

Figure 7.1: Left 

✻ 

Space for Right Figure 

3.0 cm 

✛ 

❄ 

6.0 cm ✲ 

Figure 7.2: Right


\hfill 


\begin{picture}(6.0,3.0) ... ... \end{picture}\par 

\caption{Right} 


\end{figure} 

The two figures along with their captions are each set in a minipage 

environment of widths 5 and 6 cm. The minipages are separated from one 

another by an \hfill space. The positioning argument t has the effect 

that the minipages are aligned along their first lines (Section 4.7.3). The 

entire structure within the figure environment floats as a single entity. 

The question might now arise as to why, since the two figures have 

unequal heights and are supposedly aligned vertically along their top 

lines, their bottom edges are at the same level. The explanation is that 

a picture environment establishes an LR box (Section 4.7.1) to contain 

all the picture commands, and that is viewed by L AT E X as a single line of 

output text with the baseline at the bottom edge of the picture. In both 

minipages, the picture environment is the first entry and is therefore 

the first logical line of text. It is these baselines that are taken for the 

vertical alignment of the minipages. 

If the two pictures were to be aligned along their top edges, it would be 

necessary to include a dummy first line (Section 4.7.4) in each minipage 

before the picture environments. This could be something like \mbox{}, 

for example. 

The application of box commands within a float permits completely 

free positioning. If the caption text is to appear, say, beside the table or 

figure, instead of above or below it, the objects may be put into minipages 

or parboxes with suitable alignment arguments. Here is an example: 

\begin{table}[b] 

\centerline{\bfseries Results and Seat Distribution of the...} 

\mbox{\small 

\begin{minipage}[b]{7.7cm} 

Pros 

140 

Results and Seat Distribution of the 2003 General Election 

Cons 

186 

Ups 

Downs 

60 

42 

Seat Distribution 

in the Assembly of 

2004–2006 

Party Votes % 

Results: 

2003 General 

Election 

Pros 15 031 287 37.7 

Cons 12 637 418 31.7 

Ups 6 499 210 16.3 

Downs 4 486 514 11.3 

Others 1 201 078 3.0

7.6. References to figures and tables in text 177 


\mbox{}\\ \setlength{\unitlength}{0.75cm} 

\begin{picture}(5.75,5.0) ... ... ... \end{picture} 

\end{minipage} \hfill 

\parbox[t]{3.2cm}{\makebox[0cm]{}\\{\bfseries Seat ...} ...} 


\hspace{-3cm} 

\begin{minipage}[b]{7cm} 

\parbox[b]{2.5cm}{{\bfseries Results:}\\ General Election..} 

\hfill 

\begin{tabular}[b]{|l||r|r|} ... ... \end{tabular} 

\end{minipage} } 

\end{table} 

Here the vertical boxes are nested inside one another. The left side, consisting 

! of graphics and a title at the upper right, is contained in a minipage of width 

7.7 cm, in which the picture is in another minipage of width 4.4 cm while the text 

is in a parbox of width 3.2 cm. Both are aligned with the top lines. A dummy line 

containing \mbox{} (Section 4.7.4) is added before the first picture to provide 

a top line with which the second parbox can be aligned. 

The right-hand side consists of a minipage of width 7 cm, shifted to the left 

by 3 cm, aligned with the left-hand minipage along the bottom lines. It contains 

a parbox of width 2.5 cm for the text as well as a table, which is automatically a 

vertical box. Both of these are aligned with the bottom lines. 

7.6 References to figures and tables in text 

The automatic numbering of tables and figures has the consequence that 

the author does not know the numbers at the time of writing. Since he 

or she would like to be able to refer to these objects by number, as ‘see 

Figure 3’ or ‘Table 5 illustrates’, another means of referencing must be 

found. It is not sufficient to keep track of the number of \caption calls 

that have been made, since the document may not be written in the order 

in which it finally appears, and new figures or tables may be inserted 

during revision, or some removed. 

These problems are solved with the L AT E X cross-reference system, described 

in more detail in Section 9.2.1. The basic commands are 

\label{name} \ref{name} 

which assign a keyword name to the figure or table number that may be 

used as reference in the text. The keyword name may be any combination 

of letters, numbers or symbols. The assignment is made with the \label 

command given anywhere within the caption text of the \caption command; 

in the main text, the command \ref inserts the number that is 

associated with its keyword. 

This is best shown with an example. The budget table on page 174 

was actually written as


\caption{\label{budget04} Computer Center ... } 

so that Table \ref{budget04} produces ‘Table 7.1’ when used in the 

main text. 

There is a second referencing command \pageref to generate the 

page number where the referenced object is to be found. For example, a 

few lines up, the input text on page \pageref{budget04} was used to 

create the text ‘. . . on page 174’. 

7.7 Some float packages 

7.7.1 Placing a float right here 

Package: The float placement parameter h, described on page 170, only permits a 

here float to be placed at the current location, it does not force it to appear 

there. With Frank Mittelbach’s package here, an additional parameter H 

is provided which places the float immediately where it appears in the 

source text. If there is not enough room for it on the current page, a 

\pagebreak command is first issued. This will most likely lead to an ugly 

page break. 

7.7.2 Flowing text around a float 

Package: The package floatflt by Mats Dahlgren (based on the earlier and now 

floatflt obsolete floatfig by Thomas Kneser) permits one to mix text and floats 

side by side, with the text flowing around the float at the end, as in the 

next paragraph. 

A figure is placed within 

the text as shown in the box 

\begin{floatingfigure}[pos]{wth} 

at the right. Here pos de- 

Figure commands with or without a 

termines on which side the 

\caption command 

figure is to appear; it may 

\end{floatingfigure} 

be r (right), l (left), p (right 

for odd, left for even pages), or v (like p unless a package option is given, 

in which case, it is used). The mandatory argument wth is the width 

of the figure itself; in fact, some additional space will be included for a 

border between text and figure. It is advisable to include a \centering 

command at the start of the figure material. 

The material in the floatingfigure environment need not actually 

be a figure. In the above example, some boxed text in a minipage was 

used for demonstration purposes. However, if the \caption command 

is included in the environment, then a figure number will be printed, in 

sequence with those of the regular figure environment. 

Rather than giving the pos parameter each time, one can specify a 

default value as an option when loading the package with \usepackage:

7.7. Some float packages 179 

rflt (right), lflt (left), or vflt (variable, i.e. p). If no option is given, the 

default is vflt. 

\begin{floatingtable}[pos]{ 

\begin{tabular}{table specs} 

table entries 

\end{tabular}} 

Optional \caption command 

\end{floatingtable} 

To include a table with text, 

the usage is slightly different, as 

shown at the left. That is, the entire 

table appears in place of the 

wth for a figure, and only the caption 

in placed in the environment 

itself. 

Both these environments are 

to be placed at the start of the paragraph that is to flow around the figure 

or table. The flowing cannot start in the middle of a paragraph. 

There is some very tricky coding to get the flowing text to work properly, 

and needless to say, there are many things that can go wrong. Please 

read the accompanying manual for more details. 

7.7.3 Defining new floats and styles 

Package: For authors who are not satisfied with the standard float style, or who 

float need additional floating environments, we recommend the float package 

by Anselm Lingnau. With this package, one defines a new float type with 

\newfloat{name}{loc}{ext}[within] 

where name is the name of the new float environment, loc is the default 

placement letters (selection of tbhpH), ext is the extension of the file 

containing the list of this float type, and the optional within is the name 

of a counter (e.g., chapter) within which the numbering is reset. 

For example, to define a plate environment that behaves just like 

figure but which is numbered separately as Plate xxx, with the numbers 

reset within each chapter, give 

\newfloat{plate}{p}{lop}[chapter] 

\floatname{plate}{Plate} 

The first line defines the new environment, with default placement on a 

float page. The second states that the captions associated with this float 

are to start with the word ‘Plate’; without this, they would be labeled as 

‘plate’, the same as the environment name itself. 

The information for the list of plates is stored in a file with extension 

.lop, in the same way that the list of figures is in a file with extension 

.lof (Section 3.4.4). This list is printed by reading in this file with the 

command 

\listof{plate}{List of Plates} 

where ‘List of Plates’ is the list title. 

The default placement parameters for any float environment, including 

the standard figure and table ones, can be redefined with


\floatplacement{name}{loc} 

The float package also defines 4 float styles: 

plain much like standard L AT E X except the caption is always below the 

float contents; 

plaintop 

like plain but with the caption always above the contents; 

boxed in which the contents are in a framed box with the caption below 

it; 

ruled with the caption above, and with rules above, below, and between 

caption and float contents. 

One selects a float style by issuing \floatstyle{style}, where style is 

one of the above possibilities. The style then applies to all new float 

environments defined with \newfloat, until a new style is issued. 

Note: the style selection affects the subsequent float definitions, not 

the float environments that are then used. To change the float style for 

the standard figure and table floats, issue 

\floatstyle{ruled} 

\restylefloat{figure} 

\restylefloat{table} 

The float package also contains the H placement parameter for ‘here 

and I mean it’, so it is not necessary to include the here package with it.

8 

User Customizations 

L AT E X allows the user to define his or her own commands and environments. 

However, since these make extensive use of the L AT E X counters and lengths, 

we will first present a more detailed discussion of these objects and how 

they may be manipulated. 

8.1 Counters 

8.1.1 L AT E X counters 

L AT E X manages a number of counters by giving them initial values at the 

start and by changing these values when certain commands are called. 

Most of these counters have the same name as the commands that alter 

them: 

part chapter paragraph figure enumi 

section subparagraph table enumii 

subsection page footnote enumiii 

subsubsection equation mpfootnote enumiv 

The meanings of most of these counters are obvious from their names 

and need no further explanation. The counters enumi . . . enumiv refer to 

the four levels of the enumerate environment (Sections 4.3.4 and 4.3.5), 

while the counter mpfootnote controls the footnote numbering within 

the minipage environment (Section 4.10.4). 

The value of a counter is an integer number, usually non-negative. A 

command may output several numbers at once: the current \subsection 

command outputs 8.1.1, which addresses three counters in all. For example, 

the \subsection command increments the value of the subsection 

counter by one, and prints the values of the chapter, section, and 

subsection counters, separated by periods. At the same time, this command 

sets the subsubsection counter to zero. 

181

182 Chapter 8. User Customizations 

8.1.2 User-defined counters 

The user may create new counters with the command 

\newcounter{counter name}[in counter] 

where counter name is the name of the newly established counter. This 

may be any combination of letters that is not already the name of an 

existing counter. Thus none of the names of the L AT E X counters listed 

above may be used as counter name nor any name of a previously defined 

user counter. 

The optional argument in counter is the name of another counter 

that already exists (L AT E X or user defined) and has the effect that the newly 

defined counter is reset to zero whenever in counter is incremented by one 

of the commands \stepcounter or \refstepcounter (see below). For 

example, the subsection counter is reset to zero whenever the section 

counter is incremented. 

The \newcounter command may not appear in any file that is read in 

with the \include command (Section 9.1.2). It is therefore best to put all 

\newcounter commands into the preamble. 

8.1.3 Changing counter values 

Every counter, whether L AT E X or user defined, has an initial value of zero; 

this can be altered with the following commands: 

\setcounter{counter}{num} 

This command is self-explanatory: the specified counter is assigned 

the integer value num. 

\addtocounter{counter}{num} 

With this command, the value of the counter is increased by the 

integer num, which may be positive or negative. 

\stepcounter{counter} 

The value of the counter is increased by one, and at the same 

time all its sub-counters (that is, those that have this counter as 

their in counter) are reset to zero (see above). 

\refstepcounter{counter} 

This command has the same effect as \stepcounter but also 

makes counter the current counter for the cross-referencing command 

\label (see Section 9.2.1). 

This last command may be applied, for example, within the figure 

or table environments when the \caption command is missing and yet 

there is to be a reference in the text to the figure or table number using the 

\ref command. Then \refstepcounter{figure} or \refstepcounter 

{table} is given within that float environment to bring the corresponding

8.1. Counters 183 

counter to the right value and to allow that value to be assigned a keyword 

with the \label command (Section 9.2.1). 

The value of a counter may be treated as a number with the command 

\value{counter} 

which does not change the value at all. This command is mostly used 

in connection with \setcounter or \addtocounter. For example, if the 

user-defined counter mypage has been created, it may be set to the same 

value as the page counter page by giving \setcounter{mypage}{\value 

{page}}. 

8.1.4 Printing counter values 

The numerical value in a counter can be printed with the commands 

\arabic{counter} as an Arabic number, 

\Roman{counter} as a capital Roman numeral, 

\roman{counter} as a lower case Roman numeral, 

\alph{counter} as a lower case letter, 

\Alph{counter} as a capital letter, 

\fnsymbol{counter} as a footnote symbol. 

For the commands \alph and \Alph, the numbers 1. . . 26 correspond to 

the letters a. . . z and A. . . Z. It is up to the user to ensure that the counter 

value lies within this range. For \fnsymbol, the numbers 1. . . 9 are output 

as the symbols * † ‡ § ** †† ‡‡. Here again, the user must 

take care that the counter does not reach a value of 10 or more. 

For each counter, a command of the form 

\thecounter 

is also available, consisting of \the immediately followed by the name 

of that counter, such as \thepage. This type of command is initially 

identical to \arabic{counter}, but may redefined to be composed of 

several counter commands. In the document classes book and report, 

for example, the command \thesection is defined in terms of both the 

chapter and section numbers: \arabic{chapter}.\arabic{section}. 

Here is the result of printing \thesection at this point: 8.1. 

The automatic printing of counter values such as the page, equation, 

or sectioning numbers is accomplished by means of calls to the appropriate 

\thecounter commands. If a different format for some automatic 

numbering is desired, say alphabetical equation numbers, the definition 

of the corresponding \thecounter command can be altered using the 

methods described in Section 8.3. 

Exercise 8.1: Take your standard exercise.tex file and print out the final values 

of the L AT E X counters with \arabic{counter} commands at the end. Change some 

the values with the \setcounter and \addtocounter commands and print the 

values out once more.


8.2 Lengths 

It has been constantly pointed out in all the descriptions of the length 

parameters such as \parskip or \textwidth that new values may be 

assigned with the \setlength command. Some of these parameters 

expect rubber length values that may stretch or shrink. These are mainly 

parameters that produce vertical spacing. The types of length units, 

both fixed and rubber, are described in detail in Section 2.4. This will 

not be repeated here, but rather additional commands for assigning and 

handling lengths are discussed in this section. 

The standard L AT E X method for assigning a value to a length parameter 

is with the command 

\setlength{\length cmd}{length spec} 

where length spec may be a length specification (with units) or another 

length parameter. In the latter case, \length cmd takes on the current 

value of that other parameter. Thus with \setlength{\rightmargin} 

{\leftmargin} the right-hand margin in a list environment is set to 

the same value as that of the left-hand margin. 

Lengths may be increased with 

\addtolength{\length cmd}{length spec} 

which adds length spec to the value of the length parameter \length cmd. 

A negative value for length spec decreases \length cmd by that amount. 

Once again, another length parameter may be used for length spec, with or 

without a preceding minus sign, and its value will be added or subtracted. 

A decimal number just before a length parameter multiplies its value by 

that quantity: 0.5\textwidth means half the width of the text column 

and 2\parskip twice the inter-paragraph spacing. 

With the command 

\settowidth{\length cmd}{text} 

the length parameter \length cmd is set equal to the natural length of a 

piece of text. 

Similarly the commands 

\settoheight{\length cmd}{text} 

\settodepth{\length cmd}{text} 

set the \length cmd equal to the height and depth of the text above and 

below the baseline, respectively. 

Finally, the command 

\stretch{decimal num} 

yields a rubber length that is decimal num times as stretchable as \fill 

(Section 2.4.2). 

A user-defined length parameter is created with

! 

\newlength{\new len cmd} 

8.3. User-defined commands 185 

which establishes \new len cmd as a length with a value of 0pt. All the 

above commands may be used to manipulate its value further. 


\addvspace{length spec} 

inserts extra vertical spacing of the amount length spec at that point where it 

appears. If more than one such command is given, the total inserted spacing will 

be that of the largest argument and not the sum of them all. This command may 

only be given between paragraphs. Its application for user-defined commands 

and environments lies in the generation of structures that should behave like 

paragraphs. 

8.3 User-defined commands 

New commands may be defined or redefined under L AT E X with the commands 

\newcommand{\com name}[narg][opt]{def } 

\renewcommand{\com name}[narg][opt]{def } 

The first version is used to define a command \com name that does not 

yet exist. Its name may be any combination of letters that do not form 

the name of another command. The second version redefines an already 

existing command \com name. In both cases, an error message is printed 

if the incorrect variant is called. The first optional argument narg is a 

number between 1 and 9 specifying how many arguments the new or 

altered command is to have. A second optional argument opt gives the 

default value for an optional argument that the new command may take. 

The actual definition of the command is contained in the text def. 

8.3.1 Commands without arguments 

We will first illustrate the use of the \newcommand without the optional 

argument [narg]. This form is applied when a fixed combination of 

L AT E X or user commands is to be repeated frequently as a command with 

its own name. For example, the structure x1, . . . , xn, called an x-vector, 

often occurs in mathematical formulas and is formed in math mode with 

x_1,\ldots,x_n. Typing 

\newcommand{\xvec}{x_1,\ldots,x_n} 

creates a new command named \xvec that may be called and used just 

like any other command. When called, it inserts the sequence of text and 

commands, in this case x_1,\ldots,x_n, into the current text exactly as


if one had typed it oneself. In fact, this is precisely what happens: when 

\xvec is called, it is its definition that goes into the L AT E X processing. 

Since the new command \xvec contains a math command (the subscript 

command ), it may only be called within math mode. Thus $\xvec$ 

is required to produce x1, . . . , xn in text mode. It might therefore seem a 

good idea to include the switching to math mode in the definition itself, 

as 

\newcommand{\xvec}{$x_1,\ldots,x_n$} 

Now \xvec yields x1, . . . , xn. However, this command may only be applied 

in text and never in math mode. There is a trick to enable the command 

to be called in both modes: define it as 

\newcommand{\xvec}{\ensuremath{x_1,\ldots,x_n}} 

Now both \xvec and $\xvec$ are allowed, both with the same result. 

The above text mode example was actually written as \xvec{}, since 

TEX treats it as a command without arguments, terminating its name with 

the first non-letter that it finds. If this character is a blank, it only ends 

the command name and does not insert interword spacing (Section 2.1). 

Thus \xvec and ... produces ‘x1, . . . , xnand . . . ’ without any spacing 

between. This problem is solved by adding a space command \ or the 

empty structure {} after the command name, in this case as \xvec\ 

\xvec{}. 

or 

Package: 

It would also have been possible to include the blank in the definition of 

\xvec, as {\ensuremath{x_1,\ldots,x_n} }. Now the blank following 

the command name is still removed, but the command itself inserts one 

to make up for it. However, this is not recommended practice since this 

‘programmed’ blank will always be present, even when some punctuation 

or other symbol directly follows the command. 

A better solution is provided by the xspace package in the tools 

xspace collection (Section B.5.4). With this package, one adds the command 

\xspace at the end of a definition where a space may appear: 

\newcommand{\xvec}{\ensuremath{x_1,\ldots,x_n}\xspace} 

This command prints a blank unless it is followed by punctuation, in 

which case it does nothing. 

The different versions of the above example were all illustrated with 

the command \newcommand, although in fact this command may only be 

used once to initiate a user-defined command that does not already exist. 

Once \xvec has been created, a revised definition can only be given with 

the command \renewcommand. This was indeed done with the second 

and subsequent definitions of \xvec. 

Following this example, the user may employ \newcommand (or alternatively 

\renewcommand) to combine any set of commands and text to 

form a command under a new name that may then be called whenever


necessary. In this way, a considerable amount of typing can be avoided 

and the possibility of error reduced, especially for complex mathematical 

structures. 

If it is not known whether or not a command with the chosen name 

already exists, one may use instead 

\providecommand{\com name}[narg][opt]{def } 

which has the same syntax as \newcommand and \renewcommand. The 

difference is that if the command already exists, the new definition will 

be ignored. The opposite effect (overwriting the current definition of a 

command without knowing if it already exists) can be achieved by first 

calling \providecommand to ensure the command exists, and then issuing 

the true definition with \renewcommand. However, this should be done 

only with great care! 

Exercise 8.2: Define the commands \iint, \iiint, 

and \idotsint, to make the multiple integrals shown 

at the right as displayed formulas, or as the following 

text formulas: , , · · · 

 

 

· · · 

Exercise 8.3: Change the \thechapter, \thesection, and \thesubsection 

commands so that for the document classes book and report the chapter numbering 

is done with capital letters, such as B, the section numbering with capital 

Roman numerals after the chapter letter, in the form B–III, and the subsection 

numbering with lower case Roman numerals following a comma: B–III,v. 

Hint: the original versions of these commands in book and report are defined 

as 

\newcommand{\thechapter}{\arabic{chapter}} 

\newcommand{\thesection}{\thechapter.\arabic{section}} 

\newcommand{\thesubsection}{\thesection.\arabic{subsection}} 

Now apply \renewcommand to make the required changes. 

8.3.2 Commands with arguments 

In addition to the vector x1, . . . , xn, there are equivalent vectors y1, . . . , yn 

and z1, . . . , zn in mathematics. It would be possible to define commands 

\yvec and \zvec following the pattern for \xvec. However, it is also possible 

to define a generalized vector command and to specify the variable 

part as an argument. In the present example, the variable part is the letter 

x, y, or z. A command with one variable part is created with the optional 

argument [1]. For example, 

\newcommand{\avec}[1]{\ensuremath{#1_1,\ldots,#1_n}} 

defines the general vector command \avec{arg}. Calling \avec{x} yields 

x1, . . . , xn while invoking \avec{y} prints out y1, . . . , yn. The character


#1 in the command definition is a dummy argument representing the text 

of arg that replaces all occurrences of #1 when the command is called. 

By imagining an x or a y at each location of the #1 in the definition, the 

desired structures in each case can be recognized. 

The digit 1 in the dummy argument #1 seems at first to be rather 

pointless. In fact, for a command with only one argument it really has 

no meaning. However, its function becomes more obvious for commands 

with multiple arguments. Let us say, for example, that we want a command 

to generate structures such as u1, . . . , um as well as v1, . . . , vn. This 

requires two arguments, one to specify the letter u, v, etc., and a second 

to determine the last subscript n, m, and so on. Such a command is 

created with 

\newcommand{\anvec}[2]{\ensuremath{#1_1,\ldots,#1_#2}} 

and invoked as \anvec{u}{n} for u1, . . . , un and as \anvec{v}{m} for 

u1, . . . , um. The optional argument [2] for \newcommand says that the 

command being defined contains two arguments; in the definition part, 

#1 is replaced by the first argument and #2 by the second. By imagining 

u or v in place of #1 and n or m where #2 stands, one can see how the 

command \anvec{arg1}{arg2} operates. 

This pattern may be carried on for even more arguments. With 

\newcommand{\subvec}[3]{\ensuremath{#1_#2,\ldots,#1_#3}} 

a command \subvec is defined with three arguments. It should be clear 

from the definition that calling \subvec{a}{i}{j} produces ai, . . . , aj. 

A command argument that consists of only a single character need not be 

! put into curly brackets { } but may be given directly. If it is the first argument, 

it must be separated from the command name with a blank, as usual. Thus the 

sequence \subvec aik is the same as \subvec{a}{i}{k}, and \subvec x1n 

produces the same structure x1, . . . , xn as our first user-defined example \xvec. 

Arguments must be enclosed in curly brackets { } when they contain more 

than one character, since the brackets indicate that the contents are to be treated 

as a unit. Thus \subvec{A}{ij}{lk} prints out Aij, . . . , Alk. The three replacement 

arguments are A for #1, ij for #2, and lk for #3. 

Why does \subvec{A}{ij}{lk} produce Aij, . . . , Alk and not the expected 

result Aij, . . . , Alk? The answer is that, although the arguments within curly 

brackets are set as units into the definition text, the brackets themselves are 

not. The command text after replacement is \ensuremath{A_ij,\ldots,A_lk} 

so that only the first characters following the subscript symbols _ are actually 

lowered. In order that both letters be lowered, they must be seen as a unit within 

the command text, that is as A_{ij},. . . ,A_{lk}. This may be achieved with 

the \subvec command by including an extra set of brackets in the arguments: 

\subvec{A}{{ij}}{{lk}}. A better solution, however, is to put the brackets in 

the definition to begin with: 

\renewcommand{\subvec}[3]{\ensuremath{#1_{#2},\ldots,#1_{#3}}} 

which will always produce the desired result with only a single set of brackets 

per argument: \subvec{A}{ij}{lk} prints Aij, . . . , Alk.

8.3.3 Commands with an optional argument 


As we have seen, many L AT E X commands may take optional arguments, including 

! the command \newcommand itself. It is also possible to define user commands 

with one optional argument. The advantage of this is that, although an argument 

is provided, in most applications it will usually take some standard value that 

need not be given explicitly. 

As an example, the user-defined vector command \subvec in the last section 

has three arguments, for the letter and for the first and last subscripts. However, 

it may be that the letter is normally x, so it makes sense to include it as an 

optional argument, which is to be specified only for a different letter. This is 

accomplished with 

\renewcommand{\subvec}[3][x] 

{\ensuremath{#1_{#2},\ldots,#1_{#3}}} 

The difference between this and the previous definition is the addition of [x] 

after the [3] argument. This states that the first of the three arguments is to be 

optional, and its standard value is x. Now \subvec{i}{j} prints xi, . . . , xj while 

\subvec[a]{1}{n} produces a1, . . . , an. 

There may only be one optional argument in the user-defined command, and 

it will always be the first one, the #1 in the definition. 

8.3.4 Additional examples of user-defined commands 

In the above explanation of user-defined commands, a very simple case 

of a vector structure was taken as an example. We would now like to 

demonstrate some more complex situations, in which counters, lengths, 

and even some special T E X commands are applied. 

Example 1: In Section 5.4.6, the T E X commands \atop and \choose were 

presented as useful mathematical commands even for L AT E X applications. 

Unfortunately, the syntax of these commands deviates considerably from 

that of the similar L AT E X command \frac. However, 

\newcommand{\latop}[2]{{#1\atop#2}} and 

\newcommand{\lchoose}[2]{{#1\choose#2}} 

define two commands \latop and \lchoose that yield the same results 

with a syntax like that of L AT E X: \latop{upper}{lower}. 

Example 2: The command \defbox{sample text} is to set a box width 

equal to the length of the text sample text. A subsequent call to the 

command \textbox{text} centers text within a frame with the same 

width as sample text. 

\newlength{\wdth} 

\newcommand{\defbox}[1]{\settowidth{\wdth}{#1}} 

\newcommand{\textbox}[1]{\framebox[\wdth]{#1}}


First, a new length parameter \wdth is created, then \defbox is defined 

so that \wdth is set equal to the length of its argument (Section 8.2), and 

finally \textbox makes a framed box of that same width containing its 

argument, centered. (Do not name the length parameter \width, for this 

already exists, Section 4.7.5.) 

as wide as this text\\ 

\defbox{as wide as this text}\textbox{}\\ 

\textbox{text}\\ 

\textbox{longer text} 

as wide as this text 

text 

longer text 

Example 3: A footnote command \myftnote is to be created that behaves 

as the normal command \footnote{text} in putting text into a footnote, 

but instead of using numbers as the marker, it should take the symbols 

∗ † ‡ § ∗∗ †† ‡‡ one after the other, starting again with the symbol ∗ 

on each new page. First a new counter must be established that will be 

reset to zero every time the page counter is incremented. This is done 

with (see Section 8.1.2) 

\newcounter{myfn}[page] 

making a user-defined counter myfn that is set to zero every time the 

page counter is incremented. ∗ Next the command 

\renewcommand{\thefootnote}{\fnsymbol{footnote}} 

redefines the footnote marker to be that symbol in the sequence given 

by the counter footnote (Sections 4.10.2 and 8.1.4). Now the actual new 

footnote command can be constructed with 

\newcommand{\myftnote}[1]{\setcounter{footnote}{\value{myfn}}% 

\footnote{#1}\stepcounter{myfn}} 

yielding the desired results. The user-defined command \myftnote possesses 

one argument, which is passed to the L AT E X \footnote command 

after the L AT E X counter footnote has been set equal to the value of the 

user counter myfn. Once the command \footnote has been executed, 

the counter myfn is then incremented by one with \stepcounter{myfn}. 

This counter, however, is reset to zero whenever the page counter is 

incremented, that is, whenever a new page begins. 

The footnote on the previous page was generated with the command 

\myftnote as described. It is now used here † and again here ‡ , demonstrating 

how the symbols have been reset on a new page. 

Example 4: A command \alpheqn is to be set up so that once it has been 

called, the subsequent equations will all have the same number but be 

∗ Actually the page counter is not incremented exactly at the end of the page; LAT E X reads 

in the whole paragraph before it decides if and where a page break should occur. This can 

cause problems with resetting the myfn counter near the top of a page. 

† another footnote 

‡ and yet another footnote


followed by letters a, b, . . . , separated by a hyphen ‘-’ from the number. 

The command \reseteqn restores the numbering scheme to its original 

style. Thus a sequence of equation numbers could be 4, 5, 6-a, 6-b, 7. 

\newcounter{saveeqn} 

\newcommand{\alpheqn}{\setcounter{saveeqn}{\value{equation}}% 

\stepcounter{saveeqn}\setcounter{equation}{0}% 

\renewcommand{\theequation} 

{\mbox{\arabic{saveeqn}-\alph{equation}}}} 

\newcommand{\reseteqn}{\setcounter{equation}{\value{saveeqn}}% 

\renewcommand{\theequation}{\arabic{equation}}} 

The example should be easy to comprehend with the help of the commands 

and counters in Section 8.1. The current value of counter equation 

is saved in the counter saveeqn and then incremented, while equation 

itself is set to zero. The form of the equation marker, \theequation, is 

redefined using these two counters. The equation numbering routines will 

operate on equation as usual, leaving saveeqn unchanged. The resetting 

command \reseteqn puts the value of saveeqn back into equation and 

restores the definition of \theequation. 

This example is only appropriate for the document class article. For 

report and book, the definition of \theequation is 

\arabic{chapter}.\arabic{equation} 

The necessary modifications are left as an exercise for the user. 

The \mbox command in the first \renewcommand{\theequation} defining 

! the combined equation number is necessary because the result will be printed 

in math mode, where the hyphen ‘-’ is interpreted as a binary operator (minus 

sign) with extra spacing between it and its two ‘operands’, \arabic{saveeqn} 

and \alph{equation}. Thus 6 − a would be output instead of 6-a. The \mbox 

command causes a temporary switch out of math into text mode. 

Example 5: In Section 5.4.10, the international standards for mathematical 

typesetting are outlined, in which tensor and matrix variables should 

be set in a sans serif typeface, preferably italic or slanted. The math 

alphabet command \mathsf accomplishes this only for an upright font. 

The solution is to create a new math alphabet \mathsfsl for use in a 

\tensor command. 

\DeclareMathAlphabet{\mathsfsl}{OT1}{cmss}{m}{sl} 

\newcommand{\tensor}[1]{\mathsfsl{#1}} 

The NFSS system of font attributes is presented in Section A.1, and the 

\DeclareMathAlphabet command is explained in Section A.3.3. 

The Computer Modern sans serif fonts do not possess an italic form 

which is why the slanted shape is taken instead. If one were to select 

italic with {i} in place of {sl}, L AT E X would automatically substitute the 

slanted version.


In Section 5.4.10, several new commands are suggested to simplify 

typing math according to the ISO standard. These are: 

\newcommand{\me}{\mathrm{e}} 

\newcommand{\mi}{\mathrm{i}} 

\newcommand{\dif}{\mathrm{d}} 

\renewcommand{\vec}[1]{\boldsymbol{#1}} 

These permit the constants e, i, and differential operator d to be printed 

upright, and not italic. The redefinition of the \vec command requires 

the A M S math package amsbsy, either directly or as part of the amsmath 

package (Chapter 12). 

Here one could also include the abbreviation for the electron volt, 

mentioned on page 30, and its multiples: 

\newcommand{\eV}{\mbox{e\hspace{-.12em}V}} 

\newcommand{\keV}{\mbox{k\eV}} 

\newcommand{\MeV}{\mbox{M\eV}} 

If these commands are used frequently in many different documents, 

we recommend saving them in a separate file, say isomath.tex, to be 

input at the start of those documents, with \input{isomath}. 

Exercise 8.4: Define L AT E X commands \Lbrack and \Lbrace in the same manner 

as in Example 1 for \latop and \lchoose corresponding to the T E X commands 

\brack and \brace. These T E X commands behave as \choose from Section 5.4.6 

except that they enclose their contents in square brackets [\Lbrack] or curly 

braces {\Lbrace}. (Note that the names \lbrack and \lbrace are already 

defined and so should not be used for these commands.) 

Exercise 8.5: Generalize Example 4 with a command \vareqn{num}{type} to 

make the subsequent equation numbers have the value num followed by a running 

number in square brackets printed as \alph . . . \Roman, as given by the argument 

type. For example, 33[A], 33[B], would result from calling \vareqn{33}{\Alph}. 

Exercise 8.6: Generalize the integral commands in Exercise 

8.2 to include an argument to represent the area 

of integration set centered below the entire symbol. 

Thus \iint{(D)}, \iiint{V}, and \idotsint{G} 

should produce: 

 

(D) 

 

 

V 

 

· · · 

G 

Hint: the second command can be made simply with a subscript on the middle 

integral (but see \limits in Section 5.2.5). For the other two, negative horizontal 

shifting of the subscripted symbol is needed using \hspace{-..}. 

Exercise 8.7: The \tensor command defined in Example 5 above sets tensor 

variables in a slanted, sans serif font. If matrices are to be set instead in an 

upright, sans serif font, define a command \mtrx to produce this. 

Note: do not name this command \matrix because that already exists to produce 

matrices, not their variable names.

8.3.5 Conditional text 


Package: Practiced TEX users will be familiar with the conditional commands that 

ifthen are available, both for TEX and LAT EX. However, their usage is not always 

straightforward and often requires extensive knowledge of TEX’s deeper 

principles. Leslie Lamport has provided a package named ifthen, extended 

by David Carlisle, which not only simplifies their application, but 

also gives them a LAT EX syntax. 

The package is loaded as usual with the command 

\usepackage{ifthen} 

in the preamble. It then makes available the two commands \ifthenelse 

and \whiledo, which have the following syntaxes: 

\ifthenelse{test}{then text}{else text} 

\whiledo{test}{do text} 

In both cases, test is a logical statement (explained below); for the first 

command, then text or else text is inserted into the text depending on 

whether test is 〈true〉 or 〈false〉. For the second command, the do text is 

inserted (executed) as long as test evaluates to 〈true〉. (The do text must 

alter the inputs to test or it will never stop!) The texts may also contain 

commands, or even define or redefine commands. 

There are four types of basic logical statements which may be combined 

to form more complicated ones. 

Testing numbers 

To compare two numbers or commands that evaluate to numbers, simply 

put one of the relational operators between them, which stand 

for less than, equals, and greater than, respectively. The value of a 

counter may be tested by putting its name as the argument of the \value 

command. Examples: 

\newcommand{\three}{3} 

\ifthenelse {\three = 3} {O.K.} {What?} 

\ifthenelse {\value{page} < 100 } 

{Page xx} {Page xxx} 

The first case prints ‘O.K.’ since \three is equal to 3; in the second 

case Page xx is printed if the current page number is less than 100, else 

Page xxx. (The blanks above are added for clarity, since spaces between 

arguments are always ignored.) 

Whether a number is even or odd can be tested with \isodd 

\ifthenelse {\isodd{\value{page}} 

{odd} {even}


Testing text 

To test if two commands evaluate to the same piece of text, or if a 

command is defined as a certain string of text, use 

\equal{string1}{string2} 

where string1 and string2 are texts or commands that reduce to text. For 

example, with 

\ifthenelse {\equal{\name}{Fred}} {Frederick} {??} 

the text Frederick is inserted if \name has been defined to be Fred (with 

\newcommand), otherwise two question marks are printed. 

Testing lengths 

Another logical statement compares two lengths. 

\lengthtest{relation} 

where relation consists of two lengths or length commands separated by 

a relational operator . For example, 

\newlength{\horiz} \newlength{\vert} 

\newlength{\min} 

. . . . . . . 

\ifthenelse {\lengthtest{\horiz > \vert}} 

{\setlength{\min}{\vert}} {\setlength{\min}{\horiz} 

sets \min to be the smaller of \horiz and \vert. 

Testing switches 

A boolean switch is a parameter that is either 〈true〉 or 〈false〉, also called 

a flag. Three commands exist to handle them: 

\newboolean{string} creates a new switch 

\setboolean{string}{value} assigns a value true or false 

\boolean{string} tests its value 

The last of these is used as test in \ifthenelse and \whiledo. 

There are a number of internal L AT E X switches that may also be tested 

(but never reset!). The most useful of these are @twoside and @twocolumn 

for checking if two-side or two-column modes are active. Since they 

contain the character @, they may only be used inside a class or package 

file (Appendix D).

Combining logical statements 

8.4. User-defined environments 195 

Any of the above logical statements may be combined to form a more 

complex statement by means of the logical operators 

\and \or \not 

which should be straightforward to anyone familiar with boolean logic. 

For example, to set \textwidth to 10 cm if two-column mode is active or 

if \paperwidth is greater than 15 cm and the page counter is below 100, 

\ifthenelse {\lengthtest{\textwidth > 10cm} \or 

$ \lengthtest{\paperwidth > 15cm} \and 

\value{page} < 100 $ } 

{\setlength{\textwidth}{10cm}} {} 

will accomplish this. 

Such conditionals are fairly complicated, and they are most appropriate 

for defining new commands that may have alternative actions, or for 

including in package and class files (Appendix D). 

A relatively simple example: suppose one is uncertain whether British 

or American spelling is wanted by the publisher. One can write the 

document with both included in the text, with a switch in the preamble 

to make the final selection. 

\newboolean{US} 

\setboolean{US}{true} %For American spelling 

%\setboolean{US}{false} %For British spelling 

\newcommand{\USUK}[2]{\ifthenelse{\boolean{US}}{#1}{#2}} 

Now \USUK is a command that prints its first or second argument according 

to the setting of the flag US. Thus in the text one may write 

... the \USUK{color}{colour} of music ... 

which will yield American spelling if \setboolean{US}{true} has been 

specified, otherwise the British version. In fact, different sections of the 

work could have different spelling simply by changing the value of the 

switch at the appropriate point. 

As an example of \whiledo: one wishes to write some text n times, 

where both the text and n are to be variable. 

\newcounter{mycount} 

\newcommand{\replicate}[2]{\setcounter{mycount}{#1} 

\whiledo{\value{mycount}>0}{#2\addtocounter{mycount}{-1}}} 

Now \replicate{30}{?} will print 30 question marks. 

8.4 User-defined environments 

Environments may be created or changed with the commands


\newenvironment{env name}[narg][opt]{beg def }{end def } 

\renewenvironment{env name}[narg][opt]{beg def }{end def } 

where the arguments have the following meanings: 

env name: the name of the environment; for \newenvironment, it may 

not be the same as any existing environment or command name, 

whether L AT E X or user-defined. For \renewenvironment, on the other 

hand, there must already be an environment bearing this name. Any 

changes to L AT E X environments should only be undertaken if the user 

knows what he or she is doing. 

narg: a number between 1 and 9 that states how many arguments the 

environment is to have; if the optional argument narg is omitted, 

the environment is to have no arguments. 

opt: the default text for the first argument (#1) if it is to be optional; this 

behaves the same as for \(re)newcommand (page 185). 

beg def: the initial text to be inserted when \begin{env name} is called; 

if this text contains entries of the form #n, with n = 1, . . .,narg, then 

when the environment is started with the call 

\begin{env name}{arg 1}...{arg n}... 

each occurrence of #n within beg def is replaced by the text of the 

argument arg n. 

end def: the final text that is inserted when \end{env name} is called; 

here the dummy arguments #n are not allowed since they are only 

to appear in the beg def text. 

8.4.1 Environments without arguments 

Just as for user-defined commands, environments without the optional 

argument narg will be illustrated first. A user-defined environment named 

sitquote is created with 

\newenvironment{sitquote}{\begin{quote}\small 

\itshape}{\end{quote}} 

which sets the text appearing between \begin{sitquote} text \end 

{sitquote} in the typeface \small\itshape, and indented on both 

sides from the main margins, as demonstrated here. 

In this case, beg def consists of the command sequence \begin{quote} 

\small\itshape while end def is simply \end{quote}. Now the call 

\begin{sitquote} text \end{sitquote} is the same as 

\begin{quote}\small\itshape text \end{quote}


which produces the desired result. 

This example does not appear to be very practical since the same 

effect can be achieved with less typing by adding \small\itshape at 

the beginning of the quote environment. However, it is more consistent 

with the concept of logical markup introduced in Section 1.2.2. This 

special quotation environment may be used throughout the document 

without worrying about its typographical details, which are specified in 

its definition located in the document preamble. Doing it this way not 

only ensures consistency, it also simplifies any substitution that may be 

demanded later by a typographical expert. 

Let us expand the previous example somewhat as follows: 

\newcounter{com} 

\newenvironment{comment} 

{\noindent\slshape Comment:\begin{quote}\small\itshape} 

{\stepcounter{com}\hfill(\arabic{com})\end{quote}} 

where now beg def contains the text and commands 

\noindent\slshape Comment:\begin{quote}\small\itshape 

and end def is 

\stepcounter{com}\hfill(\arabic{com})\end{quote} 

where com is a user counter created by the \newcounter command. Now 

since the command \begin{comment} inserts the text beg def at the start 

of the environment and \end{comment} the text end def at the finish, it 

should be clear that 

\begin{comment} This is a comment. 

Comments should ... 

... in round parentheses. 

\end{comment} 

will generate the following type of structure: 

Comment: 

This is a comment. Comments should be preceded by the word Comment: 

the text being in a small, italic typeface, indented on both sides 

from the main margins. Each comment receives a running comment 

number at the lower right in round parentheses. (1) 

The user should examine the sequence of commands with the replacement 

text of this example to see precisely what the effect of this 

environment is. Two weaknesses should become apparent: what would 

happen if \begin{comment} were called in the middle of a line of text 

without a blank line before, and what would happen if the last line of the 

comment text were so long that there was no more room for the running 

comment number on the same line? 

The following revision removes these two problems:


\renewenvironment{comment} 

{\begin{sloppypar}\noindent\slshape Comment: 

\begin{quote}\small\itshape} 

{\stepcounter{com}\hspace*{\fill}(\arabic{com})\end{quote} 

\end{sloppypar}} 

With the \begin{sloppypar} command, the call to the environment 

always starts a new paragraph in which no overfull line can occur upon 

line breaking. If the comment number does not fit on the last line of 

text, a new line begins with the number right justified because of the 

command \hspace*{\fill}. Again the user should examine carefully 

just what is inserted into the processing between the \begin{comment} 

and the \end{comment}. 

8.4.2 Environments with arguments 

Passing arguments over to an environment is carried out exactly as for 

commands. As an example, the comment environment will be modified so 

that the name of the person making the comment is added after the word 

Comment:, and this name will be an argument when the environment is 

invoked. 

\renewenvironment{comment}[1] 

{\begin{sloppypar}\noindent\slshape Comment: #1 


{\stepcounter{com}\hspace*{\fill}(\arabic{com})% 

\end{quote}\end{sloppypar}} 

The text 

\begin{comment}{Helmut Kopka} This is a modified ... 

... environment argument \end{comment} 

now produces 

Comment: Helmut Kopka 

This is a modified comment. Comments should be preceded by the 

word Comment:, followed by the name of the commenter, with the text 

of the comment being in a small, italic typeface, indented on both sides 

from the main margins. Each comment receives a running comment 

number at the lower right in round parentheses. The name of the 

commenter is transferred as an environment argument. (2) 

This example will now be modified once again by interchanging the 

comment number and the name of the commenter. Placing the running 

number after the word Comment: is no problem, for it is simply necessary 

to insert those commands from {end def } at the location of the #1 dummy 

argument. However, putting the symbol #1 where the comment number 

used to be will produce an error message during the L AT E X processing since


this violates the syntax of the \newenvironment command: ‘No dummy 

arguments shall appear within the {end def}.’ If the dummy argument 

comes after \begin{quote}, the name will be printed at the wrong place, 

at the beginning of the comment text. 

There is a trick to solve this problem: 

\newsavebox{\comname} 

\renewenvironment{comment}[1] 

{\begin{sloppypar}\noindent\stepcounter{com}\slshape 

Comment \arabic{com}\sbox{\comname}{#1} 


{\hspace*{\fill}\usebox{\comname}\end{quote}\end{sloppypar}} 

The commands \newsavebox, \sbox, and \usebox are all described in 

Section 4.7.1. Here \comname is the name of a box that has been created 

with \newsavebox, in which the first argument, the commenter’s name, 

is stored. With this new definition, a comment now appears as follows: 

Comment 3 

In this form, every comment is assigned a sequential number after 

the word Comment. The comment text appears as before, while the 

name of the commenter is entered as the environment argument and 

is placed at the right of the last line. Helmut Kopka 

The implementation of more than one argument for environments is 

the same as for commands and needs no further explanation. 

8.4.3 Environments with an optional argument 

Similarly, environments may be defined with an optional argument, just 

as commands. To take our last example once more, if we feel that most 

comments will be made by Helmut Kopka, we could alter the first line of 

the definition to 

\renewenvironment{comment}[1][Helmut Kopka]{..}{..} 

Now it is only necessary to specify the name of the commenter if it is 

someone else. With 

we obtain 

\begin{comment}[Patrick W. Daly]More than ... 

... appropriate.\end{comment} 

Comment 4 

More than one person may want to make a comment, but perhaps 

one person makes more than others do. An optional argument for the 

name is then appropriate. Patrick W. Daly


Exercise 8.8: Extend the definition of the comment environment so that a page 

break cannot occur either between the heading ‘Comment n’ and the comment 

text or between the comment text and the commenter’s name. 

Exercise 8.9: Create a new environment making use of the minipage environment, 

to be named varpage and possessing one argument that is a sample text 

to determine the width of the minipage. The call 

\begin{varbox}{‘As wide as this sample text’} 

. . . . . . . . . . . . . . . . . . .\end{varbox} 

should pack the enclosed text into a minipage that has a width equal to that of 

the text ‘As wide as this sample text’. 

Hint: a user-defined length parameter must first be established, say \varwidth. 

See Section 8.2 for details on assigning lengths to text widths. 

Exercise 8.10: Generate an environment named varlist with two arguments 

that behaves as a generalization of the sample list in Section 4.4.3. The first 

argument is to be the item word that is printed on each call to \item; the second 

is the numbering style of the item numeration. For example, with the call 

\begin{varlist}{Sample}{\Alph} . . . . . \end{varlist} 

every \item command within the environment should produce the sequence 

‘Sample A’, ‘Sample B’, . . . . The indentation should be 1 cm larger than the width 

of the item word, which itself should be left justified within the label box. 

Hint: once again a user-defined length, say \itemwidth, is necessary for this 

solution. After the length of the item word has been stored with \settowidth, 

the indentation may be set with 

\setlength{\leftmargin}{\itemwidth} 

to be equal to the width of the item word, and then with 

\addtolength{\leftmargin}{1cm} 

to be 1 cm larger. Length assignments for \labelwidth and \labelsep may be 

similarly set to appropriate values. 

All further details may be obtained from Section 4.4. 

8.5 Some comments on user-defined structures 

8.5.1 Reusing sets of definitions 

User definitions are created to alter page formats, to reorganize the document 

layout, to invent new structures, and to define shortcuts. That 

is, they either directly specify typographical elements like the text width, 

or indicate how new or existing logical elements are to be rendered typographically. 

As such, they belong in the preamble, where they can be 

conveniently redefined as needed, effecting the entire document. 

Once a set of user definitions has been established, what is the best 

way to reuse them? There are several possibilities:

8.5. Some comments on user-defined structures 201 

1. Write a new class file based on an existing one, incorporating the 

your own changes. This should be reserved for major changes to the 

layout or document organization, which will be used very frequently. 

2. Write your definitions to a package file with the extension.sty to be 

loaded with the \usepackage command. You can take advantage of 

special package features, like internal commands and options. 

3. Write the definitions to a file with the extension.tex and load it with 

\input{filename}. No extra features are possible, and the file can 

be loaded anywhere within the source file. 

4. Just copy the definitions from one source file to the next. 

Methods 1 and 2 are more sophisticated, and should be considered for 

user definitions that have a wide generality. They are explained in Appendix 

D. Method 3 is perfectly good for a personal collection of shortcut 

commands, or common (re)definitions that are applied to many documents. 

By having one depository for all such customizations, it is easier 

to maintain them centrally. 

Finally method 4 is applicable for a short list of uncomplicated definitions. 

8.5.2 Unwanted spaces 

Occasionally user-defined structures generate spacing where none was 

expected, or more spacing than was desired. This is almost always due 

to blanks or new lines in the definition, included only to improve the 

legibility of the input text but interpreted as spacing when the structure 

is invoked. 

For example, if the % character had been left out of the first line of the 

\myftnote definition on page 190, a new line would have been added to 

the command text at that point, and converted into a blank. This blank 

would be inserted between the previous word which should receive the 

footnote marker and the call to the \footnote command that actually 

generates the marker, with the result that it would be displaced from that 

word (for example, wrong ∗ instead of right ∗ ). 

At this point we should like to point out that many L AT E X commands are 

! invisible, in that they do not produce any text at the point where they are called. 

If such an invisible command is given separated by blanks from the surrounding 

text, it is possible that two blanks will appear. 

For example \rule{0pt}{0pt} produces 

‘For example produces’ twice the interword spacing between ‘example’ and 

‘produces’. Invisible commands without arguments do not present this problem 

since the trailing blank acts solely as a command name terminator and disappears. 

Furthermore, the following L AT E X commands and environments always remove the 

subsequent blanks even when arguments are present.


\pagebreak \linebreak \label \glossary \vspace figure 

\nopagebreak \nolinebreak \index \marginpar table 

8.5.3 Identical command and counter names 

In the previous examples a number of counters were introduced for 

application to specific commands or environments, such as myfn for the 

command \myfootnote on page 190, or com for the environment comment 

on page 197. In these cases, the counter and the corresponding command 

or environment were given different names. This was not necessary: 

counters may have a name that is identical to that of a command or 

environment. L AT E X knows from the context whether the name refers to a 

counter or to a command/environment. 

The different names were chosen in the examples in order to avoid 

confusion for the beginner. In fact, it is reasonable to give a counter the 

same name as the command or environment with which it is coupled, a 

practice that L AT E X itself constantly employs (Section 8.1.1). In the abovementioned 

examples, the counters could just as easily have been named 

myfootnote and comment to emphasize their interdependence with the 

command and environment of the same names. 

8.5.4 Scope of user definitions 

User structures that are defined within the preamble are valid for the 

entire document. Command and environment definitions that are made 

inside an environment remain in effect only until its end. Even their 

names are unknown to L AT E X outside the environment in which they were 

defined. Thus if they are to be defined once again in another environment, 

it is \newcommand or \newenvironment that must be used and not the 

\renew versions. 

For command and environment names that have been globally defined, 

that is, in the preamble, all further new definitions of these names must 

be made with the \renew version. However, these subsequent definitions 

will apply only locally within the environment in which they are declared; 

outside, the previous global definitions will be valid once more. 

The same applies to structure definitions within nested environments. 

A definition in the outer environment is effective within all inner ones, but 

a new definition must be made at the deeper levels with \renewcommand 

or \renewenvironment. On leaving the inner environment, the new 

definition will no longer be operative, but rather the old one will be 

re-established. 

Warning: Structures that have been created with \newsavebox or with 

\newcounter are globally defined. If they are given within an environment, 

their definitions remain effective even outside that environment. 

Similarly \setcounter functions globally, although \savebox does not.

8.5. Some comments on user-defined structures 203 

Local definitions are really not to be recommended. It is far better to 

place all user definitions at the beginning, in the preamble, where they 

are most easily visible. A local redefinition of a standard command can 

lead to puzzling effects that are hard for others, or even for the author, 

to trace down. Of course, some special effects might just be wanted. 

8.5.5 Order of definitions 

User-defined structures may be nested within one another. If one user 

definition contains another user-defined structure, it is often the case that 

the inner one has already been defined. However, this is not necessary. 

User definitions may contain other user structures that are defined afterwards. 

What is important is that the other structure is defined before the 

first command is invoked. 

For example, a normal sequence of definitions would be 

\newcommand{\A}{defa} 

\newcommand{\B}{defb} 

\newcommand{\C}{\A \B} 

where \C is defined in terms of \A and \B. However, it is also permissible 

to write 

\newcommand{\C}{\A \B} 

normal text, but without calling \C 

\newcommand{\B}{defb} 

\newcommand{\A}{defa} 

further text with any number of calls to \A, \B and \C 

8.5.6 Nested definitions 

User definitions may be nested inside one another. A structure such as 

! \newcommand{\outer}{{\newcommand{\inner}...}} 

is permissible. The command that is defined with the name {\inner} is valid 

and known only within the command {\outer}, according to the remarks on the 

scope of definitions on the facing page. Although the T E X macros make copious 

use of nested command definitions to limit the lifetime of temporary commands, 

it is not recommended to nest L AT E X definitions excessively since it is too easy 

to lose track of the bracketing. A forgotten bracket pair will produce an error 

message on the second call to the outer command, since the inner definition still 

exists as a leftover from the first call. Nevertheless, here is an example: 

\newcommand{\twentylove} 

{{\newcommand{\fivelove} 

{{{\newcommand{\onelove} 

{I love \LaTeX!}% 

\onelove\ \onelove\ \onelove\ \onelove\ \onelove}}} 

\fivelove\\ \fivelove\\ \fivelove\\ \fivelove}}


The entry My opinion of \LaTeX:\\ \twentylove now produces: 

My opinion of L AT E X: 

I love L AT E X! I love L AT E X! I love L AT E X! I love L AT E X! I love L AT E X! 




Indenting the lines of the definition, as shown above, can help to keep track of 

the nesting levels; each line of the same level starts in the same column, ignoring 

braces. 

If both the inner and outer definitions are to be provided with arguments, 

! the symbols for the inner and outer dummy arguments must be distinguished. 

The symbols for the inner definition are ##1 ... ##9, while those for the outer 

one are the normal #1 ... #9. For example: 

\newcommand{\thing}[1]{{\newcommand{\color}[2]{The ##1 is ##2.} 

\color{#1}{red} \color{#1}{green} \color{#1}{blue}}} 

The entry The colors of the objects are\\ 

\thing{dress}\\ \thing{book}\\ \thing{car} produces 

The colors of the objects are 

The dress is red. The dress is green. The dress is blue. 

The book is red. The book is green. The book is blue. 

The car is red. The car is green. The car is blue. 

The separate definition and calling as in Section 8.5.5 is easier to follow and 

would be: 

\newcommand{\thing[1]}{\color{#1}{red} \color{#1}{green} 

\color{#1}{blue} 

\newcommand{\color}[2]{The #1 is #2.} 

If one were ever to go to a third level of definitions, the dummy arguments 

! are given as ####1 . . . ####9. And for the fourth level one writes 8 # signs!

Part II 

Beyond the Basics

9 

Document 

Management 

This chapter describes those features of L AT E X that justify the subtitle 

of Leslie Lamport’s original books, ‘A Document Preparation System’. 

Whereas in the previous chapters we have concentrated more on markup, 

logical and typographical, we now present topics that are essential for 

producing large, complex documents in an efficient manner. The subjects 

included here are the splitting of a document into several files, 

selective processing of parts of a document, cross-references to sections, 

figures, and equations, automated production of bibliographies, indices, 

and glossaries. 

9.1 Processing parts of a document 

As often pointed out, a L AT E X document consists of a preamble and the 

actual text part. Short documents, such as those that a beginner might 

have, are written to a single file by means of a text editor, and might be 

corrected after the first trial printing. As the user gathers experience and 

confidence, the L AT E X documents will rapidly expand in length until one is 

faced with the task of producing an entire book of more than a hundred 

pages. 

Such long documents could theoretically be kept in one file, although 

that would make the whole operation increasingly clumsy. The file editor 

functions less efficiently on longer files and the L AT E X processing takes 

correspondingly more time. A better idea is to split the work into several 

files which L AT E X then merges during the processing. 

9.1.1 The \input command 

The contents of another file may be read into a L AT E X document with the 


\input{filename} 

207

208 Chapter 9. Document Management 

where the name of the other file is filename.tex. It is only necessary to 

specify the full name of the file if the extension is something other than 

.tex. During the L AT E X processing, the text contained in this second file 

is read in at that location in the first file where the command is given. 

The result of the \input command is the same as if the contents 

of the file filename.tex had been typed into the document file at that 

position. The command may be given anywhere in the document, either 

in the preamble or within the text part. 

Since the \input command may be given in the preamble, it is possible 

to put the whole preamble text itself into a separate file. The actual L AT E X 

processing file could even be reduced to simply \begin{document} ... 

\end{document} with a number of \input commands. A preamble 

file makes sense if one has a series of documents all of the same type 

requiring a common preamble. This also simplifies any later change to 

the specifications that must be made in all the documents. Different 

preamble files may be prepared for various types of processing and may 

then be selected with \input{proc type}. 

A file that is read in by means of the \input command may also 

contain further \input commands. The nesting depth is limited only by 

the capacities of the computer. 

In order to obtain a listing of all extra files read in, give the command 

\listfiles 

in the preamble. The list appears both on the computer monitor and in 

the transcript file, at the end of the processing run. The version numbers 

and any other loading information are also printed. This provides a check 

as to which files have been input. See Section D.2.9 for a demonstration. 

Exercise 9.1: Put the preamble of your standard exercise file exercise.tex into 

a separate file preamble.tex. Split the text part into three files exer1.tex, 

exer2.tex, and exer3.tex. What should the main file now contain to ensure 

that L AT E X processes the whole exercise text? 

9.1.2 The \include command 

Splitting the document into several files may be practical for writing and 

editing, but when the files are merged with the \input commands, it is 

still the entire document that is processed. Even if only one file contains 

a small correction, all files will be read in and processed once again. It 

would therefore be desirable to be able to reprocess only the one corrected 

file. 

One rough-and-ready method is to write a temporary main file containing 

only the preamble (which may be read in) and an \input command to 

read in that specific file. The disadvantage is that all automatic numbering 

of page numbers, sections, figures, equations, etc. will start from 1, since

9.1. Processing parts of a document 209 

all the information from the previous files will be missing. Furthermore, 

all cross-references from other files will be absent. 

A much better method is to employ the L AT E X command 

\include{filename} 

which is only allowed within the text part of the document, together with 

\includeonly{file list} 

in the preamble, containing a list file list of those files that are to be read 

in. The file names are separated by commas and the extension.tex is left 

off. 

If filename is in the file list, or if \includeonly is missing from the 

preamble, the command \include{filename} is identical to 

\clearpage \input{filename} \clearpage 

However, if filename is not contained within the file list, \include is 

equivalent to \clearpage and the file contents are not read in. However, 

an auxiliary .aux file is read in that sets all the counters and crossreferencing 

information for that file. 

Since \include always begins a new page, the document must be split 

into files at those points where a new page occurs, such as between chapters. 

Furthermore, \include commands may not be nested: they may 

only appear in the main processing file. However, an \input command 

may be given within a file that is \included. 

The great advantage of the \include command is that the additional 

information about page, section, and equation numbers will be supplied 

by the .aux files that are read in in place of the .tex files so that the 

selective processing takes place with the correct values of these counters. 

Cross-reference information from the other files is also available so that 

the \ref and \pageref commands (Section 9.2.1) yield the correct results. 

All these values will have been determined during a previous processing 

of the entire document. 

If the changes in the file that is being selectively processed lead to an 

increase or reduction in the number of pages, the following files will also 

have to be reprocessed to correct their page numbers. The same is true if 

sections are added or removed, or if the number of equations, footnotes, 

figures, etc. is altered. 

For example, suppose file 3 ends on page 17, but after the selective 

processing it now extends to page 22. The following file 4 still begins on 

page 18, and all further files also have their original starting page numbers. 

If file 4 is now selectively processed, it will receive the correct number for 

its first page, 23, based on the stored information in the revised file 3. So 

far so good. However, if instead file 6 were to be selectively processed 

right after file 3, it would receive its starting page number from file 5, 

which has not yet been corrected and would be in error by 5. The same


applies to all other structure counters. Their correct values can only be 

guaranteed when the files have been reprocessed in their proper order. 

In spite of these restrictions, the \include command is extremely useful 

for large documents, saving considerable computation time. Longer 

documents are normally written and edited in many stages. The \include 

command permits one to reprocess selective alterations in a short time, 

even if the numbering systems temporarily go awry. This can be repaired 

later on with a complete reprocessing of the entire document, by 

deactivating the \includeonly command in the preamble. 

A file that is read in with \include may not contain any \newcounter 

declarations. This is not much of a restriction, since they should normally 

be given in the preamble. 

For example, each chapter of a book might be written to a separate 

file with names chap1.tex, chap2.tex, . . . . The processing file itself 

contains the text 

\documentclass{book} 

. . . . . . . . . . . . . . . . 

\includeonly{...} 


\frontmatter 

\include{toc} 

\mainmatter 

\include{chap1} 

\include{chap2} 

. . . 

\backmatter 

\include{back} 


where the file toc.tex contains only commands for the title page, table 

of contents, and other listings: 

\begin{titlepage} ... \end{titlepage} 

\tableofcontents \listoftables \listoffigures 

and back.tex those for bibliography and index: 

\bibliography{..} 

\printindex 

The chapters to be processed are selected by adding them to the argument 

of \includeonly, not by changing the \include commands, which must 

always be present to ensure that all the .aux are input every time. For 

example, by giving \includeonly{toc,chap3} one processes the table 

of contents and chapter 3.

9.1.3 Monitor input and output 

9.1. Processing parts of a document 211 

There are times when it is desirable that L AT E X write a message to the 

computer monitor during processing. This can be achieved with the 


\typeout{message} 

where message stands for the text that is to appear on the monitor. This 

text is printed when the L AT E X processing reaches this command. It is also 

written to the transcript file, but does not appear in the processed output. 

If message contains a user-defined command, it will be interpreted 

and its translation appears on the monitor. The same thing applies to 

L AT E X commands. This could have dire consequences if the commands, 

either user or L AT E X, are not really printable. To print the command name 

literally, precede it with the command \protect. 


\typein[\com name]{message} 

also writes the message text to the monitor, but then it waits for the 

user to enter a line of text from the keyboard, terminated by typing the 

〈return〉 key. If the optional argument \com name is missing, the line of 

text is inserted directly into the processing. In this way, one could, for 

example, reuse the same text for a letter for several addressees, entering 

the name each time from the keyboard. Suppose the text contains 

Dear \typein{Name:}\\ ... 

then what appears on the monitor is: 

✎ 

Name: 

☞ At this point, one enters the name of the 

recipient. If on successive processings 

‘George’, ‘Fred’, and ‘Mary’ are entered, the 

result will be a set of identical letters dif- 

\@typein= 

fering only in their salutations as ‘Dear 

✍ 

✌ George’, ‘Dear Fred’, and ‘Dear Mary’. 

If the \typein command contains the optional argument \com name, 

this is treated the same as 

\typeout{message} \newcommand{\com name}{entered definition} 

In this case the definition is stored under the command name \com name 

interactively and may be invoked and executed in the rest of the document 

like any other L AT E X command. 

With some experience in L AT E X methodology, it soon becomes obvious 

that interactive processing with the \typein command can be very 

practical. For example, if the preamble contains 

\typein[\files]{Which files?} 

\includeonly{\files}


the following appears on the monitor: 

✎ 

Which files? 

☞ 

Now L 

\files= 

✍ 

✌ 

ATE X waits for the user to type in the 

names of one or more files (separated by 

commas) to be processed. This avoids having 

to modify the main processing file with 

an editor every time. 

A similar procedure may be employed when a form letter is to be sent 

to various recipients. One can enter the name and address and even the 

salutation interactively. Complete forms may be processed by LAT EX in this 

way, with the entries being made from the keyboard. 

Warning: The \typein command may not be used as the argument of 

another L AT E X command! It may, however, be given in environments such 

as minipage. 

Exercise 9.2: Change the main file from Exercise 9.1 so that the files exer1.tex, 

exer2.tex, and exer3.tex may be read in with the \include command. Arrange 

that you can determine interactively which of the files is to be processed. 

Exercise 9.3: Generate output of the form 

Certificate 

Olympic Spring Games 

Walterville 1992 

Finger Wrestling 

Gold A. T. Glitter AUR 7999.9 Points 

Silver S. Lining ARG 7777.7 Points 

Bronze H. D. Tarnish CUP 7250.0 Points 

so that the following enquiries appear on the monitor one after the other 

Message Command= Input 

Sport: \@typein = Finger Wrestling 

Unit: \unit = Points 

Gold: \@typein = A. T. Glitter 

Country: \@typein = AUR 

Value: \@typein = 7999.9 

Silver: \@typein = S. Lining 

. . . . . . = . . . 

and the appropriate entries are made interactively. The third column contains 

the answers necessary to produce the above sample output. Repeat the program 

with different entries. Let your imagination run wild.

9.2 In-text references 

9.2. In-text references 213 

In a longer text, one often wants to refer to other chapters, sections, 

tables, and figures, or to pages where some other description has been 

given. One also needs an index of certain keywords that occur throughout 

the document. In the days before electronic text processing, such crossreferences 

and indices meant an enormous amount of extra work for 

the author or secretary. Today the computer can alleviate much of this 

burden. 

In the old days, referring to earlier parts of the text by page number 

was tedious but feasible. Referencing future parts that had not yet been 

written had to be limited to section numbers since the page numbers were 

not yet known, or space had to be left for the page number to be entered 

at a later date. 

The production of a book is usually a progressive and constantly 

shifting process. The manuscript may not necessarily be written in the 

order in which it is to appear, and even once the initial draft version is 

finished, there will be major changes due to new considerations by the 

author or to knowledgeable advice from reviewers. Revisions, deletions, 

and insertions of entire sections or even chapters are commonplace, not 

to mention possible reordering of parts of the text. 

L AT E X relegates all the problems associated with such major alterations 

to the past. No matter what changes the author makes, the information 

necessary for the cross-references and keyword index is stored for use at 

any point in the text. 

9.2.1 Cross-references 

As already pointed out in several places, the command 

\label{marker} 

is used to store the current value of the relevant counter (section, 

equation, etc.) at that point in the text, which may then be referred 

to at other places. The value is assigned to the key marker, which may be 

any combination of letters, numbers, and characters, even those that are 

normally special symbols. 

The page number on which the \label command was issued can be 

printed with 

\pageref{marker} 

at any point earlier or later in the document. 

If the \label command is given after a sectioning command, or within 

an equation, eqnarray, or enumerate environment, or inside the argument 

of a \caption within the figure or table environments, the 

command


\ref{marker} 

prints the number of that section, equation, figure, table, or enumeration 

that was current when marker was defined, and in the correct format. 

For enumerate, the same number is printed as is generated by the \item 

command where the \label appeared. Reference may also be made 

to theorem-like structures created by a \newtheorem command if the 

\label is given within the text of that theorem command. For example, 

a \label with marker text bo-wei was issued in the text of the Bolzano– 

Weierstrass theorem on page 80: 

\begin{theorem}[Balzano--Weierstrass] 

\label{bo-wei}...\end{theorem} 

so that the input text 

Theorem˜\ref{bo-wei} on page˜\pageref{bo-wei} 

generates the output ‘Theorem 1 on page 80’. Similarly, the input text 

for Table˜\ref{budget04} on page˜\pageref{budget04}, 

see also Section˜\ref{sec:figref} 

produces ‘for Table 7.1 on page 174, see also Section 7.6’, since that 

section contains the marker \label{sec:figref}. 

Note that the counter whose value is stored by the \label command 

depends on the context in which the command is given. This is normally 

fairly self-evident. Within regular text, it will be the counter and its value 

for the last issued sectioning command, which might not be immediately 

visible. Therefore, when reference to a section number is desired, it is 

best to place the \label command immediately after the appropriate 

sectioning command, or even within its argument, as part of the title text. 

This is in fact the recommended method. Put the \label command within 

the body of the text only when you want to refer to the page number at 

that point. 

A list of all label markers, their translations, and page numbers may 

be produced by processing the file lablst.tex, provided with the L AT E X 

installation. 

9.2.2 How cross-referencing works 

It is useful to know how the cross-referencing information is managed. What the 

! \label command actually does is to write the key name together with the current 

value of the appropriate counter(s) and the current page number to an auxiliary 

file with the root name of the document file plus the extension.aux. The \ref and 

\pageref commands obtain their information from this.aux file, which is read 

in at the beginning of the next processing by the \begin{document} command. 

The same situation as for the table of contents (Section 3.4.2) also applies here: 

on the first run, the.aux file does not exist so that no cross-reference information

9.2. In-text references 215 

can be output; instead the information is gathered and written to a new.aux file 

at the end of the run. A warning message is printed at the end if the auxiliary 

information has changed and a new run is necessary. 

If selective processing is being done with \includeonly and a set of \include 

commands, then an.aux file exists for each included files, as well as for the main 

source file. These.aux files are read in on every processing run, whether or not 

the corresponding .tex file is input. In this way, counter and cross-referencing 

information for the entire document is maintained even when only parts of it are 

processed. 

9.2.3 Referencing parts of other documents 

It is possible to refer to the marker keys in another LAT Package: 

EX document by 

xr means of David Carlisle’s xr package, which is part of the tools collection 

(Section B.5.4). With 

\externaldocument{filename} 

all the keys in the specified file become available in the current document, 

to be printed with \ref{key} as usual. Difficulties arise if there 

are duplicate key names in any of the external or current files. To circumvent 

this problem, an optional argument can be given, for example 

as \externaldocument[x-]{filename}, in which case all the markers in 

the external file will be prefixed with x-. Thus the external key intro is 

referred to with \ref{x-intro}. 

9.2.4 Smart referencing: the varioref package 

Package: When the object being referenced is so close that you know it will be 

varioref on the same or adjacent page, you very likely will want to alter the text 

accordingly. However, it might later switch from ‘same’ to ‘adjacent’ after 

another revision and you will have to change that text once more. The 

tools package varioref by Frank Mittelbach does this for you. It defines 

a command \vref{key} which is almost the same as ‘\ref{key} on page 

\pageref{key}.’ However, if \label{key} is on the current, previous, 

or next page, it prints appropriate text. Thus depending on the relative 

locations, Fig. \vref{f1} automatically produces one of 

Fig. 5 when \label on same page, 

Fig. 5 on the preceding page on previous page, 

Fig. 5 on the next page on following page, 

Fig. 5 on the facing page on opposite page with twoside option, 

Fig. 5 on page 24 when two or more pages away. 

The command \vpageref prints only the page part of the text, or ‘on 

this page’ if \label{key} is on the current page. It may also take one or 

two optional arguments: the texts for the current page and non-current 

pages. For example, the example \vpageref[above]{f1} prints one of


the example above when \label on same page, 

the example on the previous page on following page, 

the example on page 24 when two of more pages away, 

while the \vpageref[above example][example]{f1} outputs one of: 

the above example when \label on same page, 

the example on the next page on the following page, 

the example on page 24 when two of more pages away. 

A range of numbers and pages can be produced with the command 

\vrefrange, which takes two mandatory arguments, both label keys. It 

prints text like ‘5 to 7 on pages 212–214’, or ‘5 to 7 on the next page’, and 

so on. It takes one optional argument, the text to print when both labels 

are on the current page. 

The page part of \vrefrange only is printed with \vpagerefrange, 

which also takes two mandatory (label key) arguments and one optional 

(text) one. If both labels are on the same page, it behaves like \vpageref; 

the optional argument is printed if both are on the current page. 

Since English text is automatically printed by the above commands, 

they would not be very useful for documents, say, in Italian. The varioref 

package therefore accepts all the language options recognized by the 

babel system (Chapter 11) to reprogram the texts. With the option 

italian, one will get texts like in questa pagina for ‘on this page’ and 

nella pagina precedente for ‘on the preceding page’, and so on. 

All the texts are stored in special commands which can be redefined 

by the user: 

\reftextbefore previous but non-facing page 

\reftextfacebefore previous but facing page 

\reftextafter next but non-facing page 

\reftextfaceafter next but facing page 

\reftextfaraway{key} more than one page away 

\reftextpagerange{key1}{key2} page range of 2 labels 

\reftextlabelrange{key1}{key2} range of 2 different labels. 

The last three must be redefined with 1 or 2 arguments, as 

\renewcommand{\reftextfaraway}[1]{on page \pageref{#1}} 

\renewcommand{\reftextpagerange}[2]{on pages \pageref{#1}% 

--\pageref{#2}} 

\renewcommand{\reftextlabelrange}[2]{\ref{#1} to \ref{#2}} 

There is also \vreftextvario{text1}{text2} to provide altering texts 

for the above commands, so that the same text is not printed every time. 

9.3 Bibliographies 

Academic publications normally include a list of references, or bibliography, 

containing the names of other works that are cited within the text

9.3. Bibliographies 217 

by means of a running number. This is the system supported by standard 

L AT E X, described in the next section. The alternative citation method, 

with author name and year of publication, can be achieved with various 

extension packages, as described in Section 9.3.4). 

Often the bibliography has not been finalized before the main text is 

started. Especially for numerical bibliographies, it would be a great nuisance 

to have to go through the whole text and change all the numbering 

every time something was added to the bibliography. Therefore L AT E X is 

programmed not only to format the bibliography but also to keep track 

of its alterations and additions in order to modify the references in the 

text automatically. 

The bibliography is generated most conveniently and efficiently from 

a general database with the help of the BIBT E X program that should be part 

of the L AT E X installation. How to prepare such databases is explained in 

Chapter 14. The advantages of a database is that its entries can be reused 

again and again for all future documents, while the formatting style in 

each bibliography listing can vary according to the requirements of each 

document. One can, however, write it by hand without BIBT E X. 

9.3.1 Bibliography by hand 

The actual bibliography, whether created by BIBT E X or by hand, is placed 

inside the environment 

\begin{thebibliography}{sample label} 

entries 

\end{thebibliography} 

The individual entries in the bibliography each begin with the command 

\bibitem[label]{key} entry text 

Without the optional argument label, \bibitem produces a running number 

in square brackets as the label for the reference in the text. With 

label, one can give whatever indicator one wishes, such as an abbreviation 

of the author’s name, or an arbitrary reference number. The mandatory 

argument key is a reference key, much like that for the cross-referencing 

\label command (page 213), which is used to make the actual citation 

in the main text. The key name can be made up of any combination of 

letters, numbers, and symbols except commas. 

The bibliography information itself is contained in entry text, such as 

‘author, title, publisher, year, edition, page numbers’, possibly in various 

typefaces. The output text will be indented after the first line by a width 

equal to that of the sample label, so this should be as large as the longest 

label in the bibliography. For the standard application with running 

numbers, sample label should be a dummy number with as many digits


as the largest label (for example 99 if there are more than 10 but less than 

100 entries). 

For bibliographies with author–year citations (Section 9.3.4) the entries 

in the thebibliography are the same, except that the optional label must 

be present, taking a special form that will transfer the author and year 

texts to the citation commands. 

A sample (numerical) thebibliography environment could look as 

follows: 

\begin{thebibliography}{99} 

\bibitem{lamport} Leslie Lamport. \textsl{\LaTeX\ -- A Document 

Preparation System}, 2nd edition. Addison-Wesley, 

Reading, MA, 1994 

. . . . . . . . . 

\bibitem{knuth} Donald E. Knuth. \textsl{Computers and 

Typesetting Vol.\ A--E}. Addison-Wesley, 

Reading, MA, 1986 

\bibitem[6a]{knuth:a} Vol A: \textsl{The {\TeX}book}, 1986 

. . . . . . . . . 

\bibitem[6e]{knuth:e} Vol E: \textsl{Computer Modern 

Typefaces}, 1986 


Here lamport, knuth, and knuth:a have been chosen as keys. The 

sample label is given as 99 since a two-digit number produces sufficient 

indentation for the standard form of \bibitem. The entry with the key 

knuth is the sixth in the list and thus it automatically receives the label 

[6]; in order for its sub-entries knuth:a . . . knuth:e to be printed as [6a] 

. . . [6e], it is necessary to set their optional label arguments to 6a . . . 6e. 

(Yes, this does violate the automatic labeling system.) 

The bibliography is normally printed near the end of the document 

by placing the thebibliography environment at the desired location. 

For document classes book and report, the word Bibliography appears 

as an unnumbered chapter title at the beginning, whereas for article 

the word References is written as an unnumbered section heading. The 

above sample bibliography appears as: 

[1] Leslie Lamport. L AT E X – A Document Preparation System, 2nd edition. 

Addison-Wesley Co., Inc., Reading, MA, 1994 

. . . . . . . . . 

[6] Donald E. Knuth. Computers and Typesetting Vol. A–E. Addison-Wesley Co., 

Inc., Reading, MA, 1986 

[6a] Vol A: The T E Xbook, 1986 

. . . . . . . . . 

[6e] Vol E: Computer Modern Typefaces, 1986 

Exercise 9.4: Produce a bibliography with the thebibliography environment, 

with a label consisting of the first three letters of the author’s name followed


by the last digits of the year of publication. If the same author has more than 

one work in a given year, add a running letter to distinguish them, for example 

knu86c, knu86d. With such labels, it is appropriate to make the reference keys 

the same as the labels. The indentation depth should be 1.5 cm. 

Note: The indentation depth is determined by the argument sample label in the 

thebibliography environment. This is usually a dummy text with the right 

width. This width can be given precisely with \hspace{width}. 

Exercise 9.5: Copy the thebibliography environment from the above exercise 

to the end of your standard exercise file (but ahead of the \end{document} 

command). Refer to the entries in the bibliography by inserting \cite commands 

into your text. Make sure that the key in the \cite command is written exactly 

as in the \bibitem command in the thebibliography environment. 

9.3.2 Bibliography with BIBT E X 

Bibliographic databases and the BIBT E X program are fully described in 

Chapter 14. Here we just review the essential aspects for the L AT E X source 

file. 

• All the information for the literature references is placed in one or 

more databases. Each entry is given an identifying key (the key in 

the \bibitem argument) which is the handle to refer to it. 

• In the source text, citations to the references are made with various 

commands like \cite, \citet, \citep, taking the key as argument. 

• The bibliography style must be specified somewhere in the document 

with \bibliographystyle{bib style}; the bib style determines how 

the bibliography is to be formatted; the basic style is plain for 

numerical citations, or plainnat with the natbib package. 

• Give the command \bibliography{database} where the bibliography 

is to appear; database is a list of all databases to be searched. 

• Process the source file with L AT E X; run BIBT E X and then L AT E X twice. 

BIBT E X writes a bibliography following the formatting specifications of 

bib style into a thebibliography environment, storing it in a file with 

extension.bbl; this file is input by the \bibliography command. 

9.3.3 Numerical citations 

Standard L AT E X can only handle numerical bibliographies and citations. In 

this case, the citation in the text is made with the command 

\cite[extra]{key}


where key is the reference key that appears in the \bibitem command. 

With the sample bibliography on page 218, the source text 

For additional information about \LaTeX\ and \TeX\ see 

\cite{lamport} and \cite{knuth,knuth:a}. 

yields: For additional information about L AT E X and T E X see [1] and [6, 6a]. 

If the optional argument extra is included in the \cite command, this 

text is added after the label(s) but still inside the square brackets. 

The creation of a bibliographic database is described in 

\cite[Appendix B]{lamport}, while the program \BibTeX\ 

itself is explained in \cite[pages 74, 75]{lamport}. 

The creation of a bibliographic database is described in [1, Appendix B], 

while the program BIBT E X itself is explained in [1, pages 74, 75]. 

9.3.4 Author–year bibliographies with natbib 

An alternative bibliographic citation style, used in this book and in many 

scientific journals, makes reference to other published works by citing 

the author’s name and year of publication. In this case, the entries in the 

bibliographic listing are not numbered. The citation itself may be either 

parenthetical [Jones et al., 1999] or textual, as shown by Jones et al. [1999]. 

There may be some variations within this style: for example, in this book 

parentheses are used in place of square brackets and the name is left in 

normal type. None of this is supported by standard LAT EX, but rather by 

additional packages. 

There are a number of packages developed to deal with this situation, 

each requiring a different syntax for the optional label argument in the 

\bibitem command in order to transfer the author and year information 

to the citation commands. And each has its own bibliography style (.bst) 

file to allow BIBTEX to accomplish this. 

Package: The natbib package by Patrick W. Daly is the most universal of these, 

natbib being compatible with the \bibitem syntaxes, and thus with the.bst files, 

of other author–year packages such as apalike, chicago, and harvard, 

to name just a few. Furthermore, natbib can even produce numerical 

or superscript citations as well, and will also function with the standard 

(numerical) bibliography styles and \bibitem syntax, albeit only with 

numerical citations. 

Furthermore, natbib is integrated with the hyperref package (Section 

10.2.4) to make citations into automatic links to the bibliographic 

entry. 

The author–year \bibitem 

The required natbib form of the ‘optional’ label argument to \bibitem 

within the thebibliography environment is

\bibitem [short(year)long] {key} . . . 


where short is the short author list (e.g. Jones et al.) and long is an 

optional full author list. A sample bibliography might be: 

\begin{thebibliography}{99} 

\bibitem[James et al.(2001)James, Jones, and Smith]{JAM01} 

P. R. James, F. J. Jones, and R. T. Smith, ... 2001. 

\bibitem[Jones et al.(1999)Jones, Baker, and Toms]{JON99} 

F. J. Jones, H. P. Baker, and W. V. Toms, ... 1999. 

\bibitem[Jones et al.(2000a)Jones, Toms, and Baker]{JON00} 

F. J. Jones, W. V. Toms, and H. P. Baker, ... 2000a. 

\bibitem[Jones et al.(2000b)Jones, Toms, and Baker]{JON00b} 

F. J. Jones, W. V. Toms, and H. P. Baker, ... 2000b 


Such bibliographies will be automatically produced by BIBT E X when a 

bibliography style is invoked with the \bibliographystyle command 

that is compatible with natbib. Three such .bst files are provided with 

natbib, as substitutes for 3 of the 4 basic BIBT E X styles: plainnat, 

unsrtnat, and abbrvnat (Section 14.1). However, many others also exist 

as contributions for specific publishers and journals. 

Author–year citations 

There is a flexible citation syntax permitting parenthetical (\citep) and 

textual (\citet) citations, with the abbreviated or full author list, with 

optional notes both before and after the citations. For example, 

\citet{JON99} ⇒ Jones et al., (1999) 

\citet[pg.˜22]{JON99} ⇒ Jones et al., (1999, pg. 22) 

\citep{JON99} ⇒ (Jones et al., 1999) 

\citep[pg.˜22]{JON99} ⇒ (Jones et al., 1999, pg. 22) 

\citep[e.g.][]{JON99} ⇒ (e.g. Jones et al., 1999) 

\citep[e.g.][pg.˜22]{JON99} ⇒ (e.g. Jones et al., 1999, pg. 22) 

\citet*{JON99} ⇒ Jones, Baker, and Toms (1999) 

\citep*{JON00} ⇒ (Jones, Toms, and Baker, 2000) 

The citation punctuation (type of brackets, points between multiple citations, 

dates, and so on) can be selected by options when loaded, specified 

by declarations, or programmed to be associated with the bibliography 

style. See below. 

Multiple citations are available as usual, with compression of repeated 

authors: 

\citet{JON99,JAM01} ⇒ Jones et al. (1999); James et al. (2001) 

\citep{JON99,JAM01} ⇒ (Jones et al., 1999; James et al., 2001) 

\citep{JON99,JON00} ⇒ (Jones et al., 1999, 2000a) 

\citep{JON00,JON00b} ⇒ (Jones et al., 2000a,b)


In numerical citation mode, the command \citep behaves like the 

standard \cite command, printing the reference’s number in brackets, 

while \citet gives the authors’ names followed by the reference’s number. 

This is textual citation in numerical mode, allowing the author to use the 

same syntax for both citation modes. 

\citet{JON99} ⇒ Jones et al. [21] 

\citet[pg.˜22]{JON99} ⇒ Jones et al. [21, pg. 22] 

\citep{JON99} ⇒ [21] 

\citep[pg.˜22]{JON99} ⇒ [21, pg. 22] 

\citep[e.g.][]{JON99} ⇒ [e.g. 21] 

\citep[e.g.][pg.˜22]{JON99} ⇒ [e.g. 21, pg. 22] 

\citep{JON00,JON00b} ⇒ [21, 32] 

The commands \citeauthor{key} and \citeyear{key} print just 

the author names and year, respectively; a starred version \citeauthor 

prints the full author list. 

There also exist capitalized variants of these citation commands, 

\Citet, etc., for use at the beginning of a sentence, when the first author 

name begins with something like ‘von’. 

\citet{delaM89} has shown ... ⇒ de la Mar has shown . . . 

\Citet{delaM89} has shown ... ⇒ De la Mar has shown . . . 

Package options 

Many aspects of formatting the citations can be controlled by options 

specified when loading natbib with the \usepackage command: 

round citations in round parentheses (default); 

square citations in square brackets; 

curly citations in curly braces; 

angle citations in angle brackets; 

colon to separate multiple citations with semicolons (default); 

comma to use commas as separators; 

authoryear 

for author–year citations (default); 

numbers 

for numerical citations; 

super for superscripted numerical citations;


sectionbib 

puts the bibliography in a section instead a chapter; for use with 

the chapterbib package (Section 9.3.5); 

sort orders multiple citations into the sequence in which they appear 

in the list of references; 

sort&compress 

as sort but in addition multiple numerical citations are compressed 

if possible (as 3–6, 15); 

longnamesfirst 

makes the first citation of any reference the equivalent of the 

starred variant (full author list) and subsequent citations normal 

(abbreviated list). 

Note: natbib defaults to author–year citations, but if any single \bibitem 

command does not conform to any of the recognized author–year syntaxes, 

it will switch automatically to numerical citations. 

Further customizations 

There are several formatting commands in natbib that may be (re)defined 

by the author to achieve special effects in the bibliography, such as font 

or spacing between entries. In standard L AT E X, these are all fixed. 

\bibsection 

can be redefined to the desired sectioning command for introducing 

the list of references. This is normally \section* or 

\chapter*. 

\bibpreamble 

is printed after the bibliography heading but before the actual 

list of references; is normally undefined, but can be defined to 

insert any desired leading text. 

\bibfont 

can be a font declaration, e.g. \small to apply to the whole list 

of references; is normally undefined. 

\citenumfont 

can be a font declaration or command like \itshape or \textit 

for formatting numerical citations; normally undefined. 

\bibnumfmt 

is a command with an argument to format the numbers in the 

list of references; the default definition is [#1].


\bibhang 

is the indentation after the first line of each reference; change 

this with the \setlength command. 

\bibsep 

is the vertical spacing between references; change this with the 

\setlength command. 

There are many other customization possibilities and specialized features 

to natbib which can be found in the supplied documentation (process 

natbib.dtx if it not already provided as a.dvi or.pdf file) or in the 

short summary natnotes.tex. 

9.3.5 Multiple bibliographies 

Package: 

chapterbib 

In standard L AT E X, with or without BIBT E X, there can only be one bibliogra- 

phy per document. However, it is not uncommon for a book to consist of 

independent chapters, each with its own bibliography and local citations. 

The chapterbib package by Donald Arseneau solves this problem. It 

requires that the document be broken up into individual source files that 

are read into the main file with the \include command of Section 9.1.2. 

Each such file must contain its own thebibliography environment, or its 

own \bibliographystyle and \bibliography commands if BIBT E X is to 

be used. The citation commands will work locally within each file. Even 

if the same reference appears in several bibliographies, with a different 

number each time, the citations will refer to the one in the local file. 

One must process BIBT E X individually on each of the separate files, 

creating a series of.bbl files. Then the document is reprocessed with L AT E X, 

at least twice; the.bbl files are read in by the individual \bibliography 


If you cannot make use of the \include commands because you do 

not want each file to begin on a new page, you can still localize the 

bibliographies and their citations with the cbunit environment. This 

does not work with BIBT E X, however. In that case, you must still put the 

bibliographic units into separate files, and then read them in with the 

\cbinput{file} command. Set the package option draft when loading 

chapterbib, process with L AT E X, process each file with BIBT E X, remove the 

draft option, process with L AT E X at least twice. 

With classes book and report, the bibliography is placed in an unnumbered 

chapter; this is undesirable if each chapter is to have its own 

bibliography. In this case, add the package option sectionbib to change 

the bibliography to an unnumbered section instead. 

The chapterbib package is compatible with natbib except for the 

sectionbib option. The two packages clash on their redefinitions of bibliography. 

In this case, add the option to natbib instead of to chapterbib.

9.4 Keyword index 

9.4. Keyword index 225 

L AT E X itself does not produce a keyword index automatically as it does a 

table of contents, but it can prepare input data for the MakeIndex program 

(Section 9.4.3) that does generate such an index, in a form that L AT E X can 

use on a later run. 

9.4.1 The index environment 

A keyword index is formatted in the environment 

\begin{theindex} index entries \end{theindex} 

which switches to a two-column page format with a running head INDEX. 

The first page of the index carries the heading Index in the same size as 

a chapter heading for the book and report document classes, and as a 

section heading for the article class. (More precisely, the actual word 

printed is contained in the command \indexname which may be redefined 

for other languages.) The individual entries are made with commands 

\item \subitem \subsubitem and \indexspace 

followed by the keywords and their page numbers. For example, 

commands, 18 

arguments, 19, 101 

multiple, 103, 104 

replacement symbol, 20 

as environments, 42 

used as arguments for sectioning 

commands, 41, 42 

displayed text, 21–32 

\item commands, 18 

\subitem arguments, 19, 101 

\subsubitem multiple, 103, 104 

\subsubitem replacement symbol, 20 

\subitem as environments, 42 

\subitem used as arguments for 

sectioning commands, 41, 42 

\indexspace 

\item displayed text, 21--32 

If the text entry is too long for one line, it is broken and continued on 

the next line, indented deeper than all other lines, as in the above example 

‘used as argument for sectioning commands, 41, 42’. The command 

\indexspace leaves a blank line in the index. 

9.4.2 Preparing the index entries 

The theindex environment only sets up a suitable format for the index. 

The entries themselves, as well as their page numbers, are generated by 

the MakeIndex program described in Section 9.4.3. This program requires 

input information from the L AT E X file in the form of unsorted keywords 

and page numbers. The author enters this information in the text file 


\index{index entry}


where index entry is the text to be entered into the index. It may contain 

any combination of letters, numbers, and symbols including even 

command characters and blanks. This means that even commands may 

be included in the index entry, such as \index{\LaTeX\ logo}. Even 

the one command that may otherwise never be used as an argument, 

\verb, may be included. However, if index entry does contain a command, 

\index may not be used as an argument of another command. 

Normally all the \index commands are ignored by L AT E X and do absolutely 

nothing. They are activated only when the preamble contains the 


\makeindex 

in which case, a file with the document’s root name plus the extension 

.idx is opened. Now the \index commands write index entry and the 

current page number to this file in the form 

\indexentry{index entry}{pagenumber} 

It is this .idx file that is the input to the MakeIndex program. There are 

special symbols that may be used in the index entry to indicate main and 

subitems and other refinements. 

The MakeIndex program expects the \index entries to be of one of 

the three forms: 

\index{main entry} 

\index{main entry!sub entry} 

\index{main entry!sub entry!sub sub entry} 

The individual main and sub-entries may contain any combination of 

characters with the exceptions of !, @, and |. The exclamation point is the 

divider between the various entry fields. If the \index command contains 

only a main entry, this will become the text for an \item command. The 

main entries will be in alphabetical order. 

If the \index command contains a main and a sub-entry, the text 

sub entry will be assigned to a \subitem command underneath the corresponding 

main entry. The texts for the \subitem will also be in alphabetical 

order. Similarly the text of a sub sub entry will appear following 

a \subsubitem command, alphabetically ordered, below the appropriate 

sub entry text. 

The main and sub-entries may also contain special characters and even 

L AT E X commands that are to be neglected during alphabetization. This is 

indicated by an entry of the form lex entry@print entry, in which lex entry 

is used for the alphabetical ordering while print entry is the actual text 

to be output. For example, in the keyword index in this book, which was 

generated with MakeIndex, all the command names appear as input text 

and are ordered without the preceding \ character. These entries were 

made in the text as \index{put@\verb=\put=}.


An entry may be terminated with the character sequences |( or |) 

to designate the beginning and end of a range of page numbers. For 

example, 

\index{table|(} on page 95 

\index{table|)} on page 110 

produces the entry ‘table’ with page numbers given as 95–110. 

Instead of having the page number printed after the entry, a reference 

may be made to another index entry. For example, with 

\index{space|see{blank}} 

\index{table|seealso{array}} 

the entries ‘space, see blank’ and ‘table, see also array’ are made in the 

index. 

The three characters !, @, and | therefore have special functions for 

the MakeIndex program. In order to print these characters literally as 

text without their functions, the quote character " must precede them. 

For example, "! represents a literal exclamation point and not the entry 

divider. 

The quote character itself is thus a fourth special symbol and must be 

entered literally as "". However, there is a special rule in MakeIndex syntax 

that says a quote character preceded by a backslash will be interpreted 

as part of a command: thus \" may be used to put a German umlaut in 

an entry (as in \index{Knappen, J\"org}). This special rule can lead to 

additional problems at times: in this book the index entry \! had to be 

typed as "\"!. 

It is possible to specify varying fonts for the page number. For example, 

in the index of this book, page numbers are set in bold face to indicate 

the page where a command is defined or first explained. This is achieved 

with an entry of the form 

\index{blank} on page 11, and 

\index{blank|textbf} on page 22 

which in the second case puts the page number for this entry as the 

argument of a command \textbf. The line in the theindex environment 

becomes 

\item blank, 11, \textbf{22} 

Note: The vertical bar in the \index entry is not a typing error, but 

must replace the backslash as the L AT E X command symbol under these 

circumstances. 

There is a file idx.tex in the L AT E X installation, which can improve the read- 

! ability of the .idx files. By processing the idx.tex file (that is, calling latex 

idx), the user is prompted for the name of the.idx file to list:


******************************** After the root name of the.idx file has 

* Enter idx file’s first Name. * been entered from the keyboard, a two- 

******************************** column page format is output containing 

the index entry texts listed under 

\filename= 

page headings of the form Page n. No 

further information is obtained through such a formatted listing, but it is far 

easier to work with than a straightforward output of the.idx file alone. 

Even if the \index commands have no effect without the \makeindex command 

in the preamble, it is a good idea to include them in the text from the initial 

development of the document. The \makeindex command can be included later 

when the final version of the document is ready. At this point, the MakeIndex 

program can be run to produce the theindex environment from the data in the 

.idx file just as BIBTEX produces thebibliography environments. This tool is 

described in the next section. 

The LAT Package: 

EX installation also contains the package showidx, which prints 

showidx the index entries from the \index commands as marginal notes, beginning 

at the top of the page on which they occur. This is useful for going through 

a preliminary version of the document to check that all index entries are 

really on the proper page or whether additional entries have to be made. 

If showidx is used, it is a good idea to increase the width of the marginal 

note box with the declaration \marginparwidth (Section 4.10.6) in the 

preamble. Unfortunately, the present book format is not suitable for such 

a demonstration. 

9.4.3 Running MakeIndex 

The tiresome drudgery of making up the theindex environment for a 

keyword index can be dispensed with if the program MakeIndex is available. 

This program was written by Pehong Chen with the support of 

Leslie Lamport. We will give only an abbreviated description of its use 

here. More details are in the documentation accompanying the program 

package. 

The program MakeIndex processes the .idx file, producing as output 

another file with the root name of the document and the extension.ind, 

containing the complete theindex environment. It is run by clicking the 

appropriate icon in the editor shell, or from a command line with 

makeindex root name.idx or simply makeindex root name 

A subsequent LAT Package: 

EX processing outputs the index at the location of 

makeidx the \printindex command which, together with the \see command, is 

defined in the package makeidx. Thus the production of an index in this 

way requires the package file makeidx to be loaded with the \usepackage 


The alphabetical ordering of MakeIndex normally follows the standard 

ascii code, first symbols, then digits, and finally letters, where upper case 

comes before lower case. Blanks are included as symbols. There are a


number of options that allow these rules to be changed. How the options 

are invoked when the program is called depends on the computer type, 

but we assume here a hyphen preceding the option letter, as in 

makeindex -g -l root name 

The most important options are: 

-l Letter ordering: blanks are ignored when sorting; 

-c Compress blanks: multiple and leading blanks are removed as 

they are in normal L AT E X; 

-g German ordering: following the German rules in which symbols 

precede letters (lower before upper case) which precede numbers; 

the sequences "a, "o, "u, and "s (the codes for ä, ö, ü, and ß 

in the usual L AT E X adaptations for German) are treated as though 

they were ae, oe, ue, and ss, which is standard German practice; 

-s Style specification: allows the name of an index formatting file 

to be included to redefine the functioning of MakeIndex. 

The -s option reads in an index style file containing commands to 

define both the input and output of the MakeIndex program. For example, 

it is possible to change the special symbols !, @, |, and " so that different 

characters execute their functions and they themselves revert to being 

pure text. 

The style-defining file consists of a list of pairs in the form keyword attribute. 

The attribute is either one character in single quotes (for example, 

’z’), or a character string in double quotes (for example, "a string"). 

The most important keywords, together with their default definitions, are: 

quote ’"’ defines the quote symbol; 

level ’!’ defines the entry separation symbol; 

actual ’@’ defines the lexical switch symbol; 

encap ’|’ defines the dummy command symbol for page formatting. 

There are many more keywords for defining complicated output structures. 

These are described in the documentation that should be included 

with the MakeIndex program package. 

The file makeindex.tex contains a short manual by Leslie Lamport 

(but without mentioning style definitions).


9.4.4 Glossary 

A ‘glossary’ is a special index of terms and phrases alphabetically ordered 

together with their explanations. To help set up a glossary, L AT E X offers 

the commands 

\makeglossary in the preamble and 

\glossary{glossary entry} in the text part 

which function just like the commands for making up a keyword index. 

The entries are written to a file with extension .glo after the command 

\makeglossary has been given in the preamble. The form of these file 

entries from each \glossary command is 

\glossaryentry{glossary entry}{pagenumber} 

The information the .glo file can be used to establish a glossary. 

However, there is no equivalent to the theindex environment for a glossary, 

but a recommended structure is the description environment 

(Section 4.3.3) or a special list environment (Section 4.4).

10 

PostScript and PDF 

PostScript is a printing and plotting language developed by the Adobe 

Systems Inc. Unlike the printing instructions for other types of printers, 

the PostScript code may be written to an ascii file, viewed on a computer 

monitor, put online, downloaded, included in emails. It therefore 

represents a type of electronic paper. 

It is also a programming language, and thus offers enormous possibilities 

for including special effects that cannot easily be inserted with other 

output formats. Several packages exist to exploit these, packages which 

are then limited to PostScript output. However, the PostScript advantages 

are so attractive that this can be justified. 

Portable Document Format, PDF, is another Adobe product, a successor 

to PostScript, inheriting many of it properties. However, PDF includes 

many features to make it applicable to true electronic documents: internal 

and external links, bookmarks, animation, encryption, etc., features that 

have no correspondence in paper. It represents a revolutionary new type 

of information medium. 

PostScript has played a very important role in the development of 

L AT E X, setting the standards for graphics inclusion, and offering additional 

font families over the standard Computer Modern set. Now PDF carries 

on that tradition. With the pdfT E X program, PDF is actually married to T E X 

and thus to L AT E X. 

In the next sections we describe the generation of PostScript output 

with L AT E X, the usage of PostScript (type 1) fonts, and then the production 

of PDF output with all its bells and whistles. 

10.1 L AT E X and PostScript 

PostScript output is best produced with the dvips driver program written 

by Tomas Rokicki, available with most T E X installations. It is this very 

powerful and flexible program that has set the standard for DVI conversion 

programs. 

231

232 Chapter 10. PostScript and PDF 

Its two most significant features are the inclusion of graphics as encapsulated 

PostScript images, as described in Chapter 6, and the capability of 

making use of PostScript fonts in place of the Computer Modern ones (Section 

G.5). There are also enhanced drawing capabilities, as demonstrated 

in Sections 13.2.2 and 13.3.2. 

10.1.1 The dvips driver 

The dvips driver program operates on the DVI result from T E X, with or 

without L AT E X, to generate PostScript output, which may be written to a 

file or sent directly to a printer. One either clicks the appropriate icon in 

the editor shell, or issues the command from a command line prompt: 

dvips myfile 

to process myfile.dvi and generate myfile.ps. (The L AT E X source text 

was of course in myfile.tex.) The PostScript file may be viewed with 

a PostScript viewer like GhostView or sent to a PostScript printer, or 

included in an email. 

There are many options that may be included in the command line, 

appearing between dvips and the file name. For example, to output five 

pages only starting at the 10th, sending them to a file named p5-15.ps, 

give 

dvips -p 10 -n 5 -o p5-15 myfile 

You may also specify the first (-p) and last (-l) page numbers, whether 

only odd (-A) or even (-B) pages are to be printed, or in reverse order (-r). 

The page numbers with -p and -l refer to the actual page numbers in 

the document, or rather to its first occurrence. Hence -p 8 might start 

with page viii rather than page 8. Specifying -p =8 will start with the 8th 

page, regardless of its actual number. The list of options can be printed 

to the monitor with dvips --help. 

One useful option is t landscape to rotate the output into landscape 

mode, which should be done if the document has been processed with 

the landscape option in \documentclass. However, dvips has another 

way of accomplishing this: add the driver-specific command 

\special{landscape} 

in the source text itself and dvips will automatically print in landscape 

mode. 

The output is written to a file with the same root name as the input, 

but with extension .ps, unless the option -o filename is given. To send 

the output directly to a printer, specify that printer as the output ‘file’, as 

-o \lpt1. Beware, if that printer is not a PostScript printer, it will simply 

output the PostScript code directly as text, something that normally uses 

a considerable amount of paper.

10.1. L AT E X and PostScript 233 

A full manual for dvips should be available as dvips.ps or dvips.pdf; 

otherwise the source file dvips.tex can be processed (with T E X, not L AT E X!). 

The program is configured to the local installation by means of a file 

config.ps that should be located in texmf\dvips\config. This sets 

up certain parameters for the default printer, as well as various paper 

formats, indicating which is to be the default (European A4 or American 

letter). This can be overridden with -t a4 or -t letter. Another 

important configuration item is a list of font mapping files (Section 10.1.3), 

to associate T E X font names with actual PostScript fonts and their files. 

Additional configuration files are usually provided, for specific printers 

or other font mappings, all with the name config plus a different 

extension. They may be invoked on the command line with the option 

-P plus extension. For example, config.cx contains lines for a generic 

PostScript printer with resolution of 300 dpi; with -P cx, this file is input, 

overriding the default values in config.ps. Another example for adding 

PostScript fonts is given in Section 10.1.3. 

Of course, the dvips driver manages not only PostScript fonts but 

also the standard T E X pixel fonts generated by the METAFONT program 

(Section G.3). These bitmaps are stored in .pk files that need to be 

generated for the desired printer, size, and resolution. Originally, the 

user had to ensure that they were present before the driver program 

could address them. Fortunately, dvips has revolutionized this process 

by calling METAFONT itself if the needed.pk files are missing. 

However, it is the use of PostScript fonts that make dvips especially 

interesting. We describe in some detail how they are integrated into 

the T E X/L AT E X system in the next sections since almost all of this applies 

equally well to PDF output. 

10.1.2 Invoking PostScript fonts 

Under the New Font Selection Scheme of Appendix A, it is relatively 

simple to invoke the PostScript fonts, provided that the font definition 

.fd files are provided. These are normally to be found in the directory 

psnfss in the standard installations. It also contains packages for the 35 

standard PostScript fonts, plus those for a number of other commercially 

available fonts. These packages are actually quite simple; for example, 

the times.sty file contains only four lines, which are listed on page 379. 

All these packages do is redefine the three font families, \rmdefault, 

\sfdefault, \ttdefault and the bold face default attribute \bfdefault. 

The packages for the 35 standard fonts, and the three assignments 

that they make, are listed in Table 10.1 on the next page. (The 35 fonts 

include italic and bold variants of these.) 

These packages are only meaningful when the resulting DVI output is 

to be transformed with dvips to PostScript or PDF files, or when the L AT E X 

source is to be processed with pdfT E X to generate PDF directly. Other


Table 10.1: The psnfss packages and their fonts 

Package \rmfamily \sffamily \ttfamily 

times.sty Times-Roman Helvetica Courier 

palatino.sty Palatino Helvetica Courier 

newcent.sty NewCenturySchlbk AvantGarde Courier 

bookman.sty Bookman AvantGarde Courier 

avant.sty AvantGarde 

helvet.sty Helvetica 

drivers can only handle the T E X pixel fonts described in Appendix G (also 

known as type 3 fonts). Some drivers, in particular the monitor previewers 

xdvi and windvi, will also invoke programs to generate the bitmap .pk 

files from the PostScript.pfb source files. 

Why bother to use these packages when dvips is capable of including 

the pixel fonts just as well? The answer is that the standard PostScript 

fonts are well known from other applications, and there are additional 

commercially available fonts in PostScript format; they therefore offer a 

wider choice than simply the traditional Computer Modern fonts that are 

limited to T E X and L AT E X. 

Furthermore, although the pixel fonts can be included in PDF output, 

the results viewed on the monitor are simply atrocious. They are slow 

to draw and appear very fuzzy. (They do print all right, however.) PDF 

should make use of PostScript fonts exclusively. In Section G.6 we explain 

how the standard Computer Modern and Extended Modern fonts are now 

available in PostScript encoding, especially for this need. 

10.1.3 Installing PostScript fonts 

The packages listed in the previous section invoke the PostScript fonts by means 

! of the NFSS system, provided they have been properly included in the installation. 

This means that the following files must be present: 

.tfm the font metric files for both the virtual and raw fonts; 

.fd the NFSS font definition files associating the font attributes to precise 

virtual font names; these and the.tfm files are the only ones read by L AT E X 

itself; 

.vf virtual font files that instruct the DVI driver how to produce the characters; 

usually this refers to characters in a real font with different encoding; 

.map mapping files that tell the dvips driver the true internal names of the real 

fonts, plus any re-encodings or distortions that must be undertaken.

10.1. L AT E X and PostScript 235 

The.map files are a feature of dvips that has been taken over by pdfT E X and 

other programs that make use of PostScript fonts. It translates the L AT E X names 

for the fonts into the internal names used by PostScript itself. For example, the 

line 

ptmr8r Times-Roman "TeXBase1Encoding ReEncodeFont"


The encoding and family indicate the font definition file that associates these 

attributes with real font names, in this case ot1ptm.fd. In that file, one finds 

this set of attributes to be assigned to font ptmbo7t; this is the Berry name for 

Times-BoldOblique, 7-bit-text (Tables G.2 and G.3 on pages 504 and 505). 

Our sample input code is \AE; according to file ot1enc.def, this corresponds 

to character 29 in the OT1 encoding. L AT E X thus writes instructions to the .dvi 

file to output the symbol in position 29 of font ptmbo7t. And with that, L AT E X is 

finished with its part of the operation. 

The next step is to process the .dvi file with the dvips driver. When this 

program searches for a font, it first looks in the map files, then it seeks the .vf 

virtual font files, and finally it tries to find it as a .pk pixel file. In our case, it 

discovers the virtual font ptmbo7t.vf, which tells the driver that character 29 is 

to be symbol 198 of font ptmbo8r. This is the same basic font but with 8-bit-raw 

encoding. Again dvips searches for this next font, and finds it in a mapping file, 

containing the line 

ptmbo8r Times-Bold ".167 SlantFont TeXBase1Encoding 

ReEncodeFont"

10.2. Portable Document Format 237 

Where Adobe makes its money is in the sales of programs to produce 

and manipulate PDF files. Of these, the main ones are: 

• Acrobat Distiller to convert PostScript into PDF, and 

• Acrobat program to modify existing PDF files. 

In the latter case, ‘modification’ means adding links, changing colors, 

including navigational aids, and inserting notes; it does not mean altering 

the actual textual content in any way. 

Today most graphical and text processing programs are capable of 

generating PDF output. For the L AT E X user, there are three routes that 

may be taken to produce PDF output. The obvious one is.tex →.dvi → 

.ps → .pdf using dvips to generate the PostScript and then Distiller or 

Ghostscript to create the PDF file. 

The other two methods are to employ a DVI-to-PDF driver, or the pdfT E X 

program to go directly to PDF, by-passing the intermediate DVI output. 

We describe each of these in the next sections. 

Used on their own, all of these methods produce a simple PDF file 

containing nothing more than the text itself, a sort of electronic paper. 

The hypertext features that make an electronic document really live can 

be added automatically with the help of the hyperref package described 

in Section 10.2.4. 

10.2.2 The dvipdfm driver 

The driver program dvipdfm by Mark A. Wicks is the equivalent of dvips 

for PDF output. As its name implies, it converts DVI to PDF output. (The 

m suffix is to distinguish it from a previously announced dvipdf program 

that never came to fruition.) 

Graphics input can be included with the graphics or graphicx packages 

(Section 6.1) by specifying the option dvipdfm. The allowed graphics 

formats are JPEG, PNG, and PDF. 

The program is called from a command line with 

dvipdfm myfile 

to convert myfile.dvi to myfile.pdf. As for dvips, there are many 

options that may be included. It is best to read the provided manual in 

dvipdfm.pdf or dvipdfm.dvi, or produce it by processing dvipdfm.tex 

with T E X. 

The configuration is set up by means of a file named config in the 

dvipdfm directory. This is set up similarly to the config.ps file for 

dvips, including the list of mapping files, although these have a different 

syntax to those of dvips. 

Hypertext features and other PDF additions can be added with the 

hyperref package (Section 10.2.4).


10.2.3 The pdfT E X program 

The most direct method of getting PDF output is to process the L AT E X file 

with pdfT E X, a variation on the T E X program developed by Hàn Thê ´ Thành. 

The following points should be noted: 

• pdfT E X behaves just like the regular T E X program unless the declaration 

\pdfoutput=1 is issued before the first page is output; in this 

case, it generates PDF instead of DVI output and activates many new 

T E X-like commands for including PDF features. 

• Since pdfT E X combines the functionality of both the T E X program 

and a DVI driver, it must be able to deal with fonts in the way that a 

driver program would. Originally this meant that it could only handle 

PostScript type 1 fonts, but now it can also cope with METAFONT 

bitmap fonts (type 3) as well as TrueType fonts. The bitmap fonts 

however look very bad when viewed, but this is a viewer problem. 

• When started, the program reads a configuration file pdftex.cfg 

which can set many parameters, as such page size, offsets, PDF 

compression level, and specifies the names of font mapping files. 

It may also set the default output to PDF, avoiding having to set 

\pdfoutput in the source text. 

• The font map files are identical to those used by dvips (page 236); 

they relate the T E X name for the font to various characteristics, 

encoding scheme, true name, and specify the file containing the 

character drawing instructions. 

• New T E X-like commands exist to include PDF features, like compression 

level, page attributes, document information, opening setup, 

forms, annotations, links, bookmarks, and article threads. 

• Graphics can be included in PNG, TIFF, or JPEG formats; PDF files 

may be inserted provided they contain only a single page without 

any fonts or bitmaps. The graphics packages (Section 6.1) support 

pdfT E X by means of a pdftex graphics option. 

To run pdfT E X with L AT E X (often referred to as pdfL AT E X), it is necessary 

to generate a L AT E X format (Section B.1.3) using pdfT E X instead of T E X, with 

the -ini option, renaming the resulting latex.fmt to pdflatex.fmt; it 

is then called with the command 

pdftex &pdflatex 

Most installations contain a command pdflatex that translates to this 

combination. The editor shell should also have an icon that invokes this 

command.


We do not describe the new T E X-like commands to add PDF features 

to the output, preferring to use the L AT E X-like commands of the hyperref 

package. For information on the intrinsic new pdfT E X commands, see The 

L AT E X Web Companion (Goossens and Rahtz, 1999), which also explains the 

hyperref package in detail. The pdfT E X manual should also be available 

on your system under texmf\doc\pdftex\ in various formats. 

So which of the above methods is to be recommended? The PostScript 

+ Distiller is the most inelegant, involving two intermediate steps (DVI 

and PS) and requires purchasing the Distiller program. Alternatively one 

can use the free GhostView program for the conversion to PDF, but one 

still has the intermediate steps. And if one selects this route, one must be 

certain that dvips uses the PostScript version of the Computer Modern 

fonts, by setting up the mapping files accordingly. 

The dvipdfm program is in the original spirit of T E X, that uses DVI 

as a universal intermediate format for all outputs. Purists might tend 

to respect this ideal. After all, no one ever considered rewriting T E X to 

produce PostScript output directly. That said, one must consider that T E X 

was invented in the days when no one printer specification dominated 

the field. Today, PDF is much more than a printer format; it is the means 

of representing documents electronically. That alone would not justify 

preferring pdfT E X over a DVI-to-PDF converter, nor would the fact that 

it saves a processing step; the deciding argument is that pdfT E X has 

established itself as reliable, robust, and flexible. 

In the end, it is likely a question of which program one is more 

comfortable with, and which one has given the better results for the 

particular user. 

10.2.4 The hyperref package 

Package: Sebastian Rahtz has written an ambitious package hyperref to add auhyperref 

tomatic hypertext links to LAT EX documents that are intended to become 

HTML (using LAT EX2HTML, Section E.1.1 or TEX4ht, Section E.1.2) or PDF 

files (using any of the methods described above). Not only do all internal 

cross-references link to their reference points, citations are also linked to 

the list of references, table of contents to the section headings, and index 

listings to the original text. Additional links to external documents are 

possible by means of a single syntax for all types of output. And it does 

this with a uniform syntax even though all the conversion methods have 

their own syntaxes. 

Because this package redefines many aspects of the LAT EX kernel and 

of many other packages, it should always be loaded as the very last one 

to be sure that it has the final word. 

It is especially for PDF output, whether by means of PostScript plus 

Distiller, with dvipdfm, or directly with pdfTEX, that this package is extremely 

useful. Figure 10.1 on the next page gives an example of such a


Figure 10.1: Output produced by pdfT E X with the hyperref package for 

automatic links, with an outline list at the left. 

PDF file, where it should be noted that the table of content entries are all 

links to their sections, as are the entries in the left-hand outline, which is 

also generated automatically. The package provides L AT E X-like commands 

for the PDF features. 

Before we go any further, we should explain certain terms that are 

used below. 

Links are indicated as a piece of text in a different color, or surrounded by 

a frame, or may be an image, which when clicked cause the viewer 

to jump to another document or some other place in the current 

document. 

Anchors (also called targets) are the destinations for links. They are provided 

with an identifying name so that the link can find them. Some 

anchors are meant only for internal linking within the document, 

while others are visible from outside. 

Thumbnails are small images of the pages that are displayed in a thumbnail 

window by the viewer. They may be used to locate a certain 

page more quickly, especially if it has an obviously different appearance 

because of a figure it contains. Double clicking the page image 

causes the viewer to jump to that page. 

Bookmarks are special links to places in the document, associated with 

a view, usually meaning a zoom factor, or a specification like ‘fit


page to width of window’. The bookmark window displays all these, 

and by double clicking them, one jumps to that location and view. 

Bookmarks are ordered within a tree, with sub-bookmarks and so 

on. The bookmark display may have the tree opened or closed. 

Outline is the usual use for bookmarks, building a table of contents, or 

outline, for the whole document. The word ‘outline’ is used here, 

but the PDF viewers refer to ‘bookmarks’. The latter if really far 

more general. 

Options for hyperref 

There are a large number of parameters that may be configured for the 

package, either as options in the \usepackage command, or with the 

\hypersetup command, or with a configuration file. An example of 

setting options with the first method is: 

\usepackage[pdftex, 

pdftitle={Graphics and Color with LaTeX}, 

pdfauthor={Patrick W Daly}, 

pdfsubject={Importing images and use of color in LaTeX}, 

pdfkeywords={LaTeX, graphics, color}, 

pdfpagemode=UseOutlines, 

bookmarks,bookmarksopen, 

pdfstartview=FitH, 

colorlinks,linkcolor=blue,citecolor=blue, 

urlcolor=red, 

] 

{hyperref} 

Options that take a value are given with an equals sign and the value; 

for multiple values, or text, the values must be placed within curly braces, 

as shown in the example. Options that are either true or false are 

simply given (to make them true) or equated to false. Options may also 

be given with the \hypersetup command, as for example: 

\hypersetup{colorlinks, linkcolor=blue} 

Values that are colors must be a name of a defined color, as explained 

in Section 6.2; the name must be defined before it is first used. However, 

some color values must be the three numbers of the rgb model. The 

colors of link borders may only be given as rgb. The hyperref package 

automatically loads the color package if colorlinks is given. 

One option must specify the driver name for which the output is 

intended; in the above example this is pdftex, but other possibilities are: 

dvipdf dvipdfm dvips dvipsone dviwindo hypertex 

latex2html pdftex tex4ht textures vtex


The option hypertex is not for any particular driver, but rather for any 

that conforms to the HyperT E X guidelines, such as dvips. There is also a 

draft option that turns off all hypertext features. 

The next four options in the example (pdftitle–pdfkeywords) are 

entries for the PDF document information stored in the file and displayed 

only when the viewer shows the document summary. The complete set of 

such options is: 

pdfauthor pdfcreator pdfkeywords 

pdfproducer pdfsubject pdftitle 

The standard paper size options of Section 3.1.1 are all recognized 

and one of them must be given, either as an option to hyperref or as a 

global option in the \documentclass command. This is necessary to set 

up the size of the PDF page. Reminder, these are: 

a4paper a5paper b5paper 

executivepaper legalpaper letterpaper 

Here is a list of all of the remaining options, with default value in 

brackets. 

4 (false) use Acrobat 4 features; 

anchorcolor 

(black) set color of anchors (targets of internal links); 

backref 

(false) adds links in the bibliography back to the citation; with 

pagebackref, a list of pages is added to the bibliography, each 

being a link back to the citation; 

baseurl 

determines the prefix for URLs given with \href; 

bookmarks 

(true) make bookmarks out of sectioning commands; 

bookmarksnumbered 

(false) include section numbers in bookmark text; 

bookmarksopen 

(false) start document with bookmarks tree opened; 

bookmarksopenlevel 

(max) the level to which bookmarks are opened; 

bookmarkstype 

(toc) specifies which ‘contents’ file used for bookmarks; 

breaklinks 

(false) allows link text to break over lines;


citebordercolor 

(0 1 0) the color of the border around citations; 

citecolor 

(green) color of citation links when colorlinks given; 

colorlinks 

(false, but true for tex4ht and dviwindo) link and anchor 

texts are to be colored, according to color specifications for the 

various types of links; without this, link text is surrounded by a 

frame; 

debug (false) output diagnostic messages; 

draft (false) turn off all hyperlinking; 

extension 

(dvi) default extension of linked files using the xr package (Section 

9.2.3); the package tries to guess whether the file specified 

by \externaldocument should be.dvi or.pdf; with PDF output, 

the viewer will open that external file and switch to it; 

filebordercolor 

(0 .5 .5) color of the border around file links; 

filecolor 

(cyan) color of file links when colorlinks given; 

frenchlinks 

(false) use small caps instead of color for links; 

hyperfigures 

(false) make figures into hyperlinks; 

hyperindex 

(true) make index entries into links back to referenced text; 

hypertexnames 

(true) use guessable (obvious) names for links; 

implicit 

(true) redefine L AT E X internal commands; 

linkbordercolor 

(1 0 0) color of the border around links; 

linkcolor 

(red) color of link text when colorlinks given; 

linktocpage 

(false) the page numbers in the table of contents (or list of 

figures/tables) are the links, rather than the text itself;


menubordercolor 

(1 0 0) color of the border around Acrobat menu links; 

menucolor 

(red) color for Acrobat menu links when colorlinks given; 

naturalnames 

(false) use L AT E X-computed names for links; 

nesting 

(false) links are allowed to be nested, although none of the 

existing drivers support this; 

pageanchor 

(true) an anchor (target) is placed automatically on every page, 

something that is necessary of the table of contents is to act as 

links; 

pagebackref 

(false) adds to each entry in the bibliography a list of page 

numbers linked to their citations; 

pagebordercolor 

(1 1 0) color of the border around page links; 

pagecolor 

(red) color of page links when colorlinks given; 

pdfborder 

(0 0 1, but 0 0 0 with colorlinks) width of the border around 

links in PDF files; the first two numbers are always zero, the third 

is the line thickness in points; 

pdfcenterwindow 

(false) positions the document window in the center of the 

computer monitor when the PDF file is opened; 

pdffitwindow 

(false) resizes the document window to fit the first page when 

the PDF file is opened; 

pdfhighlight 

(/I) determines how the Acrobat menu buttons behave when 

clicked; /I for inverse, /N for none, /O for outline, /P inset 

highlighting; 

pdfmenubar 

(true) makes the menu bar in the PDF viewer visible; 

pdfnewwindow 

(false) a new window is opened when the PDF viewer switches 

to another file;


pdfpagelayout 

(SinglePage) determines the page layout for the PDF viewer; 

other values are OneColumn (all pages in a continuous column), 

TwoColumnLeft (two columns, with odd numbers on the left), 

TwoColumnRight (two columns, with odd numbers on the right); 

pdfpagemode 

(None) determines initial display in the PDF viewer; other values 

are UseThumbs (show thumbnails), UseOutline (show bookmarks), 

FullScreen (no tool or menu bars, window is full screen); 

pdfpagescrop 

a set of four numbers determining the crop box when the the 

PDF file is opened, similar to a bounding box; 

pdfpagetransition 

(R) determines how the pages are changed; possible values are 

Split (like opening a curtain), Blinds (like Venetian blinds), Box 

(a rectangle opens from the center), Wipe (vertical line moves 

across page), Dissolve (old page dissolves into new), Glitter 

(variation on dissolve), R (replace); additional parameters can be 

added where appropriate: /Dm (with Split and Blinds) followed 

by /H (horizontal) or /V (vertical), /Di (with Wipe and Glitter) 

followed by 90, 180, 270 for the direction of motion, /M (with 

Split and Box) followed by /I (inwards), /O (outwards); see 

page 344 for a simplified interface to this feature; 

pdfstartpage 

(1) the first page number to be displayed when the PDF file is 

opened; 

pdfstartview 

(Fit) determines the initial view when the PDF file is opened; 

takes values of Fit (fit page to window), FitH (fit to width), FitV 

(fit to height), FitB (fit to bounding box), FitBH (fit to bounding 

box width), FitBV (fit to bounding box height), XYZ (no change); 

pdftoolbar 

(true) makes the tool bar in the PDF viewer visible; 

pdfview 

(XYZ) determines the page view when a link moves to a new page; 

takes same values as pdfstartview; 

pdfwindowui 

(true) makes the user interface elements (bookmarks, thumbnails) 

in the PDF viewer available; if false, they are permanently 

switched off;


plainpages 

(true) makes the anchors for the pages to be their Arabic form; if 

false, the page anchors are as the page numbers are formatted, 

highly recommended if the document contains both Roman and 

Arabic page numbers; 

raiselinks 

(false) with the hypertex driver, makes the border around links 

the actual height of their contents, rather than the current baseline 

separation; 

runbordercolor 

(0 .7 .7) color of the border around ‘run’ links; 

unicode 

(false) enables Unicode encoded PDF strings; 

urlbordercolor 

(0 1 1) color of the border around URL links; 

urlcolor 

(magenta) color of URL links when colorlinks given; 

verbose 

(false) issues many extra messages. 

Using a configuration file 

Very often a user will find that he or she needs the same set of options in 

almost every document. Rather than issuing that set again and again, it is 

possible to prepare a configuration file named hyperref.cfg containing 

a \hypersetup command with all the option settings that one normally 

wants. One effectively resets the default values. Any settings in the L AT E X 

source text itself will override those defaults. 

The hyperref.cfg file must be located somewhere where it will always 

be found by the L AT E X processing. It might very well be that a file of that 

name already exists, and will be read in in preference to yours. To check 

just which location is being used, examine the transcript.log file for the 

full path name, and then edit that version of hyperref.cfg. 

Linking commands 

The main purpose of the hyperref package is to add the automatic links 

between cross-referencing commands and tables of contents. However, 

it also enables access to PDF features with L AT E X-style commands. One 

of these is the ability to add manual links within the file and to other 

documents, locally or on the Web. The commands to achieve this are:


\hyperbaseurl{url pre} 

sets a prefix for all external URLs specified by other commands, 

so that the common part of a set of URLs need only be given 

once; this is the same as the baseurl option, but can be changed 

within the document; 

\href{url}{text} 

makes text into a link to the full URL given by url prefixed with 

the current base URL; special characters like # and & can be type 

in url directly; 

\hyperimage{url}{image url} 

inserts the image referenced by image url; 

\hyperdef{category}{name}{text} 

establishes text as a target (anchor) with name category.name; 

\hyperref{url}{category}{name}{text} 

makes text a link to url#category.name; 

an alternate syntax is \hyperref[key]{text} to make text an 

internal link to the location marked with \label{key}; 

\hypertarget{name}{text} 

establishes text as an internal target (anchor) with the name 

#name; 

\hyperlink{name}{text} 

makes text a link to the internal target #name. 

The cross-referencing commands \ref{key} and \pageref{key} (Section 

9.2.1) automatically create links to the corresponding \label{key} 

location. As pointed out above, one can also create a manual link with 

\hyperref[key]{text}. 

The hyperref package also provides starred versions \ref* and 

\pageref* which function as the regular ones but without creating a 

link. 

Adding Acrobat buttons 

An electronic document is active, as indicated by the links within it for 

jumping to other sections. But it is also possible to insert other types of 

‘actions’, in the form of buttons which, when clicked, do something, like 

flipping to the next page, opening another document, or even quitting. 

The hyperref package can add such functionality for the Acrobat 

viewers, by means of the command 

\Acrobatmenu{action}{text}


Table 10.2: Actions available from the Acrobat menus (after Goossens 

and Rahtz, 1999, page 48) 

Acrobat Menu Available options for \Acrobatmenu 

File Open, Close, Scan, Save, SaveAs, Optimizer:SaveAsOpt, 

Print, PageSetup, Quit 

File→Import ImportImage, ImportNotes, AcroForm:ImportFDF 

File→Export ExportNotes, AcroForm:ExportFDF 

File→DocProps GeneralInfo, OpenInfo, FontsInfo, SecurityInfo, 

Weblink:Base, AutoIndex:DocInfo 

Edit Undo, Cut, Copy, Paste, Clear, SelectAll, Ole:CopyFile, 

TouchUp:TextAttributes, TouchUp:FitTextToSelection, 

TouchUp:ShowLineMarkers, 

TouchUp:ShowCaptureSuspects, TouchUp:FindSuspect, 

Properties 

Edit→Prefs GeneralPrefs, NotePrefs, FullScreenPrefs,Weblink:Prefs, 

AcroSearch:Preferences(Windows)or, 

AcroSearch:Prefs(Mac), Cpt:Capture 

Edit→Search AcroSrch:Query, AcroSrch:Indexes, AcroSrch:Results, 

AcroSrch:Assist, AcroSrch:PrevDoc, AcroSrch:PrevHit, 

AcroSrch:NextHit, AcroSrch:NextDoc 

Edit→Fields AcroForm:Duplicate, AcroForm:TabOrder 

Document Cpt:CapturePages, AcroForm:Actions, CropPages, 

RotatePages, InsertPages, ExtractPages, ReplacePages, 

DeletePages, NewBookmark, SetBookmarkDest, 

CreateAllThumbs, DeleteAllThumbs 

View ActualSize, FitVisible, FitWidth, FitPage, ZoomTo, 

FullScreen, FirstPage, PrevPage, NextPage, LastPage, 

GoToPage, GoBack, GoForward, SinglePage, OneColumn, 

TwoColumns, ArticleThreads, PageOnly, ShowBookmarks, 

ShowThumbs 

Tools Hand, ZoomIn, ZoomOut, SelectText, SelectGraphics, Note, 

Link, Thread, AcroForm:Tool, Acro Movie:MoviePlayer, 

TouchUp:TextTool, Find, FindAgain, FindNextNote, 

CreateNotesFile 

Window ShowHideToolBar, ShowHideMenuBar, ShowHideClipboard, 

Cascade, TileHorizontal, TileVertical, CloseAll 

Help HelpUserGuide, HelpTutorial, HelpExchange, HelpScan, 

HelpCapture, HelpPDFWriter, HelpDistiller, HelpSearch, 

HelpCatalog, HelpReader,Weblink:Home 

Help(Windows) About


where text becomes a button to execute action when clicked. Of course, 

text can be more than mere text, it is often a graphics that really does 

look like a button. 

The actions available are those provided by the Acrobat and Acrobat 

Reader programs. These are not documented by Adobe, but are listed in 

Table 10.2 on the facing page, which is taken from Goossens and Rahtz 

(1999), who credit Merz (1998) for having figured them out. These actions 

are therefore specific to these viewers only, and may change with later 

versions. They may all be executed by the viewers themselves, by means 

of pull-down menus in the menu bar, indicated in the left column of 

Table 10.2. 

The main purpose of this command is to set up navigational controls 

on each page of a document that is meant essentially for screen viewing. 

The regular tool bar may be turned off, which is why the document 

itself must provide it. This is best put into a footline, and that is most 

conveniently done with the fancyhdr package (Section 3.2.2). 

\documentclass[a4paper]{article} 

\usepackage{fancyhdr} 

\usepackage[pdftex,colorlinks, 

pdftoolbar=false,pdfmenubar=false, 

pdfwindowui=false, 

pdfpagemode=FullScreen, 

]{hyperref} 

\pagestyle{fancy} 

\cfoot{\navbar} 

\newcommand{\navbar}{% 

\Acrobatmenu{FirstPage}{$\Leftarrow$} 

\Acrobatmenu{PrevPage}{$\leftarrow$} 

\Acrobatmenu{NextPage}{$\rightarrow$} 

\Acrobatmenu{LastPage}{$\Rightarrow$} 

\Acrobatmenu{Close}{$\bullet$} 

} 

The above produces a footline containing 

⇐ ← → ⇒ • 

(by default in red) as clickable buttons to jump to the first page, to the 

previous page, to the next page, to the last page, respectively. And clicking 

the bullet • closes the document. Of course, it looks even better if icons 

are made up and inserted with \includegraphics.

11 

Multilingual LAT E X 

The original versions of T E X and L AT E X were set up for the English language, 

American variant. The built-in hyphenation patterns were for English 

only, and several English words such as ‘Figure’ and ‘Bibliography’ were 

included explicitly in certain commands. This in fact violates the rules 

of good programming which forbid doing anything explicitly. In Europe, 

L AT E X users quickly made customizations for their particular languages. 

The adaptation for German, german.sty, is a set of macros from many 

contributors collected together by H. Partl of the Technical University of 

Vienna, and has become the standard for the German-speaking T E X Users 

Group (DANTE). It contains some facilities for other languages, including 

French and English. 

The key to the german package and other language adaptations is 

the fact that the standard L AT E X document styles (and later classes for 

L AT E X 2ε) had been altered so that explicit English words in the output were 

replaced by reprogrammable command names. These names had become 

standardized among European users for application to all languages. The 

modified L AT E X version by J. Schrod from Darmstadt was known as IL AT E X, 

for International L AT E X. These names and their standard English values 

are to be found on page 459. 

As of December 1, 1991, these naming features became part of the L AT E X 

standard. They represent good programming practice, since they allow 

even English-speaking users to change certain titles easily, for example 

‘Abstract’ into ‘Summary’ or ‘Contents’ into ‘Table of Contents’. Naturally, 

they were taken over, and extended upon, by L AT E X 2ε. 

Another important feature for multilingual usage is the T E X \language 

counter which allows more than one set of hyphenation patterns to be 

stored in the format with initex (Section B.1.3). Different patterns are 

activated by setting \language to the appropriate number. This is now 

standard for all T E X version over 3.0, which should now be universal. 

Individual language packages such as german maintained by Bernd 

Raichle and french by Bernard Gaulle have now been superseded to a 

large extent today by the universal multilanguage system: babel. 

251

252 Chapter 11. Multilingual L AT E X 

11.1 The babel system 

The babel system has been developed by Johannes Braams at the Dutch 

PTT Research Laboratories. Its main purpose is to provide a standard 

means of switching languages with a flexible method for loading the 

hyphenation patterns. It has been written to work with both T E X and 

L AT E X. 

The three main requirements for a single language adaptation are: 

translation of the explicit English words, special commands to simplify 

certain features of that language, and selection of the appropriate hyphenation 

patterns. To these babel adds: removal of the special commands 

and testing for the current language. The language-specific files must 

conform to the babel selection command structure. The concept of ‘dialects’ 

is also included, that is, two ‘languages’ that share a common set 

of hyphenation patterns. Examples are German and Austrian as well as 

British and US English, which only differ in the form of the dates. 

At present, the babel package contains definition files for the following 

languages (more can be added at any time): 

Afrikaans, Bahasa, Basque, Breton, Bulgarian, Catalan, Croatian, 

Czech, Danish, Dutch, English, Esperanto, Estonian, Finnish, French, 

Galician, German, Greek, Hebrew, Hungarian, Icelandic, Irish Gaelic, 

Italian, Latin, Lower Sorbian, North Sami, Norwegian, Polish, Portuguese, 

Romanian, Russian, Scottish Gaelic, Spanish, Slovakian, 

Slovenian, Swedish, Serbian, Turkish, Ukrainian, Upper Sorbian, 

Welsh 

The necessary definition files are delivered with the installation. The 

hyphenation patterns for the various languages must be obtained from 

another source, however. For most of these files, there are no special 

commands for accents or punctuation: they only redefine the naming 

commands for those words that normally print English text. 

The german.sty package has been a great inspiration for babel. However, 

because of the name conflict, it was necessary to create a new 

germanb.sty (and ngermanb.sty for the new German grammar rules) for 

babel applications, which contains the same set of special commands, 

but with a few babel additions. Similarly, the babel French file had to be 

named frenchb.sty to distinguish it from the existing french.sty. 

To exploit the babel features most fully, one should produce a L AT E X 

format (Section B.1.3) that incorporates it directly. To this end, the 

hyphen.cfg supplied with babel must be used with the language.dat 

file configured for the local languages and hyphenation patterns that are 

to be pre-loaded. Only in this way can one conveniently get the right word 

division for each language. 

Alternatively, one can use a normal L AT E X format and load babel as 

a regular package, as illustrated below. The switching of hyphenation

patterns will then be questionable. 

The babel files 

11.1. The babel system 253 

The babel installation provides the following files for its implementation: 

babel.def contains some of the basic macros for running babel, and 

must be loaded by the first language file read in; the remaining 

macros are either already in the format itself (loaded from 

hyphen.cfg) or read in from switch.def; babel.def determines 

which is the case and reacts accordingly; 

switch.def contains the additional babel macros, to be read in if these 

macros are not already stored in the L AT E X format; 

hyphen.cfg contains the same macro definitions as in switch.def, plus 

some more that are to be run by initex; if this file is available during 

the creation of the L AT E X format (Section B.1.3), these macros are 

built into that format and file switch.def is not necessary at run 

time; 

babel.sty is a master package that loads the language files specified as 

options; 

language.dat contains a list of languages and the file names of their 

hyphenation patterns; this file is read in by hyphen.cfg during 

the initex run; it is the only file that may (must!) be edited for the 

particular installation (Section 11.2); 

esperant.ldf... the language definition files. 

esperant.sty... compatibility files that read in the.ldf files. 

Invoking babel 

The babel package is loaded with the desired languages as options: 

\usepackage[english,esperanto]{babel} 

Alternatively, the language names may be used as global options, something 

that is recommended if there are other packages that take the 

languages as options, as 

\documentclass[english,esperanto]{article} 

\usepackage{babel,varioref} 

In both cases, the last named language is the one that is immediately 

active. 

The recognized language names that may be used as options are:


afrikaans 

bahasa 

basque 

breton 

bulgarian 

catalan 

croatian 

czech 

danish 

dutch 

english 

=USenglish 

=american 

=UKenglish 

=british 

=canadian 

esperanto 

estonian 

finnish 

french 

=francais 

=canadien 

=acadian 

galician 

german 

=germanb 

=ngerman 

=austrian 

=naustrian 

greek 

=polutonikogreek 

hebrew 

magyar 

=hungarian 

icelandic 

irish 

italian 

latin 

lowersorbian 

samin 

norsk 

=nynorsk 

polish 

portuges 

=portuguese 

=brazilian 

=brazil 

romanian 

russian 

scottish 

spanish 

slovak 

slovene 

swedish 

serbian 

turkish 

ukrainian 

uppersorbian 

welsh 

Those names preceded by = are synonyms for the language option 

above them. (The equals sign is not part of the option name.) For 

example, both hungarian and magyar load the file magyar.ldf. 

If the babel package is loaded, all it does is read in the specified language 

packages; they in turn read in babel.def and possibly switch.def 

(if its macros are not already in the format). 

Language switching commands 

The normal user needs to know very little about the babel internal operations. 

The new high-level user commands are: 

\selectlanguage{language} 

\begin{otherlanguage}{language} text \end{otherlanguage} 

\begin{otherlanguage*}{language} text \end{otherlanguage*} 

\foreignlanguage{language}{text} 

\iflanguage{language}{yes text}{no text} 

The \selectlanguage command and the otherlanguage environment 

are two ways to switch to another language, with all its features and translations. 

The \foreignlanguage command and the otherlanguage* environment 

also switch to language but without the translations or date 

changes; they are meant for setting short sections of text in a language 

different from the regular one. 

The \iflanguage executes yes text if language is current otherwise 

no text. 

In addition, \languagename contains the name of the currently selected 

language. 

Of course, each language definition file can also have its own special 

commands to assist typing accented and special characters. For example,

11.1. The babel system 255 

in German, one may type "a as a shorthand for \"a. To find out what 

special commands are available for each language, and any other special 

considerations, one must read the documentation for that language. 

Documentation should be found in the texmf\doc\tex\generic\babel 

directory, or you can process the desired language.dtx file in the texmf\ 

source\tex\generic\babel directory. (These paths might be different 

on your system.) 

Contents of a language definition file 

Although one does not normally need to know anything about how the 

switching mechanism functions, we will outline it here for the interested 

user. 

Two additional internal babel commands are: 

\addlanguage{lang num} and 

\adddialect{lang num 1}{lang num 2} 

where lang num is an internal language number for specifying the set of 

hyphenation patterns. The command \addlanguage sets its argument 

equal to the next available language number. The form of lang num used 

in babel is \l@language; for example, \l@english. This command is 

executed on those language names listed in language.dat during the 

initex run. The command \adddialect sets the first argument equal to 

the second so that the two languages make use of the same set of hyphenation 

patterns. For example, in english.ldf there is the command 

\adddialect{\l@american}{\l@english}. 

A language definition file must provide four commands: 

\captionslang to redefine the naming commands like \tablename 

(see Section D.4.1); 

\datelang to define \today (see Section D.4.2); 

\extraslang to define any language-specific commands; 

\noextraslang to remove the language-specific commands. 

If the language lang is not one of those with prestored hyphenation 

patterns in the current format file (that is, if \l@lang is undefined), it is 

set to be a ‘dialect’ of language number 0. 

Finally, the language definition file calls \selectlanguage{lang} to 

activate that language. This is accomplished by 

• calling \language\l@lang to select the set of hyphenation patterns, 

• invoking \originalTeX in order to remove any existing languagespecific 

commands, 

• activating \captionslang, \datelang, and \extraslang to invoke 

the language-specific names, date, and commands,


• redefining \originalTeX to be \noextraslang so that the next 

language switch will remove those features specific to lang. 

11.2 Contents of the language.dat file 

As has been mentioned before, with T E X version 3.0 and later it is possible 

to store more than one set of hyphenation patterns in the format file. 

The counter \language is used to switch between them by setting it to a 

different number. 

However, there is no standard to dictate which languages belong to 

which numbers. If a package were to assume a certain ordering, it would 

most certainly function incorrectly at installations other than the one 

for which it was designed. The babel system invokes a more reliable 

procedure. 

During the production of the format by initex (Section B.1.3), the file 

hyphen.cfg is input, which in turn reads in language.dat, the only file 

to be tailored to the local installation. This file contains a list of languages 

to add as well as the name of the file with the hyphenation patterns and 

the name of any additional file to be included. It also indicates names of 

dialects that use the same hyphenations by prefixing the name with an 

equals sign. For example, if language.dat contains 

=USenglish 

american ushyphen.tex 

english ukhyphen.tex 

=UKenglish 

=british 

french frhyphen.tex 

german dehypht.tex 

ngerman dehyphn.tex 

the hyphenation patterns stored in files ushyphen.tex, ukhyphen.tex, 

frhyphen.tex, dehypht.tex, and dehyphn.tex are loaded under the 

language numbers 0 to 4 respectively, and \l@american, \l@english, 

\l@french, \@german, and \l@ngerman are defined as numbers 0 through 

4 for use with the \selectlanguage command. Languages \l@USenglish 

and \l@UKenglish are synonyms for the current language, in this case 0 

and 1, respectively. Similarly \l@british is identical to \l@UKenglish.

12 

Math Extensions with 

A M S-LAT E X 

The American Mathematical Society, A M S, has supported the development 

and usage of T E X since its first release. It in fact owns the T E X logo 

as a registered trademark. 

Shortly after T E X 82 became available, the A M S produced a macro 

package for generating a special format amstex, described in The Joy 

of T E X by Spivak (1986). The macros in this A M S-T E X complement the 

mathematical typesetting features of Plain T E X by adding additional ones 

and simplifying others. 

However, A M S-T E X is not a documentation preparation system like 

L AT E X, describing the logical layout of a document by means of markup 

commands, but rather is simply an extension to Plain T E X. 

The great popularity of T E X as a text formatting program is due primarily 

to the availability of L AT E X as a user-friendly interface to the underlying 

T E X machinery. Many authors therefore asked the A M S to provide L AT E X 

with the same mathematical features as in A M S-T E X. The A M S-L AT E X 

project was thus launched in 1987, with version 1.0 completed three 

years later by Frank Mittelbach and Rainer Schöpf, together with Michael 

Downes of the A M S. A M S-L AT E X has been fully converted to L AT E X 2ε with 

version 1.2 in 1996. 

The A M S-L AT E X collection consists of three parts: packages for extending 

mathematical typesetting, extra classes for articles and books 

published by the A M S, and supplemental fonts for additional symbols, 

math alphabets, and Cyrillic fonts. 

In the next sections, we give an overview of the math extensions available 

in the package amsmath; a more detailed user’s manual is delivered 

with the collection, in the file amsldoc.tex. How to invoke the additional 

fonts is described below in Section 12.4, or in the A M S-supplied manual 

found in the file amsfndoc.tex. The extra classes will not be described in 

this book; A M S authors should refer to the instructions in the document 

instr-l.tex provided with A M S-L AT E X. 

257

258 Chapter 12. Math Extensions with A M S-L AT E X 

12.1 Invoking A M S-L AT E X 

If the \documentclass statement at the beginning of the LAT Package: 

EX document 

amsmath selects one of the AMS classes amsbook, amsart, or amsproc, then most 

of the AMS-LAT EX features are loaded automatically. These features may 

still be employed with other classes by including the main extension 

package amsmath by means of 

\usepackage[options]{amsmath} 

in the document’s preamble. The list of allowable options is described 

below in Section 12.2.8. 

The amsmath package defines many of the new math typesetting features 

itself, but it also loads a number of other packages from the A M S- 

L AT E X collection that contain further extensions, such as amsopn, amstext, 

and amsbsy. These packages could be loaded separately without amsmath 

if only their limited features are wanted. On the other hand, the packages 

amscd and amsthm are not included in amsmath and must be loaded explicitly 

if their features are desired. Simply add their names to the list of 

packages in \usepackage. 

The following Section 12.2 describes the new commands and environments 

made available with amsmath and its associated packages. We call 

these the standard features of A M S-L AT E X. Further extensions added by 

other packages are explained afterwards. 

The examples in this chapter employ the user-defined commands \mi, \me, 

! and \dif defined in Section 5.4.10 for printing upright i, e, and d in math mode. 

12.2 Standard features of A M S-L AT E X 

This relatively long section presents those standard features of A M S-L AT E X 

that are activated by loading the amsmath package, or by selecting one of 

the A M S classes amsart, amsbook, or amsproc. 

12.2.1 Additional font switching commands 

Standard LAT Package: 

EX provides the math alphabet commands \mathcal, \mathrm, 

amsbsy \mathbf, \mathsf, \mathit, \mathtt for changing fonts within math 

mode (Section 5.4.2). With AMS-LAT EX, one may also use the commands 

\boldsymbol{symbol} and \pmb{symbol} 

to print symbol in a bold face, provided there is an appropriate bold 

font for it. Whereas the command \mathbf sets only Latin letters, 

numbers, and Greek upper case letters in bold, these commands also 

affect math symbols and Greek lower case letters. Compare the result 

of $\mathbf{\nabla\times V\,d\sigma}$ (∇ × V dσ ) with that

Package: 

amstext 

12.2. Standard features of A M S-L AT E X 259 

of $\boldsymbol{\nabla\times V\,d\sigma}$ (∇ × V dσ). With the 

standard LAT EX command \mathbf, only the letters V and d appear bold, 

while the other characters remain in normal weight. Not only that, these 

letters are upright, and not italic as required by international standards. 

With \boldsymbol, all symbols are bold and italic. 

Those symbols for which no bold face font exists will remain in normal 

weight with the \boldsymbol command. Such symbols are, for example, 

those that come in two sizes (Section 5.3.7) like , , , and so on. The 

command \pmb (poor man’s bold) simulates a bold face even for these 

symbols by printing them several times 

slightly 

 

displaced. The result of 

$\pmb{\sum\;\int\;\bigcup}$ is . 

These commands are defined in the package amsbsy, which may be 

loaded separately without amsmath. 

A short piece of normal text can be given within a formula with 

\text{short text} 

In contrast to \mbox{short text} for standard L AT E X, the \text command 

switches font size correctly when used as superscripts and subscripts. 

Thus .._{\text{Word}} sets Word lower and changes to \scriptstyle 

font size. To achieve the same effect in standard L AT E X, the font size must 

be given explicitly, as .._{\mbox{\scriptstyle Word}}. Furthermore, 

the name \text is more precise than \mbox. 

This command is defined in the package amstext. Like amsbsy, it may 

be loaded on its own without amsmath if none of the other A M S-L AT E X 

features are wanted. 

Another command to insert normal text inside a displayed equation is 

\intertext{insert text} 

The insert text is inserted as a left-justified line of text between those 

of the formula. The alignment of the formula lines remains unaffected, 

something that is not guaranteed if one closes the displayed equation, 

inserts text and reopens the equation. For example: 

(x + iy)(x − iy) = x 2 − ixy + ixy − i 2 y 2 

On the other hand 

= y 2 + y 2 

since i 2 = −1 is true. 

(x + iy) 2 = x 2 + 2ixy + i 2 y 2 = x 2 + 2ixy − y 2 

(x − iy) 2 = x 2 − 2ixy + i 2 y 2 = x 2 − 2ixy − y 2 

\begin{align*} 

(x+\mi y)(x-\mi y) & = xˆ2 - \mi xy + \mi xy - \mi ˆ2yˆ2\\ 

& = yˆ2 + yˆ2\quad\text{since}\quad \mi ˆ2=-1 

\quad\text{is true.}\\


\intertext{On the other hand} 

(x +\mi y)ˆ2 & = xˆ2 + 2\mi xy +\mi ˆ2yˆ2 = xˆ2 + 2\mi xy-yˆ2\\ 

(x -\mi y)ˆ2 & = xˆ2 - 2\mi xy +\mi ˆ2yˆ2 = xˆ2 - 2\mi xy-yˆ2 

\end{align*} 

The alignment of the lines on the first equals sign is maintained through 

the interruption with the line of text ‘On the other hand’. The environment 

align (Section 12.2.6) is one of several new ones provided by A M S-L AT E X to 

replace the standard L AT E X environment eqnarray. Note that \intertext 

may only be issued immediately after the \\ for starting a new line. 

12.2.2 Multiple mathematical symbols 

Mathematical formulas often require the same symbol to appear several 

times, such as multiple integral signs, or symbols and arrows to be stacked 

above or below a mathematical expression. Usually the distances between 

these symbols depends on the mathematical meaning, something that 

L AT E X cannot automatically recognize. A M S-L AT E X provides a number of 

structures to help the user find the right spacing without tedious trial and 

error. 

Multiple integrals 

With AMS-LAT EX, the commands \iint, \iiint, \iiiint, and \idotsint 

output multiple integrals, with upper and lower limits being added in the 

usual manner. In text formulas, they are printed as , , , ··· , 

while in displayed formulas, they appear as 

 

f (x, y) dS \[ \iint\limits_S f(x,y)\,\dif S \] 

 

S 

 

f (x, y, z) dV 

V 

 

f (x, y, z, t) dG 

G 

 

· · · 

U 

f (x1, . . . , xk) dU 

\[ \iiint\limits_V f(x,y,z)\,\dif V\] 

\[ \iiiint\limits_G f(x,y,z,t) 

\,\dif G \] 

\[ \idotsint\limits_U 

f(x_1,\ldots,x_k)\,\dif U \] 

whereby the command \limits may be left off if the option intlimits 

has been specified when amsmath was loaded (see Section 12.2.8).

Multiline limits 


One often needs multiline limits or indices for summations, as in the 

examples below. 

 

∆p1p2···pn−k 

q1q2···qn−k 

k0,k1,...≥0 

k0+k1+···=0 

a0k0a1k1 · · · 

A M S-L AT E X provides a command \substack for this purpose, with syntax 

\substack{1st line\\2nd line\\...\\ last line} 

where the command must immediately follow the ˆ or _ shifting commands, 

and be entirely enclosed in curly braces { }. The index for the 

left-hand example above was generated with 

\Delta_{\substack{p_1p_2\cdots p_{n-k}\\q_1q_2\cdots q_{n-k}}} 

The lines of text printed by \substack are centered horizontally, as 

is apparent in the right-hand example above. This was produced with 

\[ \sum_{\substack{k_0,k_1,\ldots\ge0\\ k_0+\k_1+ 

\cdots=0}}a_{0k_0} a_{1k_1}\cdots \] 

On the other hand, the environment subarray offers more control over 

the horizontal alignment: 

\begin{subarray}{pos} 1st line\\2nd line\\...\\ last line 

\end{subarray} 

The argument pos may be c for centered, or l for left-justified lines. 

 

i∈Λ 

i


\sideset{pre}{post}\symbol 

where pre and post are the superscripts and subscript commands to be 

added before and after the \symbol, respectively. They must contain 

raising and lowering operators ˆ and _. 

The product symbol is given daggers above and asterisks 

below with \sideset{_\dagˆ*}{_\dagˆ*}\prod, as shown 

at the right. 

Additional shifting commands are 

∗∗ 

\overset{char}{\symbol} and \underset{char}{\symbol} 

which place the arbitrary char above or below the \symbol in the size 

appropriate for superscripts and subscripts. Thus $\overset{*}{X}$ 

produces ∗ 

X and $\underset{*}{X}$ yields X. ∗ 

Extended arrows 

The amsmath package provides a number of commands to produce extra 

long arrows for combining with mathematical expressions. The commands 

\overleftarrow{expr} \underleftarrow{expr} 

\overrightarrow{expr} \underrightarrow{expr} 

\overleftrightarrow{expr} \underleftrightarrow{expr} 

produce lengthened arrows pointing left and right, as well as double 

arrows above and below the mathematical expression expr. 

−−−→ 

ABCD = AB + BC + CD 

−→ −→ −→ 

←−−− 

ABCD = DC + CB + BA 

←− ←− ←− 

←−−−→ 

ABCD = DCAB 

←−−−→ 


\overrightarrow{ABCD} & = & 

\underrightarrow{AB} + 

\underrightarrow{BC} + 

\underrightarrow{CD} \\ 

\overleftarrow{ABCD} & = & 

\underleftarrow{DC} ... 

\end{eqnarray*} 

The lower arrows are actually too close to the expression, as is seen on 

the right-hand side of the last example. This should also be the case for 

all the other examples with a lower arrow, except that we have added a 

strut to push them down somewhat: 

\underrightarrow{\rule[-2pt]{0pt}{2pt}AB} ... and 

\underleftarrow{\rule[-2pt]{0pt}{2pt}DC} ... 

A smaller font size will be used for the arrows when they appear in 

exponents, indices, superscripts, and subscripts: 

† 

†

−−→ r dϕ = 2πr 

0.2π 


\int_{\overrightarrow{0.2\pi}} 

r\,d\varphi = 2\pi r 

There are two more commands for horizontal arrows of variable length: 

\xleftarrow[below]{above} \xrightarrow[below]{above} 

which place the mandatory above in superscript size over the arrow, and 

the optional below in subscript size beneath it. 

A n+µ−1 

←−−−− B n±i−1 

−−−→ C 

T 

\[ A \xleftarrow{n+\mu-1} B 

\xrightarrow[T]{n\pm i-1} C \] 

The package amscd (Section 12.3.2) offers further possibilities for 

combining arrows and text. 

Stacked accents 

Attempting to place multiple accents over a character with standard L AT E X, 

say with $\hat{\hat{A}}$, results in a misplacement: ˆ Â. The same 

applies to the other mathematical accents commands \check, \breve, 

\acute, \grave, \tilde, \bar, \vec, \dot, and \ddot (Section 5.3.9). 

The amsmath package provides a set of math accent commands with the 

same names but capitalized, 

\Hat \Breve \Grave \Bar \Dot 

\Check \Acute \Tilde \Vec \Ddot 

which can be safely combined with one another for the expected results: 

$\Hat{\Hat{A}}$, $\Breve{\Bar{B}}$ and $\Tilde{\Tilde{C}}$ 

produce Â, ˘¯B, and ˜C. 

Three or four dots in a row over a symbol are often used for time 

derivatives of third and fourth order. They can be placed with the commands 

\dddot{sym} and \ddddot{sym} 

In this way ... 

u and .... 

u are produced with $\dddot{u}$ and $\ddddot{u}$. 

Continuation dots 

In standard L AT E X, one has the commands \ldots and \cdots for printing 

three continuation dots, either on the baseline or raised to the center of 

the line. A M S-L AT E X offers a number of additional possibilities. The most 

general of these is the \dots command which adjusts the vertical height 

according to the symbol that follows it. If this is an equals sign or binary 

operator, such as + or −, the dots are raised, as with \cdots, otherwise 

they are on the baseline, as with \ldots.


$a_0+a_1+\dots+a_n$ =⇒ a0 + a1 + · · · + an 

$a_0,a_1,\dots,a_n$ =⇒ a0, a1, . . . , an 

If the continuation dots come at the end of a formula, there is no following 

symbol to determine the height of the dots. In this case, one must 

manually indicate the height by means of one of the commands \dotsc 

(comma), \dotsb (binary), \dotsm (multiplication), or \dotsi (integral). 

The \dotsc is to be used with commas, so $A_1,A_2,\dotsc$ produces 

A1, A2, . . . ; the \dotsb sets them for a binary operator, thus 

$A_1+A_2+\dotsb$ yields A1 +A2 +· · · ; the command \dotsm is actually 

identical to \dotsb, but it is logically meant to be applied to multiplication, 

$A_1A_2\dotsm$ makes A1A2 · · · ; finally \dotsi places the dots at 

the mean height of an adjacent integral sign. 

12.2.3 Fractions 

\[ \int_{A_1}\int_{A_2}\dotsi \] 

The T E X fraction commands \atop, \choose, and others may be allowed 

in standard L AT E X, but not in A M S-L AT E X. With the amsmath package, only 

those fraction commands described here may be used. 

Basic fraction commands 

In addition to the regular L AT E X command \frac{over}{under}, A M S- 

L AT E X provides the commands \tfrac and \dfrac with the same syntax. 

These are effectively the same as \frac but with the font set to 

\textstyle or \displaystyle, respectively (Section 5.5.2). We demonstrate 

their effects with some examples from the A M S-L AT E X user’s manual. 

\[ \frac{1}{k}\log_2 c(f)\qquad 

\tfrac{1}{k}\log_2 c(f)\qquad 

\sqrt{\frac{1}{k}\log_2 c(f)}\qquad 

\sqrt{\dfrac{1}{k}\log_2 c(f)} \] 

1 

k log 2 c(f ) 

Binomial expressions 

1 

k log 2 c(f ) 

 

1 

k log 2 c(f ) 

 

A1 

 

A2 

· · · 

 

1 

k log 2 c(f ) 

A binomial expression looks something like a fraction, but is enclosed in 

round parentheses and is missing the horizontal rule. The basic command 

in the amsmath package is 

\binom{over}{under}


which functions in the same way as \frac and the other fraction commands. 

 

\[ \binom{n+1}{k} = \binom{n}{k} n + 1 n n 

= + 

+ \binom{n}{k-1 \] 

k k k − 1 

Similarly there are the commands \tbinom and \dbinom analogous to 

\tfrac and \dfrac. 

User-defined fractions 

The amsmath package provides a powerful tool for defining fraction-like 

structures: 

\genfrac{left brk}{right brk}{thickness}{mathsize}{over}{under} 

where left brk and right brk are the parenthesis characters on the left and 

right, thickness is the thickness of the horizontal line, and {mathsize} is 

a number 0–3 representing the math sizes \displaystyle, \textstyle, 

\scriptstyle, and \scriptscriptstyle, respectively. The last two 

arguments, over and under, are the texts in the two parts of the fraction, 

the same arguments as in \frac and \binom. 

If the thickness is left blank, the standard thickness for L AT E X fractions 

is used. If mathsize is empty, the font size is determined automatically 

by the normal rules in Section 5.5.2. 

Rather than repeating \genfrac with the same first four arguments 

time and again, one should define new fraction commands with it. For 

example, the following definitions are given in amsmath.sty: 

\newcommand{\frac}[2]{\genfrac{}{}{}{}{#1}{#2}} 

\newcommand{\dfrac}[2]{\genfrac{}{}{}{0}{#1}{#2}} 

\newcommand{\tfrac}[2]{\genfrac{}{}{}{1}{#1}{#2}} 

\newcommand{\binom}[2]{\genfrac{(}{)}{0pt}{}{#1}{#2} 

As a further example, consider the redefinition of the command \frac 

\renewcommand{\frac}[3][]{\genfrac{}{}{#1}{}{#2}{#3}} 

in which the line thickness is now an optional first argument; without this 

optional argument, the command behaves as normal. Thus 

 

\[ \binom{n}{m} = 

yields n n! 

= 

\frac[2pt]{n!}{M!(n-m)!} \] 

m 

M!(n − m)! 

Continued fractions 

Continued fractions can be made in A M S-L AT E X with the command 

\cfrac[pos]{over}{under} 

whereby the denominator under may contain further \cfrac commands.


produces 

\[ a_0 + \cfrac{1}{a_1 + \cfrac{1}{a_2 + 

\cfrac{1}{a_3 + \cfrac{1}{a_4 + 

\dotsb }}}} \] 

1 

a0 + 

1 

a1 + 

1 

a2 + 

1 

a3 + 

a4 + · · · 

If the optional argument pos is missing, the numerator over is centered 

on the horizontal rule; otherwise it may take values of l or r to left or 

right justify the numerator. 

12.2.4 Matrices 

Standard L AT E X possesses the environment array for producing arrays 

and matrices. The amsmath package provides the additional environments 

pmatrix, bmatrix, Bmatrix, vmatrix, and Vmatrix, which automatically 

add the enclosing braces (), [], {}, | | and around the array, and in the 

right size. For completeness, there is also a matrix environment with no 

braces. 

In contrast to the standard array environment (Section 4.8.1), these 

matrix environments do not require an explicit column specification as 

argument. By default, up to 10 centered columns may be used without 

any argument. (This maximum number may be changed by giving the 

special counter MaxMatrixCols a new value with either \setcounter 

or \addtocounter.) Otherwise the matrix environments are used in the 

same way as the array environment. 

The following example is taken from Section 5.4.3 on page 134 and is 

recast here using A M S-L AT E X constructs. 

(1,2,...,n) 

p1


\hdotsfor[2.0]{4}\\ 

a_{q_kq_1} & a_{q_kq_2} & \dots & a_{q_kq_k} 

\end{vmatrix} \] 

Comparing this input text with that on page 134, one sees that it 

is simpler and easier to follow. The only new command used here is 

\hdotsfor which has the syntax 

\hdotsfor[stretch]{n} 

and which prints a continuous line of dots through n columns. The 

optional argument stretch is a multiplicative number to increase the dot 

density, being 1.0 by default. 

\[ \begin{matrix} a & b & c & d & e\\ 

x & \hdotsfor{3} & z 

\end{matrix} \] 

a b c d e 

x . . . . . . . . . z 

Compare the standard dot spacing above with that from \hdotsfor[2.0] 

in the previous example. 

The initial letter of each of the xmatrix environments indicates the 

type of braces that enclose it: pmatrix for (round) parentheses, bmatrix 

for (square) brackets, Bmatrix for (curly) braces, vmatrix for vertical 

lines, and Vmatrix for double vertical lines. They appear as 

r s t 

u v w 

x y z 

⎧ 

⎪⎨ r s 

⎫ 

t ⎪⎬ 

u 

⎪⎩ 

x 

v 

y 

w 

⎪⎭ 

z 

⎛ 

r 

⎜ 

⎝u 

s 

v 

⎞ 

t 

⎟ 

w⎠ 

x y z 

 

 

r 

 

u 

 

x 

s 

v 

y 

 

t 

 

 

w 

 

z 

where each matrix has been produced with 

⎡ 

r 

⎢ 

⎣u 

s 

v 

⎤ 

t 

⎥ 

w⎦ 

x y z 

 

 

r 

 

u 

 

x 

s 

v 

y 

 

t 

 

 

w 

 

z 

\[ \begin{xmatrix} r & s & t\\ u & v & w\\ x & y & z 

\end{xmatrix} \] 

where xmatrix is set to matrix, pmatrix, bmatrix, Bmatrix, vmatrix, 

and Vmatrix one after the other. 

To generate a small array within a text formula, one can apply the 

smallmatrix environment. In this way 

a b c 

e m r can be made with 

$ \bigl( \begin{smallmatrix} a & b & c\\ e & m & r 

\end{smallmatrix} \bigr) $ 

12.2.5 User extensions and fine adjustments 

Function names 

Standard LAT Package: 

EX recognizes a number of preprogrammed function names 

amsopn (Section 5.3.8) that are printed in math made by placing a backslash in


front of that name: arccos, arcsin, arctan, arg, cos, cosh, cot, coth, csc, 

deg, det, dim, exp, gcd, hom, inf, ker, lg, lim, liminf, limsup, ln, log, max, 

min, Pr, sec, sin, sinh, sup, tan, tanh. Not only do these names appear in 

an upright font, as is required for function names, but the spacing with 

adjacent parts of the mathematical expression is adjusted automatically. 

A M S-L AT E X provides some more function names, as variations on the 

standard \lim name: 

\varlimsup lim \varinjlim lim −→ 

\varliminf lim \varprojlim lim 

←− 

These functions may take on limits with the raising and lowering operators 

ˆ and _; for example \varliminf_{n\to\infty} for lim n→∞ . 

It is also possible to define new function names with the same font 

and spacing properties as the predefined ones. The command 

\DeclareMathOperator{\cmd}{name} 

which may only be issued in the preamble, before \begin{document}, 

defines a command \cmd that prints the function name name. For 

example, to define a function name \doit, give 

\DeclareMathOperator{\doit}{doit} 

and then $A=3\doitˆ2(B)$ yields A = 3 doit 2 (B). Note that superscripts 

and subscripts are printed beside the operator; if they are to be printed 

as limits, that is, above and below the operator in displayed math mode, 

use the *-form to define them. For example: 

\DeclareMathOperator*{\Lim}{lim} 

\[ \Lim_{n\to-\infty}ˆ{n\to+\infty}\] 

n→+∞ 

lim 

n→−∞ 

The name text need not be identical to the command name. In particular, 

it may contain special characters not allowed in command names. 

Modulo expressions are printed in standard L AT E X with \bmod and 

\pmod commands, and are complemented in A M S-L AT E X by \mod and 

\pod. The possibilities are: 

z ≡ x + y mod n 2 

z ≡ x + y (mod n 2 ) 

z ≡ x + y mod n 2 

z ≡ x + y (n 2 ) 

z \equiv x+y \bmod{nˆ2} 

z \equiv x+y \pmod{nˆ2} 

z \equiv x+y \mod{nˆ2} 

z \equiv x+y \pod{nˆ2} 

The automatic parentheses are missing with \mod, while with \pod the 

name ‘mod’ is omitted. Furthermore, \pmod is redefined for text formulas 

to reduce the preceding space: $y\pmod{a+b}$: y (mod a + b). 

The \DeclareMathOperator command and additional function names 

are defined in the amsopn package, which may be loaded on its own without 

amsmath.

Fine-tuning roots 


The positioning of an index to a root sign is not always ideal under 

standard L AT E X. In β√ k, for example, the β could be somewhat higher and 

shifted slightly to the right. The A M S-L AT E X commands 

\leftroot{shift} \uproot{shift} 

cause such manual displacements, where shift is a number specifying 

the size in small, internal units. Negative numbers represent a shift in 

the opposite direction. Compare the above standard result with that of 

$\sqrt[\leftroot{-1}\uproot{3}\beta]{k}$: β√ k. 

The size of the root sign depends on its contents. If they hang below 

the baseline, the root sign extends lower down than for contents that have 

no depth. Note the differences between √ x, √ y, and √ z. Some publishers 

want all root signs be at the same height, as √ x + √ y + √ z. This is accomplished 

with the T E X command \smash which places its argument in a box 

of zero height and depth. The A M S-L AT E X version of this command allows 

an optional argument b or t to zero only the depth or height, respectively. 

The above example is produced with $\sqrt{\smash[b]{y}}$. The option 

b is taken because we want only the depth to be ignored, not the 

height of the letter y. 

Spacing adjustment 

With standard L AT E X, there are a number of commands to fine-tune the 

spacing in a math formula (Section 5.5.1). These are \, \: \; \quad and 

\qquad for increasing amounts of positive spacing, and \! for negative 

spacing. With the amsmath package, the first three still exist, but may 

also be called with the more obvious names \thinspace, \medspace, and 

\thickspace. There is also \negthinspace as an alias for \!. 

The complete set of spacing commands are summarized in the table 

below taken from the A M S-L AT E X manual. 

Short Command Short Command 

form name Demo form name Demo 

\, \thinspace \! \negthinspace 

\: \medspace \negmedspace 

\; \thickspace \negthickspace 

\quad 

\qquad 

The general math spacing command is 

\mspace{mu} 

which inserts space in mathematical spacing units ‘mu’ (=1/18 em). For 

example, \mspace{-9mu} puts in negative spacing of 1/2 em.


Vertical bars 

In standard L AT E X, the commands | and \| are used for single and double 

vertical bars, | and . However, these symbols are often used as delimiters 

(that is, like braces) in which case different spacing requirements are 

needed. In particular, a distinction must be made between the left and 

right delimiter in expressions like |a| and v. The L AT E X commands are 

only appropriate for single appearances, like p|q or f (t, x)|t=0. 

The amsmath package defines the delimiter commands \lvert, \rvert 

for a single bar, and \lVert, \rVert for a double bar. They are useful for 

defining commands that set their arguments in such delimiters, as 

\newcommand{\abs}[1]{\lvert#1\rvert} 

\newcommand{\norm}[1]{\lVert#1\rVert} 

Now $\abs{a}$ produces |a|, and $\norm{v}$ v. 

A similar recommendation can be made for the standard commands 

\langle and \rangle. By defining 

\newcommand{\mean}[1]{\langle#1\rangle} 

one gives $\mean{x}$ to generate 〈x〉, rather than $$ which produces 

< x >. 

Boxed formulas 

A formula may be placed in a box with the command 

\boxed{formula} 

For example, 

produces 

\[ \boxed{\int_0ˆ\infty f(x)\,\dif x \approx 

\sum_{i=1}ˆn w_i \meˆ{x_i} f(x_i)} \] 

∞ 

12.2.6 Multiline equations 

0 

f (x) dx ≈ 

n 

wie xif (xi) 

Equations consisting of several lines which are horizontally aligned at set 

points, such as the equals sign, can be generated in standard L AT E X with 

the eqnarray and eqnarray* environments (Section 5.4.7). Many authors 

consider these to be far too limited for publications with complicated 

multiline equations. A M S-L AT E X therefore provides a range of further 

alignment environments for formulas extending over a single line: 

align gather falign multline alignat split 

i=1


With the exception of split, all exist in a standard and a *-form. As for 

eqnarray, the standard form adds an automatic equation number to each 

line, while the *-form does not. The standard L AT E X equation environment 

for single line formulas is also available in A M S-L AT E X in a *-form. It may 

be used in combination with multiline environments. 

Common features of alignment environments 

All the alignment, or multiline, environments switch to math mode at the 

start and back to text mode at the end, except for split which must 

be called in math mode. A new line is forced in the formula with the 

\\ command, as usual; an optional argument \\[len] can be added to 

increase the line spacing by len, again as usual. 

The automatic numbering with the standard forms can be suppressed 

for single lines by adding \notag before the \\ line break. Alternatively, 

the line can be given a desired marker with \tag{mark}. For example, 

with \tag{$\dag$}, the marker is (†). Using the *-form instead, the 

marker text is printed without the parentheses. 

The vertical position of the equation number or marker can be adjusted 


\raisetag{len} 

which moves the marker upwards by len for that line only. A negative 

value moves it downwards. 

The multline environment 

The multline environment is a variant of the equation environment for 

single formulas that are too long for one line. The line breaks occur where 

the user forces them with the \\ command. The first line is left justified, 

the last right justified, and lines in between are centered. However, if the 

option fleqn has been given, all the lines appear left justified. 

The equation number, if present, appears at the right of the last line 

by default or if the option reqno has been selected; if the option leqno 

has been chosen, the number is placed at the left of the first line. (See 

Section 12.2.8 for the amsmath options.) 

It is possible to shift individual lines fully to the left or right with 

the commands \shoveleft{formula} and \shoveright{formula}. The 

entire formula text for that line, except the terminating \\, is placed in 

their arguments. 

The left and right margins for the formula are set by the length parameter 

\multlinegap which is initially 10 pt. This may be altered by the 

user with the \setlength or \addtolength commands.


An equation with five lines could be broken to look as follows: 

First line — left justified 

Second line — horizontally centered 

Third line — pushed to the left 

Fourth line — pushed to the right 

Last line — right justified (12.1) 

\begin{multline} 

\framebox[.75\columnwidth]{First line --- left justified}\\ 

\framebox[.6\columnwidth]{Second line --- horizontally 

centered}\\ 

\shoveleft{\framebox[.6\columnwidth]{Third line --- pushed 

to the left}}\\ 

\shoveright{\framebox[.6\columnwidth]{Fourth line --- pushed 

to the right}}\\ 

\framebox[.75\columnwidth]{Last line --- right justified} 

\end{multline} 

A real equation would contain mathematical expressions and not the 

\framebox commands in the above demonstration. 

The split environment 

Like multline, the split environment is meant for a single equation 

that does not fit on one line. Line breaks are again forced with the \\ 

command; the difference is that in each line there is an alignment marker 

& such that the lines are horizontally positioned to line up the markers. 

The split environment does not switch into math mode, nor does it 

produce an equation number. It is intended to be applied within another 

math environment, such as equation or gather. This is why there is an 

equation* environment in AMS-LAT EX. The equation number, if present, is provided by the outer environment. 

It is applied to the entire multiline formula, which by default, or with the 

option centertags, is centered on the group of lines. With the option 

tbtags, it is placed either at the left of the last line, or at the right of the 

first line, depending on the further options leqno and reqno, respectively. 

(See Section 12.2.8.) 

Hc = 1 

2n 

n 

(−1) l (k − l) p−2 

l=0 

 

p 

l1+···+lp=l i=1 

ni 

× [(k − l) − (ki − li)] ki−li 

 

× (k − l) 2 − 

li 

 

p 

(ki − li) 2 

j=1 

(12.2)


\begin{equation}\begin{split} 

H_c&=\frac{1}{2n}\sum_{l=0}ˆn (-1)ˆl (k-l)ˆ{p-2} 

\sum_{l_1+\dots+l_p=l} \prod_{i=1}ˆp \binom{n_i}{l_i}\\ 

&\quad\times[(k-l) - (k_i-l_i)]ˆ{k_i-l_i}\times 

\Bigl[(k-l)ˆ2 - \sum_{j=1}ˆp (k_i-l_i)ˆ2\Bigr] 

\end{split}\end{equation} 

The alignment has been chosen to be just before the equals sign. The 

second line begins with the alignment marker & so that the equation 

continues below the = in the first line. A \quad has been inserted so 

that the × is not immediately below the equals sign. Alternatively, one 

could place the & after the = and dispense with \quad in the second line. 

However, the above is more suitable when there are continuation lines 

beginning with =. Note the centered equation number at the right. 

The gather environment 

The gather environment switches to math mode, centering each of its 

formula lines without any alignment. The formula lines are separated by 

\\ commands. Each line receives an equation number, unless the *-form 

has been used, or \notag has been issued in that line. 

1 

2 + 

4 9 n2 ∞ n2 2 3 

n 

n 

+ + · · · + 

+ · · · = 

(12.3) 

3 4 

n + 1 

n + 1 

n=1 

⎛ 

 

n2 n n 

⎜ 

converges since lim 

= lim ⎜ 1 

n→∞ n + 1 n→∞ ⎝ 

1 + 1 

⎞n 

⎟ 

⎠ = 

n 

1 

< 1 

e 

root condition 

2 + 3 

∞ 

4 n + 1 

n + 1 

+ + · · · + + · · · = 

4 9 n2 n 

n=1 

2 

(12.4) 

∞ 

 

x + 1 

diverges since 

dx = ln x − 

x2 1 

∞ = ∞ (integral condition) 

x 

c 

\begin{gather} 

\frac{1}{2} + \left(\frac{2}{3}\right)ˆ4 + \left(\frac{3}{4} 

\right)ˆ9 + \dots + \left(\frac{n}{n+1}\right)ˆ{nˆ2} + \dotsb 

= \sum_{n=1}ˆ\infty \left(\frac{n}{n+1}\right)ˆ{nˆ2} \\ 

\text{converges since}\quad\lim_{n\to\infty} 

\sqrt[n]{\left(\frac{n}{n+1}\right)ˆ{nˆ2}} = \lim_{n\to\infty} 

\left(\frac{1}{1 + \dfrac{1}{n}}\right)ˆn = \frac{1}{\me} < 1 

\tag*{root condition}\\ 

2 + \frac{3}{4} + \frac{4}{9} + \dots + \frac{n+1}{nˆ2} + 

\dotsb = \sum_{n=1}ˆ\infty \frac{n+1}{nˆ2}\\ 

\text{diverges since}\quad\int_cˆ\infty \frac{x+1}{xˆ2}\, 

c


\dif x = \left[ \ln x -\frac{1}{x}\right]_cˆ\infty = \infty 

\tag{integral condition} 

\end{gather} 

Note the use of \tag and \notag (page 271) to add text as markers to the 

second and fourth lines. 

The align environment 

The align environment is intended for multiple equations with horizontal 

alignment, usually on an equals sign or equivalent. New lines are indicated 

with \\ as usual. Each line is split into aligned columns such that the first 

column is right justified against the & character, the second left justified; 

the third column is right justified against the third &, the fourth column 

left justified again, and so on. This is the same as an array environment 

with column specification {rl rl rl ...}. 

(x n ) ′ = nx n−1 

 

1 

xn ′ 

= − n 

xn+1 √ n ′ 1 

x = 

n n√ xn − 1 

(e x ) ′ = e x 

(sin x) ′ = cos x 

(a x ) ′ = a x ln a (cos x) ′ = − sin x 

(ln x) ′ = 1 

x 

(loga x) ′ = 1 

x ln a 

(tan x) ′ 1 

= 

cos2 x 

(cot x) ′ = − 1 

sin 2 x 

\begin{align*} 

\left(xˆn\right)’ &= nxˆ{n-1} & \left(\meˆx\right)’ &= \meˆx & 

(\sin x)’ &= \cos x \\ 

\left(\frac{1}{xˆn}\right)’ &= -\frac{n}{xˆ{n+1}} & 

\left(aˆx\right)’ &= aˆx\ln a & (\cos x)’ &= -\sin x\\ 

\left(\sqrt[n]{x}\right)’ &= \frac{1}{n\sqrt[n]{xˆn -1}} & 

(\ln x)’ &= \frac{1}{x} & (\tan x)’ 

&= \frac{1}{\cosˆ2 x}\\ 

& & (\log_a x)’ &= \frac{1}{x\ln a} & (\cot x)’ 

&= -\frac{1}{\sinˆ2 x} 

\end{align*} 

The align* environment is used here to prevent the lines from being 

numbered, something that is not appropriate for such a collection of 

formulas. The alignment within each of the three column pairs is on the 

equals sign. The input for the last line begins with a double && to produce 

an empty column pair. 

Occasionally a set of formulas is to be aligned on several equals signs 

in one line, as in the equations below for the volume V , inertial moment Iz, 

and mass M of an arbitrary body, in Cartesian, cylindrical, and spherical 

coordinates. In this case, the second and third parts are separated by a


double && so that the left-hand sides of these column pairs are empty: 

the equals signs are always on an odd-numbered alignment marker. 

 

 

V = dv = dx dy dz = ρ dx dρ dφ 

V 

 

Iz = ρ 2 dv = 

V 

 

= r 2 sin θ dr dθ dφ (12.5) 

 

(x 2 + y 2 

) dx dy dz = ρ 3 dz dρ dφ 

 

= r 4 sin 3 θ dr dθ dφ (12.6) 

 

 

M = δ dv = δ dx dy dz = δρ dz dρ dφ 

V 

 

= δr 2 sin θ dr dθ dφ (12.7) 

\begin{align} 

V &= \int\limits_V\dif v &&= \iiint\dif x\,\dif y\,\dif z 

&&= \iiint \rho\,\dif x\,\dif \rho\,\dif \phi \notag \\ 

&&&&&= \iiint rˆ2 \sin\theta\,\dif r\,\dif\theta\,\dif\phi\\ 

I_z &= \int\limits_V \rhoˆ2\,\dif v &&=\iiint(xˆ2 + yˆ2) 

\,\dif x\,\dif y\,\dif z 

&&= \iiint \rhoˆ3\,\dif z\,\dif \rho\,\dif \phi \notag \\ 

&&&&&= \iiint rˆ4\sinˆ3\theta\,\dif r\,\dif\theta\,\dif\phi\\ 

M &= \int\limits_V \delta\,\dif v &&= \iiint \delta 

\,\dif x\,\dif y\,\dif z 

&&= \iiint \delta\rho\,\dif z\,\dif\rho\,\dif\phi\notag \\ 

&&&&&=\iiint\delta rˆ2\sin\theta\,\dif r\,\dif\theta\,\dif\phi 

\end{align} 

There are two variations on the align environment, falign and 

alignat. The first has exactly the same syntax as align but it inserts 

so much spacing between the column pairs that the entire line is filled 

out. The alignat environment is just the opposite: no spacing is inserted 

automatically between the column pairs. It must take the number 

of column pairs as a mandatory argument, but otherwise the syntax of 

the contents is the same as that for align. The example above with the 

volume, inertial moment, and mass of a body could just as well have been 

given with 

\begin{alignat}{3} formula text \end{alignat} 

(In fact, this is precisely what was done in order to fit it within the line 

width of this book.) 

If one or more columns are empty, as in this example, it is possible 

to control their widths precisely in the alignat environment by adding


explicit spacing between the two & characters in one of the lines. See 

Section 5.5.1 for spacing in math mode. 

In summary, for the align environment and its variants, the first, 

third, fifth, etc. & characters are alignment markers, while the second, 

fourth, sixth, etc. are column pair separators. 

Nested alignment environments 

We have already pointed out on page 272 how the split environment 

is to be placed inside an equation environment. The same is true for 

the environments aligned and gathered, which may be used as building 

blocks within formulas. Their contents and behavior are otherwise the 

same as their related environments. 

Both of these environments take an optional argument pos 

\begin{aligned}[pos] lines \end{aligned} 

\begin{gathered}[pos] lines \end{gathered} 

which takes values of t or b to determine the vertical alignment (top or 

bottom) when they appear beside other elements. When no pos is given, 

they are centered. In this way 

α = aa 

β = bbbbb 

γ = g 

is produced with 

versus δ = dd 

η = eeeeee 

ϕ = f 

versus 

s = x + y 

d = u − v − w 

p = x ◦ y 

\begin{equation*} 

\begin{aligned} \alpha&=aa\\ \beta&=bbbbb\\ \gamma&=g 

\end{aligned} 

\qquad\text{versus}\qquad 

\begin{aligned}[t] \delta&=dd\\ \eta&=eeeeee\\ \varphi&=f 

\end{aligned} 

\qquad\text{versus}\qquad 

\begin{gathered}[b] s= x+y\\ d= u - v - w\\ p = x\circ y 

\end{gathered} 

\end{equation*} 

The cases environment 

Although it is possible with standard LAT EX to produce structures of the 

form 

 

0 

Pr −j = 

r ! (−1) 

if r − j is odd, 

(r −j)/2 if r − j is even. 

(12.8)


as demonstrated by a similar example in Section 5.4.1 on page 132, the 

A M S-L AT E X cases environment allows a simpler input: 

\begin{equation} 

P_{r-j}=\begin{cases} 0 & \text{if $r-j$ is odd,}\\ 

r!\,(-1)ˆ{(r-j)/2} & \text{if $r-j$ is even.} 

\end{cases} 

\end{equation} 

There may be more than two cases in the environment, as in the 

example reproduced here from page 132: 

⎧ 

⎪⎨ −1 : x < 0 

y = 0 

⎪⎩ 

+1 

: 

: 

x = 0 

x > 0 

12.2.7 Equation numbering 

Numbering hierarchy 

\[ y = \begin{cases} -1 &:\quad x0 

\end{cases} \] 

With the standard L AT E X classes book and report, equations are given 

a double number with the chapter designation and then a sequential 

number starting at 1 for each new chapter. For the article class, the 

equations are numbered sequentially throughout the work. 

With the amsmath package, it is possible to alter this hierarchy. For 

example, if an article is to have the equations numbered within each 

section, with the section number, give 

\numberwithin{equation}{section} 

to redefine the equation numbers to include the section number and to 

make the equation counter reset every time the section counter is incremented. 

This is as though the equation counter had been created with 

\newcounter{equation}[section] (Section 8.1.2), something which the 

user cannot normally bring about. Furthermore, \theequation is redefined 

to be \thesection.\arabic{equation}, something that is in the 

user’s power but which is not much use if the equation counter is never 

reset. 

Subnumbering equations 

On page 190 we give an example of how equations may be subnumbered, 

that is, the main equation number stays the same and a letter is appended 

to it, as 1.8a, 1.8b, 1.8c . . . . The amsmath package provides this feature 

with the subequations environment. Numbered equations appearing 

within 

\begin{subequations} . . . \end{subequations}


will all have the same main number which is one more than that of the 

previous one, with sequential, lower case letters attached. 

Within the environment, the equation counter refers to the subnumber, 

that is, to the letters, while the main number is to be found in the 

parentequation counter. To change the format of the subnumber, say 

to 1.8-A, 1.8-B, . . . , give 

\begin{subequations} 

\renewcommand{\theequation} 

{\theparentequation-\Alph{equation}} 

. . . 

\end{subequations} 

Referencing equation numbers 

The L AT E X cross-reference system is described in Section 9.2.1 and works 

exactly the same way with A M S-L AT E X: when \label{marker} is issued in 

a mathematical formula that receives an automatic equation number, that 

number can be printed anywhere in the text with \ref{marker}, where 

marker is arbitrary text to identify that equation. 

The amsmath package adds a command \eqref{marker} to print the 

equation number in parentheses, as it appears beside the math formula. 

For example, the cases equation on page 276 is referred to as equation 12.8 

with \ref, or as equation (12.8) with \eqref. 

If \label is given immediately after the start of a subequations environment, 

the corresponding \ref commands will print the main equation 

number without the extra letter. In this way one can refer to the entire 

group of equations. Later \label commands are associated with 

individual equation lines and reference them with the letters. 

Page breaks within multiline formulas 

Unlike the standard L AT E X eqnarray environment, the A M S-L AT E X multiline 

math environments do not normally allow any page breaks to occur within 

them. The idea is that the author should have more control over where 

such breaks may occur. To allow or force a page break within a multiline 

equation, one gives 

\displaybreak[num] 

just before the line breaking command \\. Here the optional num has 

the same meaning as for the standard \pagebreak command (page 33): 

without it, a new page is forced, but it may take values of 0–4 to allow 

a break with increasing degree of encouragement, whereby 4 also forces 

the page break. 

Alternatively, one can issue \allowdisplaybreaks in the document 

preamble to allow L AT E X to break pages automatically within multiline


formulas as necessary. This command too takes an optional argument 

num with possible values between 0 and 4 which make it progressively 

easier for automatic page breaks to occur. 

Once \allowdisplaybreaks has been given in the preamble, it is still 

possible to suppress page breaks within a formula by ending an equation 

line with \\* instead of with \\. 

12.2.8 Package options for amsmath 

The main amsmath package for A M S-L AT E X recognizes a number of options 

that may be given when it is loaded with 

\usepackage[options]{amsmath} 

They are listed here as pairs with opposing effects. The member of 

each pair that is assumed if neither is given, the default, is indicated by 

underlining. 

centertags | tbtags The equation number for a split environment 

(page 272) is centered vertically by default. With tbtags, it is placed 

either to the left of the first line or to the right of the last line, 

depending on the side on which numbers are to appear. 

sumlimits | nosumlimits In displayed formulas, initial and final limits 

appear below and above the sign with the sumlimits option. With 

nosumlimits, they are placed beside the sign, raised and lowered 

with the usual ˆ and _ characters. 

 

Other 

 

symbols 

 

that are 

 

affected by these options are 

and . On the other hand, integral signs are not 

influenced by them. 

sumlimits 

∞ 

m−1 

1 

n! 

= 2; n − i = 

2n (n − m)! 

n=0 

i=0 

nosumlimits 

∞ 1 

n=0 2n = 2; m−1 n! 

n − i = 

i=0 (n − m)! 

The input text is the same for both of the above cases. 

intlimits | nointlimits Integral signs normally have their limits at 

the side; these options allow them to be placed above and below as 

for summations. 

intlimits 

a 

0 

 

a 2 − x 2 dx = 

1 

0 

a 2 

 

1 − sin 2 t d sin t = a 2 

π/2 

cos 2 t dt = πa2 

4 

0


nointlimits 

a 

0 

 

a 2 − x 2 dx = 

1 

0 

a 2 

 

1 − sin 2 t d sin t = a 2 

π/2 

0 

cos 2 t dt = πa2 

4 

where again the input text is the same for both cases. As a reminder, 

the standard L AT E X treatment of limits on integral signs is the same 

as for nointlimits. 

namelimits | nonamelimits The function names \det, \gcd, \inf, 

\lim, \liminf, \limsup \max, \min, \Pr, and \sup frequently take 

lower limits which normally appear below the name in displayed 

formulas. With nonamelimits, they are placed at the lower right. 

namelimits 

 

lim 1 + 

x→∞ 

1 

x = e = 2.7182 . . . 

x 

limx→∞ 

nonamelimits 

 

1 + 1 

x = e = 2.7182 . . . 

x 

The above options determine the standard placement of limits for the 

entire document. It is still possible to change the behavior in any particular 

case with the \limits and \nolimits commands, as in normal L AT E X. 

The remaining package options select the side for equation numbers 

and the horizontal positioning of equations. 

leqno | reqno The standard location for equation numbers is on the 

right side, at the margin; with leqno, they are placed on the left side 

of the equation. 

fleqn With this option, all displayed equations are printed flush left, set 

off from the left margin by an amount \mathindent. Without this 

option, equations are centered. (This is like the fleqn class option 

for the standard classes, page 39.) 

12.3 Further A M S-L AT E X packages 

The A M S-L AT E X packages described in this section must be loaded explicitly 

if their features are to be exploited. Unlike the amsbsy and amstext 

packages, they are not loaded automatically with amsmath. They may, 

however, be used on their own, independently of the main package. 

12.3.1 Extended theorem declarations 

Package: The amsthm package offers many additional possibilities for generating 

amsthm theorem-like declarations described for standard LAT EX in Section 4.5. 

As for standard LAT EX, a new theorem declaration is created with a 

statement like

\newtheorem{com}{Comment} 

12.3. Further A M S-L AT E X packages 281 

where the first argument is the name of the theorem type (here com) and 

the second is the title that is printed when the theorem declaration is 

invoked. For example, 

\begin{com} 

Theorem declarations can have any name. 

\end{com} 

produces the declaratory text 

Comment 1. Theorem declarations can have any name. 

In addition to the two mandatory arguments, the \newtheorem command 

may have one of two optional ones. The complete syntax is 

\newtheorem{type}[num like]{title} 

or 

\newtheorem{type}{title}[in counter] 

where num like is the name of an existing theorem-like declaration which 

is to be numbered in the same sequence as type, and in counter is a 

counter name like chapter or section to reset the numbers of the type 

declarations. 

All this is standard L AT E X so far. The amsthm package adds the following 

features. 

• A \newtheorem* is provided that defines an unnumbered theorem 

structure. 

• Three predefined theorem styles are available: 

plain in which the title and number are in bold face and the text 

italic; 

definition with title and number in bold face and the text in 

normal font; 

remark for title and number in italic and the text normal. 

The desired style is activated by first issuing \theoremstyle{style}; 

all subsequent \newtheorem statements will have this style until a 

new one is activated. 

• A \swapnumbers can be issued to cause all following new theorem 

types to have the numbers appear before the title, as 1 Comment. 

• New theorem styles may be defined by the user by means of a 

\newtheoremstyle command, or additional predefined styles may 

be loaded with package options. Since this is fairly specialized and 

complex, it is best to examine the example file thmtest.tex or read 

the documentation in amsthm.dtx.


• A proof environment is available for presenting short proofs. It is 

an unnumbered structure with the title Proof. The text is terminated 

automatically with the Q.E.D. symbol . This symbol may be altered 

by redefining the command \qedsymbol; it may be printed at any 

time by issuing \qed. 

The amsthm package has much in common with Frank Mittelbach’s 

theorem package in the tools collection of Section B.5.4. 

12.3.2 Commutative diagrams 

Package: 

amscd 

The extra A M S-L AT E X package amscd 

makes it easier to generate commutative 

diagrams like the one here at the right. 

SWΛ ⊗ T 

⏐ 

 

j 

−−−→ T 

⏐ 

End P 

(S ⊗ T )/I (Z ⊗ T )/J 

These diagrams are created within the CD environment using some 

additional arrow commands. These bear the rather unusual names @>>> 

@j>> T\\ 

@VVV @VV{\End P}V\\ 

(S\otimes T)/I @= (Z\otimes T)/J 

\end{CD} \] 

The command \End to print the function name ‘End’ is not standard. It 

must be previously defined with \DeclareMathOperator{\End}{End} 

(see page 268). 

12.3.3 References with upref package 

Package: Normally the numbers printed with the \ref and \pageref commands 

upref are in the current font, whether that be bold, italic, or upright. In order 

to ensure that the numbers are always upright, load the extra AMS-LAT EX package upref.

12.4 The A M S fonts 

12.4. The A M S fonts 283 

The A M S makes a number of fonts available to complement the regular 

Computer Modern fonts provided with the standard T E X/L AT E X installation. 

They include extra math alphabets, supplemental CM bold math italic and 

symbol fonts in smaller sizes than 10 pt, Cyrillic fonts, and additional 

symbol fonts. 

In the next sections we describe these various fonts and how to take 

advantage of them. 

12.4.1 Extra CM math fonts 

Standard T E X installations of Computer Modern fonts provide bold math 

italic cmmib10, the bold symbols cmbsy10, and math extensions cmex10 

fonts only in 10 pt size, as indicated by the suffix 10 to their names. The 

A M S has supplemented these with versions in sizes 5–9 pt. 

cmmib5 cmmib6 cmmib7 cmmib8 cmmib9 

cmbsy5 cmbsy6 cmbsy7 cmbsy8 cmbsy9 

cmex7 cmex8 cmex9 

The small caps font cmcsc10 is also given companions cmcsc8 and 

cmcsc9. 

The normal L AT E X installation automatically assumes that these fonts 

are on the system and incorporates them into the necessary NFSS font 

definition files. Substitutions will be made if they are missing. 

12.4.2 Cyrillic fonts 

The A M S Cyrillic fonts were originally used in reviews of books published 

in Russian and other Slavic languages in which the titles were to be 

rendered in the original language. In 1988, the Humanities and Arts 

Computing Center of the University of Washington redesigned them for 

general purpose Slavic studies, adding the pre-Revolutionary and accented 

letters. The overall appearance was also greatly enhanced (see Layout 8 on 

page 497). 

These fonts all bear the prefix wncy, followed by the style designation 

r (upright), b (bold), i (italic), sc (small caps), or ss (upright sans serif), 

and the design size in points. 

The best way to enable the Cyrillic fonts is simply to select font 

encoding OT2 under NFSS, as illustrated in Section A.2 on page 371. The 

.fd (font definition) files for the Cyrillic fonts have been so set up that the 

other font attributes fully parallel those of the Latin fonts. This means 

wncyr10 has all the same attributes as cmr10, except for the encoding: 

family cmr, shape n, series m. (Of course, if the CM fonts are not the


current standard ones, it will require more than just selecting the OT2 

encoding to activate these Cyrillic fonts.) 

The layout of the Cyrillic fonts has been chosen in such a way that 

the input text may be entered following the regular English transliteration 

scheme. Thus Cyrillic S is in position 83 where Latin S is normally 

situated. Typing S when a Cyrillic font is active outputs the correct 

equivalent S. Numerals and punctuation are to be found in the standard 

locations and so may be typed in as usual. ‘Sankt-Peterburg 10?’ is 

thus generated by {\cyr Sankt-Peterburg 10?}. 

Since the Cyrillic alphabet possesses more letters than the Latin, many 

of them must be transliterated with multi-letter combinations. These are 

automatically programmed into the fonts using T E X’s ligature system. For 

example, Ch is treated as a ligature for symbol 81 Q just as fi is for fi in a 

Latin font. This means that multi-letter transliterations are simply typed 

in. The input for ‘Hruwev’ is {\cyr Khrushchev}, where Kh → H and 

shch → w. The transliteration scheme is that for English; other languages 

have their own systems to reproduce the original pronunciation. For 

example, ‘Gorbaqv’ is Gorbatschow in German, Gorbaciov in Italian, and 

Gorbachev in English. These other schemes do not work with these fonts. 

Not all letters can be produced so automatically (for example, the in 

Gorbaqv) and for this reason the A M S provides a file cyracc.def containing 

macro definitions for accented letters and other special features, 

such as the hard and soft signs. When these macros are given in a Latin 

font, additional transliteration symbols appear. 

12.4.3 Euler fonts 

The fonts known as the Euler collection, named after the eighteenthcentury 

mathematician Leonhard Euler, have been collected from various 

sources by the AMS. They include a ‘blackboard’ font, representing a 

professor’s handwriting on a blackboard, a Fraktur or Gothic font, and a 

script font. Their main purpose in mathematics is to be a substitute for 

the CM calligraphic math alphabets. 

Package: To use the blackboard characters, one must load the package amsfonts 

amsfonts (which is included automatically with the amssymb package) and then employ 

the math alphabet command \mathbb. Thus $\mathbb{A B C ..}$ 

produces 

A B C D E F G H I J K L N M O P Q R S T U V W X Y Z 

Package: The package eucal redefines the \mathcal command to use the Euler 

eucal script characters in place of the CM calligraphic letters. If this package 

is loaded with the option mathscr, the command \mathcal is left unchanged 

and instead \mathscr is defined to invoke these letters. In this 

case, $\mathscr{A B C ..}$ produces

12.4. The A M S fonts 285 


Package: Finally, the package eufrak defines the math alphabet command 

eufrak \mathfrak, with which $\mathfrak{A B C ..}$ yields 


This math alphabet is also enabled with the amsfonts package. 

12.4.4 Extra math symbols 

The set of symbols in the CM math symbol fonts by no means exhausts the 

fantasies of active mathematicians. To overcome this deficiency, the A M S 

has produced two fonts, msam10 and msbm10, containing only symbols, 

arrows, and the blackboard characters. They are also available in sizes 

5–9 pt. Since these fonts originated in the days when T E X could only 

handle 128 characters in any font, that is exactly how many they contain. 

Today, they could be combined into one font of 256 characters. 

Two packages permit access to this treasure trove of hieroglyphs: 

amsfonts enables the \mathbb math alphabet command for the blackboard 

characters, and defines those symbol names which are otherwise 

only provided in the latexsym package (Section 5.3.3). 

amssymb is the more convenient package, which loads amsfonts and then 

defines names for all the symbols in the two fonts. 

For example, 

\[ \circlearrowright \Cup \lessapprox \lll \varpropto \because 

\circeq \vDash \blacktriangle \sphericalangle \] 

↻ ⋒ ⋘ ∝ ∵ ≗ ⊨ ∢ 

All the symbols and their associated names from the amssymb package 

are to be found in Tables H.20–H.26 on pages 599–601.

13 

Drawing with LAT E X 

The inclusion of graphical material from other programs is treated in 

Chapter 6. Such ‘foreign’ files do add some complications (which are 

greatly reduced today with L AT E X 2ε) and can cause portability problems. 

What would be desirable is a means to do graphics with L AT E X itself so that 

the source text file is fully self-contained. 

Standard L AT E X does contain the means to make somewhat primitive 

drawings on its own. The word ‘primitive’ should not be considered 

derogatory, for simple building blocks are the basic units for constructing 

very complicated, sophisticated structures. They are also useful for 

superimposing imported graphics or for adding embellishments to them, 

as demonstrated in Section 6.1.5. 

This chapter describes the intrinsic L AT E X drawing capabilities, and then 

explains some extensions that are available to enhance them. 

Many more possibilities exist for adding specialized diagrams, from 

chemistry, to music, to chess positions. These are described in detail in 

The L AT E X Graphics Companion by Goossens et al. (1997). 

13.1 The picture environment 

13.1.1 Picture coordinates 

The picture building blocks can only be put in place once a coordinate 

system has been established for that picture. This consists of a reference 

point or origin, and two mutually perpendicular coordinate axes, as well 

as a length unit for the coordinates. The origin is the lower left corner 

of the picture and the axes are its lower and left edges. These edges are 

referred to as the x-axis (lower) and y-axis (left). 

Once the unit of length (UL) has been specified, every point within the 

picture area can be uniquely referred to with two decimal numbers: the 

first is the number of length units along the x-axis, the second the number 

along the y-axis. 

287

288 Chapter 13. Drawing with L AT E X 

The coordinate numbers are generally positive, meaning that the point 

lies to the right and above the reference point. Since this point is the 

lower left corner of the picture, all other points should be more to the 

right and higher than it. However, negative values are also possible. A 

negative x value (a negative number for the first member of a coordinate 

pair) defines a point left of the origin, while a negative y value (the second 

member of the pair is negative) specifies a point below the reference point. 

y-axis 

✻ 

1.4 

origin 

✠ 

✛ UL ✲ 

(1.8,1.4) 

✲ x-axis 

1.8 

(−2.2,1.8) 

 

negative coordinate values 

y 

✻ 

 

(−1.8,−0.8) 

The unit length is selected with the command 

\setlength{\unitlength}{length} 

 

✛UL✲ (3.0,2.0) 

 

 

(2.5,−1.0) 

✲ x 

In the left-hand example above, the unit length was set to a value of 

1.5 cm with the command \setlength{\unitlength}{1.5cm}. The 

point (1.8,1.4) then lies 1.8 times the unit length (= 2.7 cm) to the right, 

and 1.4 times (= 2.1 cm) above the origin. In the right-hand example, the 

unit length is set to 1 cm. 

The unit of length is normally set to a convenient size such as 1 cm, 

1 mm, or 1 in and the picture is built up accordingly. Once it has been 

completed, it is possible to rescale the whole thing simply by changing 

the value of the length unit. A picture that was originally designed with 

\unitlength set to 1 cm can be enlarged by a factor 1.2 by redefining its 

length unit to be 1.2 cm. 

13.1.2 The picture environment 

Pictures are constructed within the picture environment which is started 

with 

\begin{picture}(x dimen,y dimen) 

picture commands 

\end{picture}

13.1. The picture environment 289 

where (x dimen,y dimen) is a pair of numbers that specifies the size 

(dimensions) of the picture in the x-direction (horizontal) and y-direction 

(vertical). This pair of numbers is enclosed in round parentheses! The 

unit of length is that previously selected by \unitlength. 

\setlength{\unitlength}{1.5cm} 

\begin{picture}(4,5) ... ... ... \end{picture} 

produces a picture that is 4 length units wide and 5 units high. Since the 

length unit has been set to 1.5 cm, the actual size is 6 cm wide and 7.5 cm 

high. 

Picture commands are those commands described below that are used 

to produce and position the individual picture elements. These are the 

only commands that are allowed within the picture environment, other 

than font style and size declarations (Section 4.1) and the line thickness 

commands \thicklines and \thinlines. These last determine which 

of the two available line thicknesses will become current for drawing 

lines: one may switch back and forth as one pleases. Initially thin lines 

are active. 

The value of the parameter \unitlength must not be altered within 

the picture environment, for it must remain the same for the entire 

picture. It may, however, be changed between pictures. 

If the \unitlength specification together with the picture environment 

are enclosed within another environment, such as \begin{center} 

... \end{center}, then that value of \unitlength is valid only until 

the end of the environment. A picture environment without a preceding 

\unitlength command uses the standard value of 1 pt. 

13.1.3 The positioning commands 

Picture elements are generated and positioned by means of the two commands 

\put and \multiput, which have the syntaxes: 

\put(x,y){pic elem} 

\multiput(x,y)(∆x,∆y){num}{pic elem} 

The pic elem is one of the picture element commands described in the next 

section. The arguments (x,y) are the placement coordinates, designating 

the location of the picture element within the picture coordinate system, 

in units of \unitlength. If this is 1 cm, then (2.5,3.6) means that the 

element is to be positioned 2.5 cm to the right and 3.6 cm above the lower 

left corner of the picture. 

The \multiput command generates the same picture element num 

times, moving it (∆x,∆y) each time. Thus the element is drawn at 

(x, y), (x + ∆x, y + ∆y), 

(x + 2∆x, y + 2∆y), . . . up to 

(x + [num − 1]∆x, y + [num − 1]∆y)


The coordinate pair (x, y) is incremented by (∆x, ∆y) for each successive 

placement. The values of the incrementing pair may be positive or 

negative. 

Thus \multiput(2.5,3.6)(0.5,-0.6){5}{pic elem} produces the 

pic elem a total of five times, first at the location (2.5,3.6) and then at 

(3.0,3.0), (3.5,2.4), (4.0,1.8), and finally (4.5,1.2). 

Note that the numbers for the coordinate and increment pairs are given 

in round parentheses ( , ) and that the two numbers within each pair 

are separated by a comma. The num and pic elem entries, on the other 

hand, are enclosed in curly brackets { } as usual. 

Warning: Since the comma separates the two numbers in a coordinate 

pair, it may not be used in place of a decimal point. For coordinate entries, 

decimal numbers must be written with a period, not a comma. 

13.1.4 Picture element commands 

Text within pictures 

The simplest picture element of all is a piece of text, positioned at the 

desired location within the picture. This is accomplished by putting text 

in place of pic elem in the \put or \multiput command. 

 

(1.8,1.2) 

✒ 

An arrow 

The arrow points to the location (1.8,1.2). The command 

\put(1.8,1.2){An arrow} inserts the text 

‘An arrow’ so that its lower left corner is at the 

specified position. 

The text as picture element may also be packed into a \parbox or a 

minipage environment, and the reference point for the coordinate entry 

in the \put command depends on the positioning arguments of that box: 

✁✁✕ 

\parbox[b]{..}{..} 

Reference point is the 

lower left corner of the 

last line in the parbox. 

✓ ✓✓✼ 

\parbox{32mm}{...} 

For a standard parbox, 

the reference point is 

the vertical center of the 

left edge. 

✡ ✡✡✡✣ 

\parbox[t]{..}{..} 

Reference point is the 

lower left corner of the 

top line in the parbox. 

Exercise 13.1: Produce a picture 100 mm wide and 50 mm high with \unitlength 

equal to 1 mm. Place the given texts at the following locations: (0,0) ‘The First 

Picture’, (90,47) ‘upper left’, (70,40) ‘somewhere upper right’, and put a parbox 

of width 60 mm at (25,25) containing ‘A separate exercise file with the name 

picture.tex should be created for the exercises in this Chapter.’ 

Exercise 13.2: Repeat the picture processing with a value of 1.5 mm for 

\unitlength, and with positioning arguments t and b for the parbox.

Picture boxes—rectangles 


The box commands \framebox, \makebox, and \savebox (Section 4.7.1) 

are available in the picture environment but with an extended syntax. In 

addition, there is another box command \dashbox: 

\makebox(x dimen,y dimen)[pos]{text} 

\framebox(x dimen,y dimen)[pos]{text} 

\dashbox{dash len}(x dimen,y dimen)[pos]{text} 

The dimensional pair (x dimen, y dimen) defines the width and height 

of the rectangular box in units of \unitlength. The positioning argument 

pos determines how the text is located within the box. It may take on 

values: 

[t] top The input text appears—centered horizontally—below the upper 

edge of the box. 

[b] bottom The input text appears—centered horizontally—above the 

lower edge of the box. 

[l] left The input text begins—centered vertically—at the left edge of 

the box. 

[r] right The input text ends—centered vertically—at the right edge of 

the box. 

[s] stretch The input text is stretched horizontally to fill up the box, 

and centered vertically. 

Without the optional argument pos, the input text is centered vertically 

and horizontally within the box. 

These positional values may be combined two at a time: 

[tl] top left The text appears at the upper left. 

[tr] top right The text appears at the upper right. 

[bl] bottom left The text appears at the lower left. 

[br] bottom right The text appears at the lower right. 

The order of the values is unimportant: tl has the same effect as lt. 

These box commands are to be used as pic elem in the placement 

commands \put and \multiput. The box is so placed that its lower left 

corner is at the position given by the coordinate pair in the placement 


\put(1.5,1.2){\framebox(2.5,1.2){center}}


 

(1.5,1.2) 

✒ 

center 

✛ 2.5 UL ✲ ❄ 

✻ 

1.2 UL 

The arrow indicates the point (1.5,1.2) which 

is the position of the lower left corner of 

the rectangle with width 2.5 units and height 

1.2 units. The text ‘center’ is centered both 

horizontally and vertically. UL = 0.8 cm. 

The effect of the text positioning argument is made clear with the 

following examples (UL = 1 cm): 

❅■ 

❅ 

top center 

(3.0,3.2) 

❅ ❅❘ 

center right 

bot. left 

❅■ 

(0.0,1.95) ❅ (3.0,1.95) 

(2.0,0.3) 

❅ center stretch 

❅❘ 

\put(0.0,1.95){\framebox(2,1.0) 

[t]{top center}} 

\put(3.0,1.95){\framebox(2,0.8) 

[lb]{bot. left}} 

\put(3.0,3.2){\framebox(2,0.6) 

[r]{center right}} 

\put(2.0,0.3){\framebox(2,0.6) 

[s]{stretch\hfill center}} 

The picture element \makebox is exactly the same as the \framebox 

command but without the rectangular frame. It is most often employed 

with the dimensional pair (0,0) in order to place text at a desired location. 

(See Section 4.7.1 for the effect of zero width boxes on the enclosed text.) 

(2.0,2.8) ✲flush 

left 

\put(3,1.6){\makebox(0,0){center center}} 

\put(2,0.5){\makebox(0,0)[tr]{top right}} 

(3.0,1.6) \put(4,1.0){\makebox(0,0)[b]{bot center}} 

center ❄center 

\put(2,2.8){\makebox(0,0)[l]{flush left}} 

(2.0,0.5) 

❅❘ 

top right 

bot center The combination [lb] positions the text in exactly 

✻ the same way as simply typing the text in as pic elem 

(4.0,1.0) 

without a box, as shown on page 290. 

The picture element \dashbox also produces a framed box, but with 

a dashed line around it. The argument dash len specifies the dash length. 

✟ ✟✯ 

(1.0,0.75) 

\put(1.0,0.75){\dashbox{0.2}(4,1){dashed frame}} 

dashed frame 

A dashed frame looks best when both the height 

and width are a multiple of the dash length. 

Even in the above picture box commands, the entered text may be put 

into a vertical box (\parbox or minipage). Since vertical boxes themselves 

possess an optional positioning argument b or t, which must not conflict 

with that of the picture box, the following rule must be observed: 

If a picture box command contains the positioning argument 

b or t, then the same value must be applied to the enclosed 

vertical box. If the picture box command has no positioning 

argument or only r or l, then the vertical box must be used in 

the standard (no argument) form.


The positioning argument of a picture box has the same effect on the 

enclosed vertical box as it does on a line of text, in that the whole box is 

treated as a single unit. 

Exercise 13.3: Reproduce the organization table below with the boxes and 

included text but without the horizontal and vertical lines and arrows. These 

will be part of the next exercise. 

Hint: first draw the boxes on a piece of squared paper so that the edges of the 

boxes lie on the printed rules. Select the unit of length to be the rule spacing 

of the paper. Take as the origin the lower left corner of an imaginary frame 

surrounding the entire diagram. 

Scientific 

Advisors 

✻ 

Directorate 

Managing Director 

Trustees 

Science Administration Technical Head 

Project planning and 

managing 

Projects Services Stores Phys. Plant Workshops Labs 

Development 

Analysis 

Theory 

Straight lines 

✛ 

✛ 

Library 

Computer 

facilities 

✛ ✲ ✛ ✲ 

Orders 

Receiving 

Customs 

Personnel 

Bookkeeping 

including 

general 

services 

✛ ✲ 

Staff planning 

Drafting 

Project 

hardware 

construction 

Project 

working 

groups 

In the picture environment, L AT E X can draw straight lines of any length, 

horizontally and vertically as well as at a limited number of angles. The 

syntax for this picture element reads 

\line(∆x,∆y){length} 

For horizontal and vertical lines, length specifies how long the line is to be 

in length units. For lines at an angle, it has a somewhat more complicated 

meaning, as is explained below. The line begins at that spot given by the 

placement coordinates in the \put or \multiput command. 

 

✠ 

(0,0) 

(7.5,0) 

❅ 

❅❘ 

\thicklines 

\put(0,0){\line(1,0){6}} 

\put(0,0){\line(0,1){1}} 

\put(6,0){\line(0,1){0.5}} 

The angle at which the line is drawn is given by the slope pair (∆x,∆y). 

The slope pair (1,0), in which ∆x = 1 and ∆y = 0, produces a horizontal 

line, while the pair (0,1) leads to a vertical line. This is illustrated in the 

above example. 

In general (∆x,∆y) has the following meaning:


y 

✟ 

✻ 

✻ 

∆y 

✟✟✟✟✟✟ 

✟ ❄ 

A ✟✟✟✟✟✟✟ 

✛ ∆x ✲ 

✲ 

x 

Beginning at a point A on the line and moving 

a distance ∆x in the x direction (horizontally), 

∆y is the distance one must move in 

the y direction (vertically) in order to rejoin 

the line. 

By specifying a slope pair (∆x,∆y), a line is 

drawn at just that angle to fulfill the above 

conditions. 

As mentioned already, the number of different slopes available is 

limited. This is because ∆x and ∆y may only take on values according to 

certain rules: 

1. The number must be a whole integer (negative or positive). 

2. Only the values 0, 1, . . . , 6 are allowed. 

3. The two numbers in the pair may not contain a common divisor. 

Pairs such as (3.5,1.2) (rule 1) and (7,0) (rule 2) are thus forbidden. 

Similarly (2,2) and (3,6) are invalid pairs by rule 3, since both numbers 

in the first pair are divisible by 2, and those in the second by 3. The 

same angles are achieved with the pairs (1,1) and (1,2) respectively. 

In all there are 25 allowed slope pairs, including (1,0) for horizontal 

and (0,1) for vertical lines, as one can verify by writing down all the 

possibilities. 

In addition, the numbers in the slope pair may be positive or negative, 

for example, (0,-1), (-2,-5). A negative ∆x in the above diagram means 

motion to the left, and a negative ∆y is for motion downwards. Thus 

\put(2,3){\line(0,-1){2.5}} draws a line starting at point (2,3) and 

going 2.5 length units vertically downwards. 

y 

✻ 

❍ 

✛ 

∆x=2 ✲ 

❍❍❍❍❍❍ ✻ ∆y=−1 

❄ 

❍ 

✛ len = 3.5 ✲ 

✲x 

For lines at an angle, the argument length 

determines the projected distance along 

the x-axis. This is made clearer with 

the help of the diagram on the left. 

\put(1.0,2.75){\line(2,-1){3.5}} 

If one draws dashed lines straight down 

from the two end points, then that part of 

the x-axis between them is the projection 

of the line on to the x-axis. 

Sloping lines must have a minimum length of about 10 pt or 3.5 mm, 

otherwise a warning is issued and no line is drawn. 

Arrows 

The arrow picture element is made with the command

\vector(∆x,∆y){length} 


which functions exactly the same as the \line command as far as the 

arguments and their limitations are concerned. The command draws a line 

from the placement location given in the \put or \multiput command 

and places an arrow head at the end. 

Just as for lines, arrows too must have a length of at least 10 pt or 

3.5 mm. Rules 1–3 apply to ∆x and ∆y as well, with the further restriction 

that the allowed numbers are limited to 0, 1, . . . , 4. This makes a total of 

13 possible angles for arrows, not counting changes in sign. 

 

✛ 

 

✒◗◗ ◗◗◗◗◗◗◗ 

Exercise 13.4: Complete the diagram in Exercise 

13.3 by including the missing horizontal and 

vertical lines and arrows. 

Exercise 13.5: Generate the figure at the right. 

The corner points are (0,5), (0,10), (5,15), (10,15), 

(15,10), (15,5), (10,0), and (5,0) and the length unit 

is 0.1 inch. 

Circles 

\begin{picture}(5,2)\thicklines 

\put(5,0){\vector(-1,0){5}} 

\put(0,0){\vector(1,1){2}} 

\put(2,2){\vector(3,-2){3}} 

\end{picture} 

✁ ✁ 

✁ 

✁ 

✁✁✁✁ 

 

 

 

 

 

✏✏✏✏✏✏✏✏✏✏✏ ❆❆ 

❆ 

❆ 

❍ 

❅❍❍❍❍❍ 

❆ 

❅❆ ❅❅❆ ❆ ✂ 

✂ ❍ 

✂ 

✂ ✂✂ 

✂ ✂ 

✂ ✂✂ 

 

 

 

 

❇ 

❇ 

❇ 

❇ 

❅❅ ❇ 

❅ ❇ 

❅ ❇ 

❅ ❇ ✟ 

❅ ❇ 

❅ ❇ ✁ 

✟ 

✟✟✟✟✟ 

❅❇ 

✁ 

✁ 

✁ ✁✁ 

✁✁ 

 

 

 

 

❍ 

✟ 

❅❅ ❍ ❅❆❆ 

❍ 

❅ 

❅ ❍ ❍❆ 

 

❍ ❍❅ 

❅ ❆ 

✟ 

✟✟✟✟✟ ❍ 

❅ ❍ 

 

❅ ❆ 

 

❅ ❆ 

❅❆ 

❅ 

❆ 

The circle picture element is produced with the commands 

\circle{diameter} 

\circle*{diameter} 

With the *-form of the command, a solid filled-in circle is printed rather 

than just an outline as for the standard form. Only certain sizes are 

available so L AT E X selects the one closest to the specified diameter entry. 

If the size is too small, a warning is output to the monitor and no circle 

is printed. 

★✥ 

✻ 

✉ ⑥ 

✧✦ 

\begin{picture}(3,1.6) 

\put(1,1){\circle*{0.2}} 

\put(1,1){\circle{1.2}} 

\put(1,1){\vector(0,1){0.6}} 

\put(2.5,1){\circle*{0.5}} 

\end{picture} 

The placement location in the corresponding \put command refers to the 

center of the circle.


Ovals and rounded corners 

The term oval is used here to mean a rectangle whose corners have been 

replaced by quarter circles; the largest possible radius is chosen for the 

circles such that the sides join together smoothly. The command to 

produce them is 

\oval(x dimen,y dimen)[part] 

The placement location in the corresponding \put command refers to the 

center of the oval. 

\put(3.0,0.75){\oval(4.0,1.5)} 

✤ 

✛ 

✻ 

1.5 UL 

Here we have set the values x dimen = 4.0 

4 UL ✲ 

✜UL 

and y dimen = 1.5 UL, whereby the unit 

length UL has been selected to be 0.8 cm. 

(3.0,0.75) The center of the oval is at the placement co- 

❄✣ 

✢ordinates 

in the \put command, at (3.0,0.75). 

The optional argument part takes on values t, b, l, or r, for making 

half ovals. 

[b] 

(1.75,4.2) 

✣ 

✛ 

✤ 3.5 UL 

✻ 

1.2 UL 

✢ 

✲ 

✜ 

\put(1.75,4.2){\oval(3.5,1.2)[b]} 

\put(1.75,2.6){\oval(3.5,1.2)[t]} 

\put(1.75,1.0){\oval(3.5,1.2)[l]} 

\put(3.25,1.2){\oval(3.5,1.2)[r]} 

[t] 

(1.75,2.6) 

 

1.2 UL The width and height specifications for half 

✲ 

❄ 

✤ 3.5 UL 

✻ 

(1.75,1.0) (1.2 UL) (3.25,1.0) 

[l] 

[r] 

✣ ❄ 

✛ 3.5 UL 

ovals are always those of the entire figure 

✜even 

though only part of it is being drawn. 

Similarly, the placement coordinates in the 

corresponding \put command refer to the 

✢center 

of the complete oval. (Unit length 

here is UL = 1 cm.) 

The argument part may also be one of the combinations tl, tr, bl, or 

br to generate a quarter oval. The order of the two letters is unimportant, 

so that lt, rt, lb, and rb are equally valid. 

✛ 3 UL ✲ 

✘\put(2.0,2.5){\oval(3.0,1.0)[tl]} 

(2.0,2.5) 

❄ [tl] 

1 UL 

 

(1.0,1.5) 

(2.5,2.5) 

[tr] 

 

(3.5,1.5) 

\put(2.5,2.5){\oval(3.0,1.0)[tr]} 

✻ \put(1.0,1.5){\oval(1.0,2.0)[bl]} 

\put(3.5,1.5){\oval(1.0,2.0)[br]} 

2 UL 

Once again the size specifications refer to the 

[bl] 

✚ 

[br] 

✛ 

1 UL 

entire oval and not just to the part that is drawn, 

✙and 

the placement coordinates in the \put command 

refer to the center of the complete oval. 

Quarter and half circles may also be drawn as partial ovals with equal 

width and height, but only up to a certain size. The following examples


demonstrate that sections of circles are possible up to a size of about 

1.5 cm. 

✬ ✩ 

★✥ 

\put(2.0,1.0){\oval(4.0,4.0)[t]} 

\put(2.0,1.0){\oval(1.5,1.5)[t]} 

 

 

★ ✥ 

\put(0.75,0.75){\oval(1.5,1.5)[bl]} 

\put(1.75,0.0){\oval(1.5,1.5)[tl]} 

\put(2.25,0.0){\oval(1.5,1.5)[tr]} 

✧ ✦ \put(3.25,0.75){\oval(1.5,1.5)[br]} 

Sections of ovals may be combined with other picture elements. The 

placement coordinates in the \put command, however, require some 

serious consideration to get them positioned properly. 

✛✘ 

 

❄ 

❄ 

 

✡ ✲ 

(2.75,0.5) 

\put(0.5,2.5){\oval(1.0,1.0)[t]} 

\put(0.0,2.5){\vector(0,-1){2.5}} 

\put(1.0,2.5){\vector(0,-1){1.5}} 

\put(0.5,2.5){\circle*{0.1}} 

\put(2.5,0.5){\line(0,1){2.5}} 

\put(2.75,0.5){\oval(0.5,0.5)[bl]} 

\put(2.75,0.25){\vector(1,0){1.25}} 

In all the above examples, the centers of the ovals have been marked 

with a black dot to indicate where they are. They are not normally a part 

of the \oval picture element. 

Exercise 13.6: Although the object 

pictured here is very offensive, it 

does make an excellent exercise for 

the picture environment. 

Hint: sizing and positioning can be 

worked out best by overlaying the 

illustration with transparent graph 

paper. 

Vertically stacked text 

✄ 

✂ 

UL=1 mm 

✤✤✜ 

✤✜✜ 

✉ ✉ ✉ ✉ ✉ ✉ ✉ ✉ 

♠ ♠ 

✣✣✢✉ 

✉ ✉ ✉ ✉ ✉ ✉ ✉ ✉✣✢✢ 

 

✂ 

❊ 

❊ 

❅ 

✂ 

✂ ✂ 

✞ 

✝ 

✎ 

✍ 

☛ ✟ 

It is sometimes necessary to write text vertically in a diagram, as in 

the example here at the right. This is carried out with the command 

\shortstack[pos]{col} 

The positioning argument can take on values l, r, or c. The standard 

is c for centered. The command is similar to a tabular environment 

with only one column. The text is entered as col, each row being 

separated from the next by \\. 

The \shortstack command is most frequently implemented for placing 

short lines of text inside a framed box, or for stacking single letters 

y –a 

x 

i 

s


vertically. The individual rows are separated from each other with the 

smallest possible vertical spacing. This means that rows with letters 

sticking up or down (like h and y) will have larger apparent gaps between 

them than rows without such letters. Adding \strut to each line, as in 

the third example, equalizes the vertical spacing. 

This 

spacing 

leaves 

some 

room 

for 

help 

Not 

really 

the 

best 

for words 

at all 

A strut 

in each 

line 

✟✟✯ ✟✟✯ 

is better 

✟✟✯ ✒ 

✻ 

(1.0,0.5) (3.0,0.5) (5.0,0.5) (8.0,0.8) (9.0,1.2) 

The placement coordinates of the corresponding \put command refer 

to the lower left corner of an imaginary box that contains the vertically 

stacked text. The first of the above texts is left justified, the second 

centered, and the third right justified. The last two on the right are 

centered. They were entered with 

\put(1.0,0.5){\shortstack[l]{This\\spacing\\leaves\\some\\...}} 

\put(3.0,0.5){\shortstack{Not\\really\\the\\best\\ ...}} 

\put(5.0,0.5){\shortstack[r]{A strut\strut\\in each\strut\\...}} 

\put(8.0,0.8){\shortstack{L\\e\\t\\t\\e\\r\\s}} 

\put(9.0,1.2){\shortstack{b\\e\\s\\t}} 

The \shortstack command may also be used outside of the picture 

environment within normal text. One possible application is for marginal 

notes, as in Section 4.10.5. 

Framed text 

The \framebox command generates a frame of a predetermined size in 

which text may be inserted at various positions (page 291). In text mode, 

there is the command \fbox for drawing a frame around text that fits 

it exactly (Section 4.7.1). This command is also available in the picture 

environment. 

The amount of spacing between the box frame and the enclosed text is 

given by the parameter \fboxsep. The placement of an \fbox by means 

of the \put command occurs in an unexpected manner, as shown below: 


\setlength{\fboxsep}{0.25cm} 

\put(0,0){\framebox(5,2){}} 

\put(1,1){\fbox{fitted frame}} 

\end{picture} 

✒ 

(1,1) 

L etters 

b est 

fitted frame 

The additional frame spacing is often unwanted in a diagram, especially 

if the frame surrounds a picture object rather than text. In this 

case, the command

\frame{pic elem} 


is used instead. The placement coordinate of the \put command then 

refers to the lower left corner as usual. 

W 

\put(0.0,0.5){\frame{TEXT}} 

TEXT 

O 

R 

\put(1.5,0.0){\frame{\shortstack{W\\O\\R\\D}}} 

D 

The contents of the \frame command can be any of the previous 

picture elements, and need not be merely text. However, in many cases 

the output comes out wrong. 

✛✘ 

\put(0,0){\frame{\vector(1,1){1.0}}} 

\put(2,0){\frame{\circle{1.0}}} ✒ 

✚✙ 

The first example produces the correct result, while the second has 

failed. In such cases, one can try putting the picture object inside a 

\makebox of suitable size and positioning as argument for the \frame 

command. However, it would then make more sense to use the \framebox 

command itself in place of \frame{\makebox...}. 

Curved lines 

Curved lines may be drawn in the picture environment with the commands 

\qbezier[num](x1,y1)(x2,y2)(x3,y3) 

which draw a quadratic Bézier curve from point (x1, y1) to (x3, y3) with 

(x2, y2) as the extra Bézier point. The curve is actually drawn as num + 1 

dots. The number of points num is an optional argument; if it is omitted, 

its value will be calculated to produce a solid-looking line. 

The meaning of the extra point can 

be illustrated with the example at 

(20,20) 

the right. The input is 


\qbezier(0,0)(20,20)(40,10) 

\end{picture} ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 

♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ ♣ 

(40,10) 

(0,0) 

The curve is drawn from (0,0) to (40,10) such that the tangents at the 

endpoints (the dotted lines) intersect at the extra Bézier point (20,20). 

Another way of stating this is that, as one moves from the first to the 

third point, one begins by heading directly towards the second point, and 

on arrival at the destination, one is moving directly away from that second 

point again. The dotted lines in the above example, which are not drawn 

by the \bezier functions, illustrate this. 

A dotted curve may be drawn by specifying the number of points num; 

some experimentation may be required to get just the right effect.


Line thickness 

There is a choice of two line thicknesses for the picture elements \circle, 

\oval, and \vector, as well as for sloping lines. These may be selected 

with 

\thicklines or \thinlines 

Each of these declarations remains in effect until countermanded by a 

call to the other one, or until the end of the environment in which it was 

invoked. Initially, \thinlines is in effect. 

The thickness of the horizontal and vertical lines may be set with the 

declaration 

\linethickness{thickness} 

to any desired size. The argument thickness is a positive length specification. 

With \linethickness{1.5mm}, all subsequent horizontal and 

vertical lines will have a thickness of 1.5 mm. 

Since frame boxes are constructed out of horizontal and vertical lines, 

the command \linethickness also affects the \framebox and \dashbox 


Saving parts of pictures 

It is possible to store a combination of picture elements as a sub-picture 

under a certain name, and to recall the whole set as often as one wants 

without having to reissue the individual commands every time. 

First each sub-picture must have a name reserved for it with 

\newsavebox{\sub pic name} 

which creates a box with the name \sub pic name for storing the picture. 

Afterwards, the sub-picture is saved with 

\savebox{\sub pic name}(x dimen,y dimen)[pos]{sub pic} 

where the arguments (x dimen,y dimen) and pos have the same meaning 

as for \makebox on page 291. 

If the picture commands sub pic are simply a piece of text, this command 

is exactly the same as the \makebox command except that the text 

is not printed but stored under the name \sub pic name. The sub pic 

may be set down anywhere within the main picture as a separate picture 

element. 

\usebox{\sub pic name} 

\newsavebox{\sub} 

\savebox{\sub}(2,1)[br]{\small Sub-Pic} 

.... 

\put(0.7,0.0){\frame{\usebox{\sub}}} 

\put(3.0,1.0){\frame{\usebox{\sub}}} 

1 UL 

✻ 

❄ 

✛ ✲ 

2 UL 

Sub-Pic 

Sub-Pic


This example may not seem very practical, since the \savebox and 

\usebox commands could have been replaced by a \framebox together 

with \multiput to achieve the same result with even less effort. However, 

the main advantage of these two commands is not for multiple setting of 

text, but rather for more complex sub pic compositions. 

It should be pointed out that a \savebox command may be given 

outside of the picture environment, and even within the preamble. Such 

a sub-picture is then available in all picture environments throughout 

the document. However, if a \savebox is defined within an environment, 

it keeps its contents only until that environment comes to an end. 

The picture elements inside a \savebox will be sized according to the 

value of \unitlength in effect at the time that the box is constructed. It 

will not be rescaled by a later change in \unitlength. 

13.1.5 Making graph paper 

Package: 

graphpap 

The package graphpap adds a new command for drawing gridded paper: 

\graphpaper[num](x,y)(lx,ly) 

which places a grid with its lower left corner at (x,y), which is lx units 

wide and ly units high. Grid lines are drawn for every num units, with 

every fifth one thicker and labeled. If num is not specified, it is assumed 

to be 10. All arguments must be integers, not decimal fractions. For 

example, \graphpaper(50,50)(200,100) produces 

150 

100 

(50,50) 

❅ 

50❅❘ 

50 100 150 200 250 

UL=0.3 mm 

Exercise 13.7: Produce a sheet of graph paper 10 cm by 15 cm with 2 mm 

separation between the lines. This sheet may be used to determine the positioning 

of picture elements when planning a new diagram. 

13.1.6 Shifting a picture environment 

The generalized syntax of the picture environment contains a further 

coordinate pair as an optional argument


\begin{picture}(x dimen,y dimen)(x offset,y offset) 

picture commands \end{picture} 

In this form, (x offset,y offset) specifies the coordinates of the lower 

left corner. This means that for all \put commands in the environment, 

the amounts x offset and y offset are effectively subtracted from the placement 

coordinates, so that the entire picture is shifted by x offset to the 

left and by y offset downwards. 

13.2 Extended pictures 

13.2.1 The epic package 

Package: With the intention of adding some higher level commands and creating 

epic a more user-friendly interface to the picture environment, Sunil Podar 

released his epic package, adding new features and enhancements. One 

idea is to be able to draw multiple objects relative to each other, with 

only a limited number of absolute coordinates, making reuse and shifting 

much easier. Another goal is to simplify the drawing of lines. 

This code was originally written for LAT EX 2.09, but functions just as 

well as a LAT EX 2ε package. Thus it is implemented with 

\usepackage{epic} 

It makes the following new commands available to the user: 

\multiputlist \matrixput \grid 

\dottedline \dashline \drawline 

\jput \picsquare \putfile 

as well as the environments: 

dottedjoin dashjoin drawjoin 

The \multiputlist command is a variation on the regular \multiput 

command; rather than placing the same element in several locations, it 

puts different elements at regularly spaced intervals. 

\multiputlist(x,y)(∆x,∆y)[pos]{Obj1, Obj2, . . . , ObjN } 

places the N picture elements Obj1, . . . , ObjN at (x,y), (x+∆x,y+∆y), 

and so on. They are actually put into a series of \makebox(0,0)[pos]{}, 

so that the optional pos argument specifies the location of the element 

relative to the plotted point (Section 13.1.4). The elements in the list are 

separated by commas, so they must be enclosed in {} if they themselves 

contain commas. 

The \matrixput command is the 2-D equivalent of \multiput, creating 

an array of a single picture element. Its syntax is:

13.2. Extended pictures 303 

\matrixput(x,y)(∆x1,∆y1){n1}(∆x2,∆y2){n2}{Obj} 

Thus, \matrixput(0,0)(3,5){3}(10,0){5}{\circle{1.5}} produces 

❝ ❝ ❝ ❝ ❝ 

❝ ❝ ❝ ❝ ❝ 

❝ ❝ ❝ ❝ ❝ 

The \grid command is similar to the \graphpaper command, but it 

will also optionally label the axes. 

\grid(width,height)(∆width,∆height)[X0,Y0] 

draws a grid of size width×height, with grid lines at intervals ∆width 

and ∆height. If the optional argument is given, the grid lines are labeled 

starting at X0, Y0, which must be integers, incremented by ∆width and 

∆height, which must also be integers in this case. The \grid command 

must be placed inside a \put command to specify its lower left corner: 

\put(0,0){\grid...}. 

A set of points can be joined by a dotted line with the command 

\dottedline[dot char]{dot gap}(x1,y1)(x2,y2). . . (xn,yn) 

where dot gap is the spacing (in \unitlength units) between the dots 

and dot char is the character used for the dot; if it not specified, the 

\picsquare (see below) is used by default, a small square. A near solid 

line can be achieved by making dot gap very small, but this could cause 

T E X to run out of memory. 

To join a set of points with a dashed line, one uses 

\dashline[stretch]{dash len}{dot gap}(x1,y1)(x2,y2). . . (xn,yn) 

where dash len is the length of the dashes (in \unitlength units), which 

are constructed as a series of dots to produce a solid looking dash. If the 

optional dot gap is given, then the dashes are made as a series of dots 

with that separation. The optional stretch is a number between −100 and 

infinity. If it is 0 or missing, the number of dashes is chosen so that 

the gaps between them are the same length as the dashes; if stretch is 

positive, the number of dashes is increased by stretch percent; if negative, 

it is reduced by that amount. With various combinations of stretch and 

dot gap, different dash patterns can be produced. 

To draw a solid-looking line through a set of points, one uses 

\drawline[stretch](x1,y1)(x2,y2). . . (xn,yn) 

which draws the lines as a series of line segments using the L AT E X line 

fonts. Since these have a limited number of angles, the result can appear 

fairly jagged. By increasing the optional stretch factor (again a percent), 

more segments are used and the appearance is improved; by decreasing 

it, the line can begin to look dashed. 

As an example of these three line drawing commands,


produces 


\dottedline[.]{3}(0,0)(10,10)(20,0) 

\dashline{3}(30,0)(40,10)(50,0) 

\drawline(60,0)(70,10)(80,0) 

\end{picture} 

. . . . . . . . . . . ❅ ❅ ❅ 

For each of the three line drawing commands, there is a corresponding 

environment, named dottedjoin, dashjoin, and drawjoin, each taking 

the same set of optional and mandatory arguments as the commands, 

except for the set of coordinates. Within these environments, one gives 

a set of \jput commands, which behaves exactly like the regular \put, 

except that all its elements are joined by the selected line type. For 

example, 

produces: 


\begin{dottedjoin}{2} 

\jput(0,0){\circle{3}} 

\jput(20,8){\makebox(0,0){\Large$\star$}} 

\jput(40,3){\makebox(0,0){\small Finish}} 

\end{dottedjoin} 

\begin{dashjoin}[20]{3} 

\jput(55,3){} 

\jput(70,10){\oval(10,5)} 

\jput(80,0){\circle*{2}} 

\end{dashjoin} 

\end{picture} 

⋆ 

. ❤. 

. . . . . . . . . . . . . . . . . . . 

Finish 

. . 

Real life data curves are generated by other software programs. These 

results can be imported into the picture environment by storing them 

in a data file, named say mycurve.put and then issuing 

☛ 

✡ 

\putfile{mycurve.put}{\picsquare} 

The data file contains a list of x y coordinates, one pair per line, with 

possible comments beginning with % as usual. The \putfile command 

places its second argument at all the points listed in the data file. 

The default dot character for plotting all lines is the \picsquare, a 

black square that scales with the current line thickness. A different dot 

character may be used with \putfile if one wishes. It is also possible to 

specify the dot character for \dottedline, but the default is \picsquare. 

✟ 

✠ 

✉

13.2. Extended pictures 305 

The default values for the optional stretch argument in \dashline and 

\drawline are initially 0, but may be changed at any time by redefining 

(with \renewcommand) \dashlinestretch and \drawlinestretch, respectively. 

This makes it possible to revise a whole set of curves with one 

command, rather than manually changing each one. 

13.2.2 The eepic package 

Package: The epic commands are still subject to many of the limitations of the 

eepic picture environment: limited number of slopes for lines, limited line 

thicknesses, restrictions on circle sizes. This is because the picture 

environment uses LAT EX fonts to ‘draw’ inclined lines and circles, making 

the results portable to all output devices. 

The eepic package, by Conrad Kwok repairs these problems by employing 

graphic commands executed by the DVI driver itself. They are 

transmitted to the .dvi file by \special commands, which are driverspecific. 

Thus greater flexibility is achieved, but at the price of loss 

of portability. The package is therefore most suitable for use with the 

PostScript dvips driver, as well as with the previewers xdvi and windvi. 

It does not work with pdfTEX, unfortunately; on the other hand, it has no 

problems with dvipdfm. 

However, eepic not only recodes the features of epic, it also adds 

some additional functionality. If one wishes to invoke these extra features 

with a driver that does not recognize the graphics \specials, one can use 

the emulation package eepicemu instead. A comparison of the results 

with these two packages is shown in Figure 13.1 on the next page. 

In either case, one must load epic with one of eepic or eepicemu: 

\usepackage{epic,eepic} or \usepackage{epic,eepicemu} 

The following standard picture elements are modified by the eepic 

package: 

\line(∆x,∆y){length} (page 293) where ∆x and ∆y may take on any 

positive or negative integer values, and not just those between ±6. 

\circle{diameter} (page 295) may have any value for diameter; the 

same for \circle*. 

\oval (page 296) may have the maximum diameter of the corners set to 

any value; this is stored in the length \maxovaldiam which the user 

may change; the default value is 40 pt. 

Similarly, all the extra commands from epic package are also revised 

internally. The restrictions on slopes never apply directly to them, but 

lines of arbitrary slope are drawn with little segments at the nearest 

available slope, or as series of dots, requiring much computer time and


¡ 

¢ 

¢ 

¡ 

 

£¥¤§¦¥¨©¤ 

 

¡ 

¢ ¡¤£ ¡¤£ ¡¤£ ¡¥£ ¡¤£ 

¡ £ ¤¢ ¤¢ ¥¢ ¥¢ ¤¢ 

 

 

 

¨¨ 

§ § ¦ ¦ ¦ ¦ ¦ 

© © 

 

¨ 

¨ 

 

 

 

§ 

§ 

§ 

§ 

§ 

§ 

§ 

§ 

¦ 

¢ 

§ 

 

 

 

Figure 13.1: Comparison of results using the true eepic drawing functions 

(top) and those with the emulation in eepicemu (bottom). (This 

example is taken from the eepic user’s manual.) 

overloading the output file. With eepic, these lines are drawn more 

efficiently. 

Additional features added by eepic (and eepicemu, even if they do 

not work very well): 

\allinethickness{dimen} 

sets the line thickness to dimen for all picture elements, including 

circles, ovals, splines, not just for straight lines; 

\Thicklines 

sets thickness of straight lines to 1.5 times that of \thicklines; 

\path(x1,y1)(x2,y2). . . (xn,yn) 

is a fast version of \drawline, without the stretch option; a solid 

line is always drawn; 

\spline(x1,y1)(x2,y2). . . (xn,yn) 

draws a curve from the first to last point, with intermediate 

points as control points; 

\ellipse(x diam,y diam) 

draws an ellipse with horizontal and vertical sizes x diam, y diam 

respectively; there is also *-form to produce a solid ellipse; 

\arc{diameter}{start}{end} 

draws the arc of a circle with diameter (in \unitlength units) 

from angle start to end, in radians, clockwise from 0 pointing 

to the right; start must be from 0 to 2π, and end from start to 

start+2π. 

¡ 

¢

13.3 Other drawing packages 

13.3.1 The XY-pic package 

13.3. Other drawing packages 307 

As indicated at the beginning of this chapter, an exhaustive description of 

many other specialized packages, for drawing or using METAFONT fonts, 

can be found in The L AT E X Graphics Companion by Goossens et al. (1997). 

Most of these are designed for T E X in general, but will also work with L AT E X. 

We mention two of these here just to show the possibilities. 

The XY-pic package can carry out general drawing entirely with extra 

METAFONT fonts, making it fully portable among all output drivers. The 

fonts are also available in type 1 coding (Section G.6) so that the package 

works just as well with PDF output. It was created by Kristoffer H. Rose 

and then extended by Ross Moore and others. 

Even an overview of XY-pic is beyond the scope of this book. We recommend 

the supplied User’s Guide and more extensive Reference Manual 

both to be found in texmf\doc\generic\xypic. The descriptions and 

illustrations in Chapter 5 of Goossens et al. (1997) are especially helpful. 

To give a flavor of the syntax in XY-pic, we give the following examples: 

\xymatrix{ 

U \ar@/_/[ddr]_y \ar@/ˆ/[drr]ˆx \ar@{.>}[dr]|-{(x,y)} \\ 

& X \times_Z Y \ar[d]ˆq \ar[r]_p & X \ar[d]_r \\ 

& Y \ar[r]ˆg & Z} 

\xy /r1.5pc/:,+*+{P};p 

@(,+(2,2)*{+}@+, +(2,-2)*{+}@+ 

,+(2,2)*{+}@+, +(2,0)*+{C}="C" 

,*\qspline{},"C",**\crvs{.} 

,@i @)\endxy 

that produce the diagrams: 

U 

x 

(x,y) 

+ 

+ C 

 

 

X ×Z Y 

y 

p X 

 

q 

r P 

+ 

 

 

g 

Y 

Z 

The first example, with \xymatrix, organizes its contents in rows and 

columns, not dissimilar to the tabular environment. In this case, the 

first row contains U in the first column, and the rest are empty; the second 

row has an empty first column (starts with &, the column separator) while 

the next two columns contain X \times_Z Y and X; the third row also 

has an empty first column, with entries Y and Z in the remaining columns. 

Everything else describes how these nodes are connected. For example,


\ar@/_/[ddr]_y means ‘draw an arrow, curving down to the node located 

two down and one right, and subscript it with y.’ And so on for the other 

connections. 

The second example, with \xy. . . \endxy gives more complicated drawing 

instructions, placing objects at given coordinates, joining them with 

arrows, lines, curves. As can be seen from both of these examples, it takes 

some practice to learn the syntax and to exploit XY-pic fully. 

13.3.2 PSTricks 

In Chapter 10 we describe how the PostScript language plays a very important 

role in T E X and L AT E X by making a wide spectrum of fonts available 

to complement the original Computer Modern fonts. But PostScript is 

also a powerful plotting language. It was in order to exploit the PostScript 

possibilities directly that Timothy van Zandt wrote the PSTricks package 

for direct use within T E X, and thus within L AT E X. 

Actually PSTricks consists of over a dozen packages. The complete 

functionality may be loaded with the package pstricks, or the individual 

packages may loaded singly. (Since PSTricks is really a T E X utility, the L AT E X 

package pstricks.sty does nothing more than load the pstricks.tex 

file containing the actual code.) Some of these packages are: 

pst-3d 3-D drawing 

pst-blur adding blurred shadows 

pst-char stroking and filling character paths 

pst-coil coil and zigzag objects 

pst-eps export objects to EPS files 

pst-fill adds \psboxfill command 

pst-gr3d drawing 3-D grids 

pst-grad gradient color fills 

pst-lens simulating effect of a lens 

pst-node placing and joining nodes 

pst-osci emulating oscilloscopes 

pst-plot plotting data 

pst-poly drawing polygrams 

pst-slpe improvement on pst-grad 

pst-text setting text along a path 

pst-tree tree commands 

As with XY-pic, it is beyond the scope of this book on document production 

with L AT E X to describe the extensive plotting features of PSTricks. 

We refer to Goossens et al. (1997), Chapter 4 for a description of most of 

these packages (some have been added since then) and otherwise to the 

supplied manuals and documentation to be found in texmf\generic\ 

source\pstricks.

14 

Bibliographic 

Databases and BIBT E X 

In academic publications, a bibliography or list of references is standard 

procedure. In Section 9.3 we describe how a bibliography can be formatted 

with the thebibliography environment and how its entries are referred 

to within the text. Often authors find that they are constantly referring to 

the same publications in most of their own papers. Similarly, researchers 

working in the same field will frequently be referring to the same set of 

papers. This means that many of the entries in the thebibliography 

environment will be repeated from one work to the next, or even among 

co-workers at one institute. 

It would be very useful if the bibliographic entries could be stored in 

one database file once and for all to be available for all documents with 

a list of references in that field. Such a database system is possible with 

the BIBT E X program supplied with the L AT E X installation. The information 

about the various publications is stored in one or more files with the 

extension .bib. For each publication there is a key that identifies it, and 

which may be used in the text document to refer to it. Such a file is called 

a bibliographic database. 

14.1 The BIBT E X program 

BIBT E X is an auxiliary program to L AT E X that automatically constructs a 

bibliography for a L AT E X document by searching one or more databases. 

To this end, the L AT E X file must contain the command 

\bibliography{database1, database2, . . . } 

at that point in the text where the bibliography is to appear. Here the 

argument database1, database2, . . . is a list of root names, separated by 

commas, of the bibliographic database files that are to be searched. The 

extension.bib is not explicitly written. 

Reference can be made to a publication in one of the databases at any 

place in the text with the commands 

309

310 Chapter 14. Bibliographic Databases and BIBT E X 

\cite{key} 

\citep{key} 

\citet{key} 

as is explained in Sections 9.3.3 and 9.3.4. The key is the database identifier 

for that publication, which of course the user must know beforehand. 

After the first L AT E X processing, the BIBT E X program must be run either by 

clicking the appropriate icon in the editor shell, or from a command line 

with bibtex plus the root name of the L AT E X file. Supposing this source 

file name is comets.tex, the call 

bibtex comets 

produces a new file with the name comets.bbl, containing the extracted 

information for those publications for which there was a \cite reference, 

packaged in a thebibliography environment to be input into the 

document on the next L AT E X run. 

Occasionally the bibliography is to include publications that were not 

referenced in the text. These may be added with the command 

\nocite{key} 

given anywhere within the main document. It produces no text at all 

but simply informs BIBT E X that this reference is also to be put into the 

bibliography. With \nocite{*}, every entry in all the databases will be 

included, something that is useful when producing a list of all entries and 

their keys. 

After running BIBT E X to make up the.bbl file, it is necessary to process 

L AT E X at least twice to establish both the bibliography and the in-text reference 

labels. The bibliography will be printed where the \bibliography 

command is issued; it in fact inputs the.bbl file. 

The formatting of the bibliography may be selected with the declaration 

\bibliographystyle{style} 

which may be issued anywhere in the document. Its only function is to 

inform BIBT E X what bibliography style file (.bst) is to be loaded to determine 

the resulting format. Many contributed.bst files exist, designed for 

specific journals and publishing houses, and it is possible to create one’s 

own (Section 14.3). However, the supplied standard ones for numerical 

citation only are: 

plain The entries in the bibliography are ordered alphabetically; each 

is assigned a running number in square brackets as the in-text 

reference marker, printed where the \cite commands are issued. 

unsrt The entries are ordered according to their first references by the 

\cite and \nocite commands. The entry for the first \cite

14.2. Creating a bibliographic database 311 

receives the number 1, that of the next \cite with a different 

key the number 2, and so on. The markings and listings are 

otherwise the same as for plain. 

alpha The ordering in the bibliography is the same as for plain but the 

markers are an abbreviation of the author’s name plus year of 

publication. A reference to Smith (1987) would appear as [Smi87] 

instead. 

abbrv The ordering and marking are the same as for plain, but the 

bibliographic listing is shortened by abbreviating first names, 

months, and journal names. 

For author–year citations and the natbib package (Section 9.3.4), 

there exist the bibliographic styles plainnat, unsrtnat, and abbrvnat, 

producing the same bibliography format as the corresponding standard 

styles, but with natbib compatibility. 

14.2 Creating a bibliographic database 

Creating a bibliographic database might seem like more work than typing 

up a list of references with the thebibliography environment; the great 

advantage is that the entries need to be included in the database once 

and only once and are then available for all future publications. Even 

if a different bibliography style is demanded in later works, all the information 

is already on hand in the database for BIBT E X to write a new 

thebibliography environment in another format. In fact, we feel that 

even for a single document, it is simpler to make an entry into the database 

than to adhere to the very precise and fiddly requirements of a literature 

list, especially regarding punctuation and positioning of the authors’ initials. 

The database entry proceeds very quickly and easily if one has a 

generalized template, as illustrated in Section 14.2.6. 

The entries in a bibliographic database are of the form 

} 

@BOOK{knuth:86a, 

AUTHOR = "Donald E. Knuth", 

TITLE = {The \TeX{}book}, 

EDITION = "third", 

PUBLISHER = "Addison--Wesley", 

ADDRESS = {Reading, Massachusetts}, 

YEAR = 1986 

The first word, prefixed with @, determines the entry type, as explained 

in the next section. The entry type is followed by the reference information 

for that entry enclosed in curly braces { }. The very first entry is the key


for the whole reference by which it is referred to in the \cite command. 

In the above example this is knuth:86a. The key may be any combination 

of letters, numerals, and symbols, except commas. The actual reference 

information is then entered in various fields, separated from one another 

by commas. Each field consists of a field name, an = sign, with optional 

spaces on either side, and the field text. The field names shown above are 

AUTHOR, TITLE, PUBLISHER, ADDRESS, and YEAR. The field text must be 

enclosed either in curly braces or in double quotation marks. However, 

if the text consists solely of a number, as for YEAR above, the braces or 

quotation marks may be left off. 

For each entry type, certain fields are required, others are optional, and 

the rest are ignored. These are listed with the descriptions of the various 

entry types below. If a required field is omitted, an error message will 

occur during the BIBT E X run. Optional fields will have their information 

included in the bibliography if they are present, but they need not be 

there. Ignored fields are useful for including extra information in the 

database that will not be output, such as a comment or an abstract of the 

paper. Ignored fields might also be ones that are used by other database 

programs. 

The general syntax for entries in the bibliographic database reads 

@entry type{key, 

field name = {field text}, 

.... 

field name = {field text} } 

The names of the entry types as well as the field names may be written 

in capitals or lower case letters, or in a combination of both. Thus @BOOK, 

@book, and @bOOk are all acceptable variations. 

The outermost pair of braces for the entire entry may be either curly 

braces { }, as illustrated, or parentheses ( ). In the latter case, the 

general syntax reads 

@entry type(key, ... ... ) 

However, the field text may only be enclosed within curly braces {...} or 

double quotation marks "..." as shown in the example above. 

14.2.1 The entry types 

The following is a list of the standard entry types in alphabetical order, 

with a brief description of the types of works for which they are applicable, 

together with the required and optional fields that they take. The 

meanings of the fields are explained in the next section. 

@article Entry for an article from a journal or magazine. 

required fields author, title, journal, year. 

optional fields volume, number, pages, month, note.


@book Entry for a book with a definite publisher. 

required fields author or editor, title, publisher, year. 

optional fields volume or number, series, address, edition, 

month, note. 

@booklet Entry for a printed and bound work without the name of a 

publisher or sponsoring organization. 

required fields title. 

optional fields author, howpublished, address, month, year, 

note. 

@conference Is the same as @inproceedings below. 

@inbook Entry for a part (chapter, section, certain pages) of a book. 

required fields author or editor, title, chapter and/or 

pages, publisher, year. 

optional fields volume or number, series, type, address, 

edition, month, note. 

@incollection Entry for part of a book that has its own title. 

required fields author, title, booktitle, publisher, year. 

optional fields editor, volume or number, series, type, 

chapter, pages, address, edition, month, 

note. 

@inproceedings Entry for an article in conference proceedings. 

required fields author, title, booktitle, year. 

optional fields editor, volume or number, series, pages, 

address, month, organization, publisher, 

note. 

@manual Entry for technical documentation. 

required fields title. 

optional fields author, organization, address, edition, 

month, year, note. 

@mastersthesis Entry for a Master’s thesis. 

required fields author, title, school, year. 

optional fields type, address, month, note. 

@misc Entry for a work that does not fit under any of the others. 

required fields none. 

optional fields author, title, howpublished, month, year, 

note. 

@phdthesis Entry for a PhD thesis. 

required fields author, title, school, year. 

optional fields type, address, month, note.


@proceedings Entry for conference proceedings. 

required fields title, year. 

optional fields editor, volume or number, series, address, 

month, organization, publisher, note. 

@techreport Entry for a report published by a school or other institution, 

usually as part of a series. 

required fields author, title, institution, year. 

optional fields type, number, address, month, note. 

@unpublished Entry for an unpublished work with an author and title. 

required fields author, title, note. 

optional fields month, year. 

Each entry type may also have the optional fields key and crossref. 

The former provides additional information for alphabetizing the entries, 

especially when the author information is missing. The author information 

is normally found in the author field but may also be in the editor 

or even organization fields. This key field has nothing to do with the 

key for identifying the entry in the \cite command. The crossref field 

gives the key for another entry in the database that shares many of the 

information fields, as illustrated in Section 14.2.3. 

14.2.2 The fields 

The fields that may be used within a bibliographic entry are listed below 

together with their meanings. They are always given in the form: 

field name = {field text} or 

field name = "field text" 

address 

The address of the publisher or other institution. For major 

publishing houses, it is sufficient to give just the city. For smaller 

publishers, giving the full address is recommended. 

annote An annotation that may be used by non-standard bibliography 

styles to produce an annotated bibliography. The standard BIBT E X 

styles ignore it. 

author The name(s) of the author(s) as described in Section 14.2.4. 

booktitle 

The title of a book when only part of it is being cited. See 

Section 14.2.4 for special considerations on capitalization. 

chapter 

A chapter or section number.


crossref 

The key of another entry in the database that shares many of the 

same field entries. See Section 14.2.3. 

edition 

The edition of a book, usually written in full and capitalized, 

as ‘Second’. The standard styles will change it to lower case as 

necessary. 

editor The name(s) of the editor(s) as described in Section 14.2.4. If 

there is also an author field, then this gives the editor of the 

book or collection in which the citation appears. 

howpublished 

States anything unusual about the method of publishing. Should 

be capitalized. Example: ‘Privately published’. 

institution 

The name of the sponsoring institution for a technical report. 

journal 

The name of a journal or magazine. Abbreviations are provided 

for the most common ones (see Section 14.2.5). 

key An addition for alphabetizing purposes when the author information 

is missing. This is not the same as the key for identifying 

the entry to a \cite command. 

month The month in which the work was published or, if unpublished, 

when it was written. Abbreviations exist in the form of the first 

three letters of the (English) names. 

note Any additional information that should be added. Capitalize the 

first letter. 

number The number of a journal, magazine, technical report, or work in 

a series. Journals are usually identified by volume and number; 

technical reports are issued a number by the institution; books 

in series sometimes have a number given to them. 

organization 

The sponsoring organization for a conference or a manual. 

pages A page number or a range of pages, in the form 32,41,58 or 

87--101 or 68+. The last form indicates page 68 and following 

pages. A single hyphen given for a range will be converted by 

the standard styles to the double hyphen to produce a dash, as 

‘87–101’. 

publisher 

The publisher’s name.


school The name of the academic institution where a thesis was written. 

series The name of a series or set of books. When citing a book from 

a series, the title field gives the name of the book itself while 

the optional series specifies the title of whole set. 

title The title of the work, obeying the capitalization rules listed in 


type The type of technical report, for example, ‘Research Note’. 

url The universal resource locator, or Internet address, for online 

documents; this is not standard but supplied by more modern 

bibliography styles. 

volume The volume number of a journal or multi-volume book. 

year The year in which the work was published or, if unpublished, in 

which it was written. It should normally consist of four numerals, 

such as 1993. 

Additional field names may be included, which BIBT E X will simply 

ignore. For example, to add the abstract of an article in the database, 

abstract = {text of an abstract} 

This can be of use in other applications as well as for database management. 

14.2.3 Cross-referencing fields 

If many entries in the bibliographic database share a common set of field 

information, such as for a number of works all appearing in the same 

conference proceedings, it is possible to refer to another entry containing 

that common set with the crossref field. For example, 

@INPROCEEDINGS{xyz-1, 

crossref = {xyz-proceedings}, 

author = {J. S. Jones}, 

title = {The First Results from the {Appleville Experiment}}, 

pages = {34--38} } 

. . . . . . . . . . . . . 

@PROCEEDINGS{xyz-proceedings, 

editor = {C. H. Kelvin}, 

title = {Proceedings of the First Conference on the 

{Appleville Experiment}}, 

booktitle = {Proceedings of the First Conference on the 

{Appleville Experiment}} 

year = 1991 }


The first entry, with key xyz-1, is to obtain all its missing fields from 

the second entry, by means of the crossref field referring to the entry 

with key xyz-proceedings. The missing fields are editor, booktitle, 

and year, those that are common to all articles in the conference proceedings. 

Note that booktitle is an ignored field for @PROCEEDINGS but 

needs to be included here since it is required for @INPROCEEDINGS. 

If an entry is referred to by two or more other entries, then it too will 

be included in the bibliography, even though its key never appeared as 

the argument of a \cite or \nocite command. 

In order for this system to function properly, the entry that is referred 

to must appear in the database(s) after all those that refer to it. It is 

therefore recommended to put all such referenced entries at the end of 

the database. Cross-references may not be nested. 

14.2.4 Special field formats 

There are some special rules for entering the texts to the fields author, 

editor, title, and booktitle. BIBT E X will process names, putting surnames 

first, abbreviating given names with initials, and so on, according 

to the instructions in the style file. Thus it is important that the program 

knows what is a given name and what a surname. Similarly for titles, capitalization 

may change depending on style and/or entry type, so BIBT E X 

must know what words are always capitalized. 

Names 

The names in the author and editor fields may be typed in either in 

the form {Given Names Surname} or as {Surname, Given Names}. That 

is, BIBT E X assumes that if there is no comma the last capitalized name is 

the surname, or family name; otherwise what comes before the comma is 

taken to be the surname. Thus the name texts "John George Harrison" 

and "Harrison, John George" are equivalent for Mr J. G. Harrison. 

However, if a person has a double surname, without a separating hyphen, 

the second form must be employed, or the double name must be enclosed 

in braces, as 

"San Martino, Maria" or "Maria {San Martino}" 

for Ms M. San Martino. 

Auxiliary words to a surname that are not capitalized, such as von or 

de, may be entered in either form: 

"Richard von Mannheim" or "von Mannheim, Richard" 

"Walter de la Maire" or "de la Maire, Walter" 

Anything enclosed in braces will be treated as a single item, something 

that is used in ambiguous cases, or when the name contains a comma or 

the word and. An example is


"{Harvey and Sons, Ltd}" 

If the name contains a Junior or some other addition, it must be 

entered {Surname, Junior, Given Name}, for example as 

"Ford, Jr, Henry" or "Ford, III, Henry" 

However, if there is to be no comma, then the Jr must be treated as part 

of a double surname, something that is not recommended at all: 

"{Filmore Jr}, Charles" or "Charles {Filmore Jr}" 

Accents within a name formed with a backslash command should 

be enclosed in braces with the backslash as the first character after the 

opening brace. In this way, the alphabetization and the formation of labels 

with the alpha bibliography style will function properly. For example, 

author = "Kurt G{\"o}del", 

year = 1931 

will produce the label [Göd31] as desired. The accent text must not be 

enclosed any deeper than shown here. 

Accents should be formed with the backslash command for most 

generality. Some language modifications have shorthands for accented 

letters (like "a instead of \"a for German) but these should be avoided in 

the database to ensure that the results will be universally understood. 

Hyphenated first names are properly abbreviated. Thus "Jean-Paul 

Sartre" becomes ‘J.-P. Sartre’. 

If an author or editor field is to contain more than one name, the 

names are separated by the word and. For example, 

author = "Helmut Kopka and Daly, Patrick William" or 

AUTHOR = {Peter C. Barnes and Tolman, Paul and Mary Smith} 

If the and is actually part of the name, the whole name must be enclosed 

in braces, as pointed out above. 

Do not insert any ˜ characters between names or initials; BIBT E X will 

do this automatically as it deems fit. 

If only initials are given for the authors given names, insert a blank between 

them, as P. W. Daly, not P.W. Daly. The latter will be interpreted 

as a single given name and will be abbreviated as ‘P. Daly’ 

If the author list is too long to type in all the names, it may be 

terminated with and others. This will be converted to the form of et al. 

prescribed in the style file. 

Titles 

The capitalization of the title depends on the bibliography style: usually 

book titles are capitalized while article titles are not. The text in the fields


title and booktitle should be written in the capitalized form so that 

BIBT E X can change to lower case as required. 

The general rules for capitalizing titles in English state that the first 

word of the title, the first word after a colon, and all other words are 

capitalized except articles and unstressed prepositions and conjunctions. 

For example: 

title="The Right Way to Learn: A Short-Cut to a Successful Life" 

Words that are always to be capitalized, such as proper nouns, must 

be enclosed in braces. It is sufficient to enclose only the letter that must 

be capitalized. The two following examples are equivalent: 

14.2.5 Abbreviations 

title = "The {Giotto} Mission to Comet {Halley}" or 

TITLE = {The {G}iotto Mission to Comet {H}alley} 

It is always possible to use an abbreviation in the field text in place of 

actual text. Some abbreviations, such as the names of the months and 

some standard journal names, are already available, and the user is free 

to define new ones for his or her personal use. 

The name of an abbreviation consists of any combination of letters, 

numbers, and symbols except 

" # % ’ ( ) , = { } 

An abbreviation is defined with the command 

@string{abbrev name = {text}} or 

@string(abbrev name = {text}) 

where abbrev name stands for the name of the abbreviation and text is 

the replacement text. For example, if the abbreviation 

@string{JGR = {Journal of Geophysical Research}} 

has been defined, the following two field statements are identical: 

journal = JGR 

journal = {Journal of Geophysical Research} 

The name of the abbreviation is not enclosed in braces or double quotes, 

otherwise the name itself will be interpreted literally as the field text. In 

both the @string command and the name of the abbreviation, the case 

of the letters is unimportant. The above abbreviation could just as well 

be defined as 

@STRING{jgr = {Journal of Geophysical Research}} 

or @StrinG{jGr = {Journal of Geophysical Research}}


and it may be referred to in the fields as JGR, JGr, JgR, Jgr, jGR, jGr, jgR 

or jgr. 

Abbreviations may be concatenated with the symbol # between them. 

Thus after giving 

@string{yrbk = {Institute Yearbook}} 

this abbreviation may be combined with other abbreviations and text, as 

in 

title = "Max-Planck˜" # yrbk # 2000 

to produce ‘Max-Planck Institute Yearbook 2000’. 

Standard abbreviations exist for the months of the year, the names 

consisting of the first three letters of the name, as jan, feb, etc. These 

are actually included within the .bst file, and can change their definition 

for styles intended for other languages. However, the name of the 

abbreviation remains the same, the first 3 letters of the English name. 

Further predefined abbreviations are available (in the .bst file) for 

some standard journal names. Those produced with custom-bib are 

especially extensive. See Section 14.3. 

The @string commands may be issued anywhere in the database 

between two bibliographic entries, but must already be defined before 

they are used. Therefore, it makes most sense to insert all the abbreviation 

definitions at the beginning of the database file. 

14.2.6 Using a template 

Typing in the text for bibliographic entries into a database might seem to 

be rather complicated since there are so many things to remember, such 

as which fields are required and which are optional, what their formats 

are, and so on. One way to simplify this task is to make up templates 

for the most common entry types that you use. A template is basically a 

complete entry with all the fields left blank. It is stored in a separate file 

on its own, to be inserted into the database file whenever a new entry of 

that type is to be made. Then the field texts are written in. 

An appropriate template for the entry type @article could be 

@ARTICLE{, 

AUTHOR = {}, 

TITLE = {}, 

YEAR = {}, 

JOURNAL = {}, 

volume = {}, 

number = {}, 

month = {}, 

pages = {},

} 

note = {} 

14.3. Customizing bibliography styles 321 

in which the is a reminder that the keyword is to go here, the 

required fields come first and are capitalized, while the optional fields are 

in lower case. 

Finally, we mention that BIBT E X has been written by Oren Patashnik, 

Stanford University, in close association with Leslie Lamport. The BIBT E X 

installation should contain the two files btxdoc.tex and btxhak.tex for 

additional information about BIBT E X. In particular, btxhak.tex contains 

instructions for writing bibliographic style files. Process these files with 

L AT E X, if the resulting.dvi or.pdf files are not already there. 

14.3 Customizing bibliography styles 

It seems that every journal and publishing house has its own set of rules 

for formatting bibliographies. The differences are really minor, such as 

whether to use commas or colons here, or to put volume numbers in bold 

or italic typeface. Nevertheless, in spite of this triviality, the rules are 

rigid for each house. 

Standard L AT E X provides only four bibliography styles, and their differences 

have more to do with sorting and labeling than with such fussy 

details. It is possible to design one’s own .bst file, by modifying existing 

ones; however, this requires a knowledge of the peculiar BIBT E X 

programming language, something that puzzles even experienced L AT E X 

users. There are some 50.bst files in the BIBT E X directories of the T E X file 

servers, and probably many more in contributed L AT E X directories. Each 

is designed for a specific journal and there is no guarantee that it would 

be accepted by another one. What is a user to do if his or her publisher 

demands a certain format that just is not to be found? How can one seek 

a given format anyway among those offered? 

Some help can be found in the custom-bib application written by 

Patrick W. Daly. The heart of this package is a generic (or master) bibliography 

style merlin.mbs containing alternative coding for a multitude of 

different bibliographic features or options. It is processed with the Doc- 

Strip program which is capable of including alternative coding according 

to selected options (see Section D.7.1). 

Because of the large number of options offered (well over 100), a menuguided 

interface is provided, called makebst.tex. When processed under 

T E X or L AT E X, this ‘program’ first asks which .mbs file is to be read, and 

then interactively constructs a DocStrip batch job to produce the selected 

bibliographic features according to menu information contained in the 

.mbs file itself. This means that there could be a number of .mbs files


to choose from, each one explaining its set of options to the user via 

makebst. 

Support for other languages is provided by means of additional .mbs 

files, one for each language, which contain the translations for such words 

as volume, editor, edition, and so on. 

Some of the features provided by the supplied generic bibliographic 

style file are: 

• numerical or author–year citations; in the latter case, one may also 

choose which \bibitem style is to be used; 

• order of references: by citation order, alphabetical by all authors, 

alphabetic by author–year label; 

• format of authors’ names: first name plus surname, initials plus 

surname, surname plus initials, reversed initials only for first author, 

and more; 

• number of author names to include before giving et al.; 

• typeface to be used for the author names; 

• position of date, parentheses or brackets around the year; 

• format of volume, number, and pages for journals; 

• capitalization of article titles as sentence or title style; 

• whether to use the word and or an ampersand &; 

• placement of commas with the word and; 

• whether to abbreviate editor, volume, chapter, etc.; 

• first and last page numbers or only first; 

• optional addition of shorthand designations of common journal 

names; 

• and much more. 

Abbreviations for many journal names in physics, optics, and geophysics 

can be provided in those .bst files produced by custom-bib, 

which then may be used in BIBT E X databases. It is unfortunate that these 

are to be found in the .bst files themselves, which means that they may 

not be universally present. Which ones are present depends on the options 

chosen by the producer of the style file. The file shorthnd.tex in 

the custom-bib installation lists all the current possible abbreviations. 

(Actually, it is generated by running L AT E X on shorthnd.ins, which then 

produces the most current list from the existing .mbs files.) To check 

what any particular .bst file contains, open it and look for the MACRO 


15 

Presentation Material 

So far we have been assuming that the output of a L AT E X project is to 

be printed paper or an electronic document looking like printed paper 

displayed on the computer monitor. 

Another important type of output is presentation material, the visual 

support displayed in front of an audience during an oral presentation. 

Traditionally, this consisted of a set of slides to the projected onto a 

screen, but in more recent times this has been replaced by viewgraphs 

(or transparencies) and an overhead projector, giving the speaker much 

more interactive control over the presentation. However, the truly modern 

presentation is done electronically, with the entire presentation stored in 

a computer directly connected to the projector; there are no slides to fall 

out of the cassette, or be inserted the wrong way around, no viewgraphs 

to spill out onto the floor. The only thing that can go wrong is that 

the computer refuses the connection, needs rebooting, or the necessary 

display program is missing. Such teething problems are occurring less 

frequently now that this form of presentation is becoming standard. 

In spite of the change in medium away from the 2 ′′ × 2 ′′ pieces of 

film, a computer presentation is still called a slide show, a term which 

today presupposes the electronic medium. We will use the word slide 

more generally to mean a page of text and/or graphics for presentation, 

whether it is to be printed on paper and then photographically rendered 

to film or transparency, or to be directly projected from the computer. 

The preparation of such a slide with L AT E X is much the same regardless 

of the final projection method; direct projection does offer additional 

features which we also address. 

We start this chapter with the standard L AT E X method for slide preparation, 

the slides class, and then describe the far more advanced seminar 

class (Section 15.2). The latter is especially suitable for interesting PDF 

electronic presentations. We then illustrate pdfscreen (Section 15.3), a 

package designed more for documents meant for electronic viewing but 

which can also produce slides. Finally we show in Section 15.4 how PDF 

presentations may be enhanced with the help of an additional program. 

323

324 Chapter 15. Presentation Material 

15.1 Slide production with SLIT E X 

The term SLIT E X was used in the old L AT E X 2.09 system to indicate a 

separate specialized version of L AT E X for producing projection material. 

This awkward and cumbersome device was needed then because of the 

way fonts were hardwired into the formats. With L AT E X 2ε, this is no longer 

necessary, and the entire functionality is now found within the slides 

class, considerably streamlined over the obsolete version, running under 

regular L AT E X, or pdfL AT E X. The word SLIT E X therefore should be retired, 

but since is is a nifty expression, we will continue to use it to distinguish 

the slides class, which is considerably different from article and the 

other standard classes. 

15.1.1 The slides class 

A slide is meant to be projected before an audience and is therefore 

something quite different from a sheet of paper to be read held in the 

hand. One could of course make up the text for slides with the normal 

L AT E X commands, but there would be problems getting the font size right, 

arranging overlays, making sure that the text does not change pages 

unexpectedly, and the question of whether book-style fonts are suitable 

for projection. The L AT E X class slides attempts to solve these. 

A different set of fonts is used by this class, ones in which the lower 

case letters are relatively larger than in the normal fonts, and with a much 

bigger base size. This has the effect of limiting the amount of text that fits 

on one page/slide, but this is in fact very good practice for presentation 

material. The text really should be restricted to keywords and abbreviated 

sentences. A full page of normal text projected on to a screen will not be 

read by the audience. 

One is not obliged to use the built-in fonts; one may still select other 

(sans serif) font packages, such as helvet for Helvetica (Section 10.1.2) 

as one pleases. They will be used in larger sizes than usual. 

Slides should also make use of color. The original SLIT E X had a complicated 

method for producing color overlays (in black and white, to be 

copied in color) which now is totally eliminated from the slides class. 

Instead, one simply makes use of the color package of Section 6.2. 

Most of the formatting commands of regular L AT E X may be used with 

slides, except for the page breaking and sectioning commands. The 

special features that are unique to this class are described in the next 

sections. 

15.1.2 The slide environments 

The source file for producing slides is structured much the same as that 

for a normal document, except that it is the slides class that is invoked

as 

\documentclass{slides} 

preamble text 


slide text 


15.1. Slide production with SLIT E X 325 

If colors are wanted the color package of Section 6.2 must be added, for 

example as 

\usepackage[pdftex]{color} 

(You will have to specify your own driver as the option here, or rely on the 

local configuration to give it automatically, as is usually the case.) 

The preamble may contain global specifications, for changing the paper 

size or selecting a page style, as usual. It may not contain any printable 

material. 

Any text that appears after \begin{document} but before the special 

slide environments described below is output to an unnumbered leading 

page that comes before any slides. This may serve as a cover page for the 

slides. 

There are three environments available for organizing different parts 

of a slide: the main slide itself, possible overlays, and additional notes to 

the slide. 

Slides 

A slide or viewgraph is created by means of the environment 

\begin{slide} text and commands \end{slide} 

The contents of a slide environment may be any text that one pleases. 

Note that the slides class makes use of its own set of character fonts. 

The standard font in the normal size is roughly equivalent to the L AT E X sans 

serif font \sffamily in the size \LARGE. All the regular font commands 

and declarations of Section 4.1 are available, along with the other display 

and list environments of Chapter 4. However, there can be no page 

breaks within the environment, for the entire text is expected to fit on 

one page (slide or viewgraph). The successive slide environments will be 

numbered consecutively. If the page should overflow, a warning message 

is printed. 

Any color commands from the color package may be employed, if 

that package has been loaded. 

Overlays 

An overlay is an addition to a slide that can be laid on top of it to fill 

in certain gaps. The idea is to create suspense during a presentation, by


filling in some key words a few minutes later, or to be able to replace 

some text by alternatives. 

The slides class generates overlays with the overlay environment, 

which functions exactly the same as the slide environment, except that 

the numbering is done as a sub-number of the last slide. Thus the overlays 

following slide 6 are numbered 6-a, 6-b, and so on. 

\begin{overlay} text and commands \end{overlay} 

The slide environment for a viewgraph that is to have overlays must 

come before the overlay environments that go with it. Both the slide 

and overlay should contain identical texts, except that certain parts are 

printed in ‘invisible’ ink by means of the two declarations 

\invisible and \visible 

These two commands function just like font declarations. They may be 

set within curly braces {} to limit their scope, or used to apply to the 

whole environment. Normally, in the slide, a few words might be made 

invisible, while in the corresponding overlay, \invisible is declared 

at the start and only those words that were blanked out in the slide are 

made visible. This may be seen in the sample on page 329. 

Note: the \invisible command does not work with fonts other than 

the default ones, nor with pdfT E X. However, for pdfT E X, there are better 

methods to achieve the same results (Section 15.4.1). 

One application for overlays is to present alternative text. One might, 

for example, show a table of cost estimates, with the figures on an overlay. 

By exchanging this overlay for another, new numbers that can be achieved 

by certain procedural changes may be fitted into the same table. 

Notes 

During a presentation, it is often necessary to refer to a list of keywords 

or other notes between the actual projected slides. The slides class has 

the note environment for producing such reminders for the speaker. 

\begin{note} text \end{note} 

Like the overlays, notes are sub-numbered with respect to the previous 

slide, but with numbers instead of letters. Following slide 4, notes are 

numbered 4-1, 4-2, and so on. 

15.1.3 Further features 

Page styles 

The L AT E X command \pagestyle{style} may also be used with the slides 

class. The following styles are available:


plain All slides, overlays, and notes have their number at the lower 

right corner. 

headings 

The same as plain except that if the clock option is selected 

(see below), a time marker is printed in the lower left corner of 

the notes. This is the default page style. 

empty The sheets are printed with no page numbers. 

In normal L AT E X, the command \pagestyle is issued in the preamble 

and is valid for the whole document. Individual pages may be given a 

different style by means of the \thispagestyle command which remains 

in effect for only one page. This command should not be employed within 

one of the environments, but rather a new \pagestyle command may be 

given outside of the slide, overlay, and note environments. 

Timing notes 

The option clock can be selected with \documentclass; it activates two 

additional commands: 

\settime{secs} 

\addtime{secs} 

and prints the time in minutes at the bottom of each note. In this way, the 

notes contain a reminder of how far advanced the presentation should be 

at that point. The value of the time marker is set and incremented with 

the above two commands, which take a number in seconds as argument. 

The initial value of the timer is 0. 

Selective slide processing 

During the construction of a slide show, one is likely to spend considerable 

time working on individual slides, ignoring the others for the time being. 

Complicated slides need to be corrected, processed, and viewed. One can 

speed this up by selecting only a few slides for processing, by placing the 


\onlyslides{slide list} 

in the preamble. The slide list stands for a list of slide numbers in 

ascending order, for example, 2,5,9-12,15, specifying the slides, or 

range of slides, that are to be processed. 

Non-existent slide numbers may also appear in the slide list. If the 

slide file contains, say, 20 slides, the command \onlyslides{1,18-999} 

will allow only slides 1 and 18–20 to be processed. Any overlays that 

belong to these slides will also be output. 

Finally, the command


\onlynotes{note list} 

in the preamble limits the notes that are output to those listed in note list. 

Supposing that slide 5 has three note environments associated with it, 

\onlynotes{5} arranges that only the note pages 5-1, 5-2, and 5-3 are 

printed. 

The \onlyslides and \onlynotes commands function similarly to 

the command \includeonly described in Section 9.1.2. An interactive 

means of selecting slides and notes to process may be set up exactly as 

illustrated in Section 9.1.3 with the \typein command: 

\typein[\slides]{Which slides to do?} 

\onlyslides{\slides} 

During processing, the message ‘Which slides to do?’ is typed to the 

monitor, and the user replies with the desired slide numbers. A similar 

scheme may be set up for \onlynotes. 

A sample slide file 

An example of input text with the slides class, with a leading page, 

overlay, and note, and using the clock option, is given here. The fourpage 

result is shown on the opposite page. 

\documentclass[a4paper,clock]{slides} 


\begin{center}\Large\bfseries 

Sample Viewgraphs 

\end{center} 

\begin{slide} 

\begin{center}\large Advantages of \texttt{slides}\end{center} 


\item Uses special fonts 

\item Forces key words instead of long text 

\invisible 

\item Supports color layers 

\visible 

\end{itemize} 

Color commands may be used {\invisible as color layers} 

\end{slide} 

\begin{overlay} 

\invisible 

\begin{center}\large Advantages of \texttt{slides}\end{center}


The Leading Page The Slide Page 

Sample 

Viewgraphs 

Advantages of 

slides 

• Uses special fonts 

• Forces key words instead 

of long text 

 

Color commands may be 

used 

The Overlay Page The Note Page 

 

 

 

 

 

• Supports color layers 

 

as color layers 

Add the overlay only for 

L AT E X 2.09 

1-a 5 min 

1-1 

1



\item Uses special fonts 

\item Forces key words instead of long text 

\visible 

\item Supports color layers 

\invisible 

\end{itemize} 

Color commands may be used {\visible as color layers} 

\end{overlay} 

\addtime{300} 

\begin{note} 

Add the overlay only for \LaTeX˜2.09 

\end{note} 


15.2 Slide production with seminar 

The slides class of the previous section some basic support for the 

production of presentation material. To improve on this, Timothy van 

Zandt has written the seminar class with many additional features. This 

was originally meant as a ‘main style’ for L AT E X 2.09, but has been upgraded 

to a L AT E X 2ε class file by Sebastian Rahtz with fixes by David Carlisle and 

Denis Girou. They only changed the interfacing so that it would work as 

a class file, leaving the functionality untouched. 

Unfortunately, the supplied manual for seminar (sem-user.tex) has 

not been updated, and is still couched in the language of L AT E X 2.09; in 

particular it confuses packages and ‘style options’. We will try to correct 

this here. 

If seminar is being used for PDF output, as for an electronic presentation, 

it is vital to include Sebastian Rahtz’ hyperref package (Section 

10.2.4), which detects the presence of seminar and adjusts the PDF 

page sizes accordingly. This should be done even if no other hyperref 

features are utilized. There are some other refinements to make seminar 

function smoothly with pdfT E X, explained in Section 15.2.7. 

15.2.1 Overview 

The seminar class is loaded as usual with 

\documentclass[options]{seminar} 

Here we summarize its main features which differ from SLIT E X.

15.2. Slide production with seminar 331 

• The slide material is put into the slide environment, for landscape 

slides, and in slide* for portrait slides. 

• Regular Roman fonts are used, but this can be changed. The options 

semhelv and semlcmss switch to sans serif family with the 

PostScript Helvetica and the standard SLIT E X fonts, respectively. 

• Regular font sizes are used, with default 10 pt as usual. However, 

the slide text is magnified on output to make it more suitable for 

projection; the magnification factor can be set, but the default is a 

factor of 2, so the effective default font size is 20 pt. 

• A slide is started automatically when the current one is full, or may 

be forced with the \newslide command; the entire slide show could 

therefore be contained within a single slide environment. 

• There is a choice of framing style for the slide text; the framing may 

be turned off with \slideframe{none} in the preamble. 

• Everything outside of the slide or slide* environments counts as 

note text. The notes are printed on separate pages, numbered n.1, 

n.2, . . . , for those following slide number n. 

• The options slidesonly, notes, and notesonly control whether 

only the slides, both slides and notes, or only the notes are printed. 

By default, both are printed. 

• With the article option (not to be confused with the class of the 

same name), the slide show is printed in miniature form, 2 per page, 

appropriate for reviewing or as a handout. There is a notesonly* 

option to be used with article to indicate the missing slides. 

We now explain these features in more detail. 

15.2.2 Magnification and lengths 

Rather than using enlarged fonts as does SLIT E X, seminar works with 

‘normal’ sizes, which are then magnified at output. One can set the 

magnification step with \slidesmag{num} in the preamble, where num 

is an integer between −5 and 9. Step 0 is one-to-one, and each step 

indicates an additional factor of 1.2. The default is step 4 (factor 2). 

By default, the basic font size is 10 pt, but this may be altered with 

the usual options 11pt and 12pt. All these are magnified on output. The 

font size can be changed at any point in the document with \ptsize{pt}, 

where pt can be one of 8, 9, 10, 11, 12, 14, or 17. If this is given within a 

slide environment, it applies only within that environment.


The article option produces a handout, with 2 slides per page, using a 

separate magnification which is set with \articlemag{num}; the default 

step is 0, no magnification. 

The output magnification applies to all sizes and lengths. This means, 

any lines drawn with \rule or spaces inserted with \hspace or \vspace 

are also magnified if they specify absolute units like pt or in. The size of 

the slide itself remains unchanged by the magnification, since the purpose 

is only to increase the size of the contents relative to the slide. To set 

lengths to values that remain fixed relative to the slide dimensions, use 

\setslidelength and \addtoslidelength in place of \setlength and 

\addtolength. For example, 

\newlength{\fixed} 

\setslidelength{\fixed}{3in} 

\addtoslidelength{\fixed}{1cm} 

\rule{1pt}{\fixed} 

draws a line of height 1 pt × magnification and of fixed length 3 in + 1 cm. 

For the article output, there are also the corresponding commands 

\setartlength and \addtoartlength 

The following length commands are slide lengths (do not magnify) and 

may be used wherever a length is required: 

\semcm 1 centimeter 

\semin 1 inch 

\textwidth width of the text area 

\textheight height of the text area 

\linewidth current width of text 

These are particularly useful with the \includegraphics command and 

its options width= and height=. 

15.2.3 Landscape and portrait slides 

In contrast to SLIT E X, the seminar class allows landscape and portrait 

slides to be mixed. However, there are several aspects that need to be 

considered. 

The slide and slide* environments determine whether the contents 

should be in landscape or portrait slides, but the actual placement will 

only be correct when it agrees with the orientation of the entire document, 

as specified by the options landscape and portrait. That means, when 

the landscape option is chosen, the landscape slides will be properly formatted, 

but the portrait ones not. The reverse applies with the portrait 

option. What one must do is to print the landscape and portrait slides separately, 

each with the appropriate option, and with the \landscapeonly 

or \portraitonly command in the preamble.


When using the dvips driver, one can print the landscape and portrait 

slides together by including the option semrot, which ensures that the 

portrait slides are rotated into the landscape orientation. The entire 

document (slide show) can be printed in one go. The portrait sheets 

are then manually rotated to the correct orientation. This would not be 

acceptable for a PDF electronic presentation, where one wants the two 

orientations to be mixed in one file, with the text always upright. Another 

solution for pdfT E X is given is Section 15.2.7. 

When printing both orientations together with the semrot option, the 

head and footlines will not normally be rotated with the text for portrait 

slides. This can be corrected by issuing \rotateheaderstrue. In this 

way, the output is the same as when one prints landscape and portrait 

slides separately. 

Selecting landscape orientation in itself does not force the output 

printer to go into landscape mode; it only sets the page dimensions so 

that the text is written along the longer side. The printer is informed 

about the landscape intentions either by a switch when the file is sent to 

it, or by an option to the DVI driver. Again, dvips can get around this, as 

explained on page 232. Using seminar and dvips, one should define 

\renewcommand{\printlandscape}{\special{landscape}} 

to allow the class to instruct the printer to go landscape when necessary. 

This also avoids the warning message on processing that the output must 

be printed in landscape mode. 

15.2.4 Slide formatting 

Here we describe the various ways to control the size, layout, and other 

formatting aspects of the slides. 

Slide size 

By default, the paper size is American letter size, 11 × 8.5 in, and the 

slide dimensions (slide writing area) are 8.5 × 6.3 in. By specifying the 

a4 option the paper size becomes 29.7 × 21 cm and the slide dimensions 

22.2 × 15.2 cm. (The a4paper option exists, but may cause an error.) 

The slide size may be altered temporarily as optional arguments to 

the slide and slide* environments, as for example, 

\begin{slide}[10cm,7cm] 

\begin{slide*}[7cm,10cm] 

Both of the above produce a slide 10 cm wide and 7 cm high.


Margins 

Normally the material within a slide is centered vertically. The declaration 


\centerslidefalse and \centerslidetrue 

turns this centering off and on, respectively. Without centering, the text 

is pushed to the top of the slide. 

The text is left justified by default, with a ragged right margin. This 

can be changed with 

\raggedslides[len] 

where len is the maximum space between the end of the line and the right 

margin. Omitting len produces a ragged right margin, while a value of 0 pt 

yields right justification; giving 2 em, say, creates a semi-ragged margin. 

The margins between the slide area and the edge of the page are set 

with the commands 

\slideleftmargin \sliderightmargin 

\slidetopmargin \slidebottommargin 

which are all set initially to 0.6 in. Since these are commands, not lengths, 

they must be changed with \renewcommand. 

Framing 

The slide area can be framed, the default being a rectangular box like that 

produced by \fbox (which is exactly what it is). The frame style can be 

chosen with 

\slideframe[cmds]{style} 

The possible values for style are plain (the default) or none. The optional 

cmds are settings to adjust the frame parameters, such as \fboxrule 

and \fboxsep. By including the fancybox package, one also has shadow, 

double, oval, and Oval as allowed values for style, with their corresponding 

parameters (Section 4.7.9). 

All frames make use of the lengths 

\slideframewidth and \slideframesep 

for the frame thickness and separation from text. These may only be 

altered (with \setlength) outside the \slideframe command; in the 

optional cmds, one must alter \fboxrule and \fboxsep directly. 

It is possible to define one’s own frame style with 

\newslideframe{new style}[cmds]{\box cmd{#1}}


where new style is the name of the new style (for use with \slideframe), 

cmds are box parameter settings, and \box cmd is a box command, like 

\fbox or \shadowbox. 

There is a semcolor option for seminar, but this is not based on 

the L AT E X color package, but depends on PostScript and the dvips driver. 

With the color package, one can make use of the \color and \textcolor 

commands for setting text in color, as well as \colorbox and \fcolorbox 

for colored boxes and \pagecolor to color the whole page (page 167). 

Two examples for defining color frame styles are 

\newslideframe{gplain}{\fcolorbox{green}{white}{#1}} 

\newslideframe{rshadow}[\color{red}]% 

{\shadowbox{\color{black}#1\color{red}}} 

The first makes use of the color frame box command \fcolorbox, while 

the second shows a trick to combine color commands with \shadowbox, 

which normally prints a black shadow box. The two \color{red} commands 

ensure all sides of the box are in red, while the \color{black}#1 

puts the box’s contents in black. 

Page styles 

Page styles determine how the running head and footlines appear on each 

page, just as in normal text. The regular L AT E X page style of Section 3.2 

(plain, empty, myheadings, and headings) are all available, and the 

default is plain, with the page/slide number centered in the footline. A 

different page style may be selected with \pagestyle{style} at any time, 

or with \thispagestyle{style} for just the current page. 

There is also an additional page style align that puts + marks at the 

four corners. 

You can define your own page styles with 

\newpagestyle{style}{headline}{footline} 

\renewpagestyle{style}{headline}{footline} 

which is preferable to using the styles in the fancyhdr package of Section 

3.2.2 since that cannot handle the magnified lengths correctly. For 

example, to define a page style with a centered text in the head, and a 

footline with a text at the left and the slide number at the right, give 

\newpagestyle{mystyle}% 

{\hfill Report\hfill}% 

{March 20, 2004\hfill\thepage} 

Note here that \thepage prints the number for the current page, whereas 

\theslide and \thenote refer to the latest slide and note numbers 

respectively; \thepage is always set to whichever of them is being printed.


Font commands can be included in the new definitions, but one should 

be aware that the commands \slideheadfont and \slidefootfont are 

always inserted into the head and footlines. These are defined to be 

\scriptsize, but may be redefined by the user. 

It is possible to select different page styles for the slides and notes. 

The \pagestyle command sets the style for both (as well as for the 

article output) while \slidepagestyle applies only to slides. 

Page breaking 

Unlike SLIT E X where each slide environment generates a single slide (and 

warnings if it is too big), the seminar package allows the text to break 

naturally into multiple slides just as normal text does into pages. A new 

slide may be forced with \newslide. 

You can add some extra tolerance for the page breaking by specifying 

\extraslideheight{len} 

where len is the additional height allowed before a break in automatically 

inserted. By setting this to be large, automatic page breaks never occur, 

and the user has full control, either with \newslide or by ending the 

environment. Setting it to 0 pt causes pages to be broken as for regular 

text. The default is 10 pt. 

15.2.5 Selective output 

One can select to output just the slides, or just the notes, or both, by 

adding one of the options slidesonly, notesonly, or notes to the 

\documentclass command. There is also a notesonly* option to print 

the notes alone in article output but with indicators for the slides. 

A problem arises if slidesonly has been selected. Any global specifications 

issued after \begin{document} and outside of a slide environment 

will be ignored. One might want to redefine parameters in this way; 

including them inside the slide environment makes the changes local to 

that one environment. To avoid this, such global commands should be 

placed inside a allversions* environment. (If you actually want to include 

text and not just non-printing specifications, use the allversions 

environment instead.) 

Just as for SLIT E X, one can output selected slides or notes. This is useful 

during development of the slide show, when one may spend much time 

designing a complicated slide and is constantly processing it to see the 

results; this can become tediously slow if the other, completed, slides are 

also complicated or include large graphics files. As for SLIT E X (page 327), 

the \onlyslides{list} command in the preamble ensures that only those 

slides whose numbers are included in list are processed. However, there 

is also a \notslides{list} command to exclude those slides in list.


The list a set of numbers in any order, separated by commas, or a 

range with a hyphen. It is even possible to include \ref commands 

for slides containing the corresponding \label, something that is very 

useful during development when the slide numbers are yet finalized. For 

example, 

\onlyslides{4,2,\ref{sum},10-999} 

outputs slides 2, 4, the slide containing \label{sum}, and all from 10 to 

999. Non-existent numbers are ignored, so 10-999 really means ‘10 to 

the end’, unless there really are so many slides in this show. 

15.2.6 Additional features 

Counters 

The seminar class provides all the standard L AT E X counters (Section 8.1) 

as well as slide and note counters, the values of which are printed with 

\theslide and \thenote. The page number is printed with \thepage, 

which is set to whichever of the others is current, or to the true page 

number for article output. 

The footnote counter is the only one that is reset to 1 for each new 

slide. If you want other counters to be reset too, say equation, specify 

this with 

\slidereset{list} or \addtoslidereset{list} 

where list is a comma-separated list of counters. The first command 

overwrites the existing list of counters, the second adds to it. 

The note environment 

As an alternative to note pages being generated by all text outside of the 

slide environment, one can issue \noxcomment in the preamble, or select 

option noxcomment in \documentclass, and then place all the notes in 

\begin{note} . . . \end{note} environments. 

Overlays 

Overlays for slides can be produced by including the options semlayer 

and semcolor. These make use of PostScript coding and therefore only 

work with the dvips driver or equivalent. 

The \overlay{n} indicates that the following text is to appear on 

overlay n; it remains in effect until the current scope is ended, at the end 

of the environment or {..} group, or when another \overlay command 

is issued.


Overlays are not so useful for a PDF electronic presentation. Instead, 

one creates the same effect by building the page successively. A method 

of doing this is explained in Section 15.4.1. 

Adding a background image 

Slide shows are made more effective with the use of color, and with a 

background pattern or picture. A background color can be added with 

the \pagecolor command from the color package. 

However, even more effective is a background image. If this is a 

graphics file named mybg.png, one can insert it as a background to all 

slides with the following trick. This does require the fancybox package 

for the \fancyput command. 

\newlength\bgwidth 

\newlength\bgheight 

\ifportrait 

\setslidelength{\bgwidth}{\paperwidth} 

\setslidelength{\bgheight}{\paperheight} 

\else 

\setslidelength{\bgwidth}{\paperheight} 

\setslidelength{\bgheight}{\paperwidth} 

\fi 

\ifarticle\else 

\fancyput(-1\semin,1\semin){\raisebox{-\bgheight} 

{\includegraphics[totalheight=\bgheight, 

width=\bgwidth]{mybg.png}}} 

\fi 

What this does is first to define and set two new lengths, \bgwidth 

and \bgheight, to be the non-magnified size of the full page, for the 

current orientation. The \fancyput command places its contents on each 

page, relative to a point 1 in from the upper left corner; with the above 

coordinates, it is shifted exactly to the upper left corner. The graphics 

is inserted here, shifted down by the page height, and expanded/shrunk 

to fit the page exactly. One must design the graphics correctly so that 

any distortion is minimal. The \ifarticle construction prevents the 

background image from being included with article output. 

This should work with any type of output, PostScript or PDF or other. 

Of course, the type of graphics file must conform to those allowed by the 

DVI driver. For PostScript, it must be an.eps file, not.png. 

See also page 345 for an alternative means of adding background 

images.

Local configuration 


The seminar class reads in the file seminar.con if it exists. This gives 

the user the chance to write all the specifications and new definitions that 

he or she always employs to this file, to have them automatically included 

in every seminar source file. 

15.2.7 Using pdfT E X with seminar 

Although the seminar class originated in the pre-L AT E X 2ε days, and many 

of its extra features are coded for PostScript and the dvips driver without 

the modern graphics and color packages, there is no problem using it 

with today’s L AT E X and those packages. This is particularly true for PDF 

output. 

However, with pdfT E X itself, there are some extra length parameters 

\pdfpagewidth and \pdfpageheight which need to be properly set to 

get the output right. This is most simply guaranteed by including the 

hyperref package (Section 10.2.4) even if no other of its features is 

needed. 

Another difficulty arises if landscape and portrait slides are mixed. 

If the goal is only to print to paper or transparencies, then the two 

orientations may be processed and output separately, as explained above. 

However, for an electronic presentation, this would be most inconvenient. 

Certainly the ability to mix the two orientations is highly desirable. 

The following code makes this possible. It revises the slide* environment 

to interchange the PDF page dimensions, so that a true portrait 

page is produced. (This is not possible with printed output where a single 

print job must have a fixed orientation.) 

\@ifundefined{pdfoutput}{\endinput}{% 

\ifcase\pdfoutput \endinput \fi} 

\newcommand*{\pdf@revpage}{% 

\@tempdima=\pdfpagewidth 

\pdfpagewidth=\pdfpageheight 

\pdfpageheight=\@tempdima 

\@tempdima=\paperwidth 

\paperwidth=\paperheight 

\paperheight=\@tempdima 

} 

\ifarticle \pdf@revpage \fi 

\renewcommand{\printlandscape}{} 

\expandafter\let\expandafter\slide@str 

\csname slide*\endcsname 

\@namedef{slide*}{\pdf@revpage\slide@str}


Store this in a file named sempdftx.sty, which can then be loaded with 

\usepackage. It is programmed to do nothing if pdfT E X is not being 

used. Alternatively, add a line \RequirePackage{sempdftx} to the local 

configuration file seminar.con to load it on every seminar run. 

Note: This file can be copied from the enclosed CD in the directory 

\books\Kopka_and_Daly\, where a version of seminar.con is also available. 

15.3 Electronic documents for screen viewing 

Package: 

pdfscreen 

Strictly speaking, this section is not about presentation material that is 

to be projected before an audience, but rather about documents that are 

deliberately intended to be read from a computer monitor. These have 

other requirements from printable documents: the page shape should be 

more like a monitor (that is, landscape instead of portrait) and navigation 

aids should be included especially if the viewer’s regular tool bars have 

been switched off. And of course, internal links should be present, as 

well as links to external web sites and documents. 

The pdfscreen package by C. V. Radhakrishnan is an effort to fulfill 

these requirements. But it is more flexible than that, for it also allows 

the entire document to be output in traditional print format. And it even 

contains a slide environment for making projection material, so it does 

indeed belong to this chapter. 

The supplied manual can be found in texmf\doc\latex\pdfscreen, 

in print and screen versions. The source file manual.tex can serve as a 

good example of how to use the package. Figure 15.2 on the opposite page 

demonstrates what a title page can look like, complete with navigation 

panel which appears on every page. 

Parameters for pdfscreen 

This package should be loaded with the article class and with the 

hyperref package. In fact, it should be considered to be an extension to 

hyperref. And it should be loaded last, right after hyperref. 

The options that may be invoked with the \usepackage command are 

screen to generate the output in the screen version, for online reading; 

print to generate the print version of the output; 

panelleft 

to place the navigation panel at the left; 

panelright 

to place the navigation panel at the right;

15.3. Electronic documents for screen viewing 341 

Figure 15.2: Title page of a pdfscreen document 

nopanel 

to suppress the navigation panel; 

paneltoc 

to include a table of contents in the navigation panel; 

nocfg to ignore any local configuration file; 

sectionbreak 

to insert a new page at the start of sections. 

In addition one can specify various color schemes for the panel and 

its buttons. Possible color options are bluelace, blue, gray, orange, 

palegreen, and chocolate. Also all the babel language options (Chapter 

11) are recognized, for labeling the navigation buttons, although only 

15 are actually supported; the others default to English. 

There are additional parameters that must be set by declarations: 

\emblema{image} 

to set the name of the graphics file used for the logo in the 

navigation panel;


\urlid{url} 

to set the web address (URL) of the home page button in the 

navigation panel; 

\screensize{height}{width} 

to set the dimensions of the screen; since there are no defaults, 

this must be given; 

\margins{left}{right}{top}{bottom} 

to set the margins for the document; again there are no defaults, 

so this must be given. 

The navigation panel 

The navigation panel appears on each page and contains the buttons for 

changing pages and linking to the author’s home page. A default panel is 

provided, but the user may redesign it as he or she pleases, by redefining 

the \panel command. For example, 

\renewcommand{\panel}{\colorbox{panelbackground} 

{\begin{minipage}[t][\paperheight][b]{\panelwidth} 

. . . . . 

\Acrobatmenu{FirstPage} 

{\addButton{.2in}{$\blacktriangleleft$}} 

. . . . . 

\end{minipage}}} 

Here the \Acrobatmenu command is from the hyperref package (see 

page 247) while \panelwidth and \addButton are from pdfscreen. The 

\panelwidth length is by default 15% of the screen width, and can be 

redefined. The \addButton{wth}{text} creates a button of width wth 

containing text. In the above example, clicking the button executes the 

viewer command to go to the first page. 

There is also a command 

\imageButton{wth}{ht}{image} 

to insert the graphics file image as a button of width wth and height ht. 

A row of small navigation buttons can be turned on at the bottom of 

the page with \bottombuttons and turned off with \nobottombuttons. 

Background image 

A graphics file can be inserted as background for the whole page with 

\overlay{image}, or a uniform background color can be selected with 

\backgroundcolor{color}. 

The navigation panel has as background that color previously defined 

as panelbackground, which the user may alter. Otherwise, the graphics

15.4. Special effects with PDF 343 

file image may be inserted as the panel background with the command 

\paneloverlay{image}. 

Making slides 

There is a slide environment that places its contents centered vertically 

within a box the full size of the text area. 

Local configuration 

Any settings that you want to use for all documents with pdfscreen may 

be written to a file named pdfscreen.cfg, which is read in on loading if 

present. 

Controlling print or screen output 

The entire file can be processed either as a document to be printed in the 

traditional way, or as an electronic document for reading on a monitor. 

This is switched by the options print and screen when loading the 

package. 

Text and specifications that are meant exclusively for one output version 

must be enclosed in print or screen environments. This is especially 

important for the \screensize and \margins commands, which 

are inappropriate in print mode. 

Other packages loaded 

The pdfscreen package makes heavy use of several other packages, which 

it loads automatically. These must then also be installed on the system: 

hyperref 

comment 

truncate 

graphicx 

color 

colortbl 

15.4 Special effects with PDF 

calc 

amssymb 

amsbsy 

shortvrb 

fancybox 

There are many fancy features that one expects today from an electronic 

presentation: sounds and animation, text dancing in from the sides, fade 

overs, and possibly fireworks. A PDF file can include many of these. 

The pdfT E X program cannot (yet) generate these directly, but there is an 

auxiliary program called PPower4, written in Java by Klaus Guntermann 

and Christian Spannagel. This is a post-processor designed to take as 

input a PDF file specially prepared with pdfT E X with some extra packages, 

and to create a new PDF file with the additional tricks.


For example, in order to build up a page containing an itemized list, 

such that each item should pop up one at a time, one adds a \pause 

command after each item, runs pdfT E X, and processes the output with 

PPower4. It splits the page up into several pages, each one containing 

the text up to the next \pause. By stepping through these pages, the 

presenter ‘constructs’ the one slide before the eyes of the audience. In 

fact, PPower4 can do much more, making a very sophisticated electronic 

slide show from pure L AT E X input. 

The PPower4 program and the necessary L AT E X support packages are 

on the T E XLive CD, but the latest versions can be obtained from http:// 

www-sp.iti.informatik.tu-darmstadt.de/software/ppower4/. At 

the time of writing, it is still under development, but is already very 

impressive. 

15.4.1 The PPower4 packages 

The PPower4 post-processor is run with only two parameters: the names of 

the input and output files. There are no additional adjustments, controls, 

or options at processing time. Everything of that sort has already been 

placed in the L AT E X source text itself with the help of supplied packages. 

These instructional commands add comments to the resulting PDF file, 

which may be viewed and printed as normal. It is the post-processor that 

reads these comments and uses them as instructions for making up the 

new output file. 

Here we describe these packages and the marker commands that they 

enable. The use of the hyperref package is presupposed, since many 

effects can be achieved directly with it. 

Page transitions 

Transitions from one page to another can be done in various ways to catch 

attention. The simplest is the plain replacement: the old page vanishes, 

the new one appears. Others involve effects that look like blinds opening, 

or like a box opening from or towards the center, or dissolving from one 

to the other. These can be set with the hyperref package with the option 

pdfpagetransition (page 245) or set to a different effect anywhere in 

the document by giving 

\hypersetup{pdfpagetransition={pars}} 

where pars are the required parameters for the desired transition. 

Marc van Dongen has prepared a file pagetrans.tex, available with 

the PPower4 collection, that defines simpler commands to execute the 

above. (If this were renamed pagetrans.sty, it would be a package and 

could be loaded with \usepackage but as is, it must be included with the 

\input command.) These commands are:

Package: 

background 

\Replace pages are simply replaced, the default; 


\Dissolve one page ‘dissolves’ into the other in a mosaic fashion; 

\VBlinds several horizontal lines sweep vertically across the screen, like 

Venetian blinds; 

\HBlinds same as \VBlinds, but vertical lines move horizontally; 

\HOSplit two vertical lines sweep horizontally out from the center; 

\HISplit as above, but the lines move into the center; 

\VOSplit two horizontal lines sweep vertically out from the center; 

\VISplit as above, but the lines move into the center; 

\OBox a box opens out from the center; 

\IBox a box sweeps into the center; 

\Wipe{angle} a single line sweeps across the screen in the direction given 

by angle, which can be 0 (left to right), 90 (bottom to top), 180 (right 

to left), 270 (top to bottom); 

\pageTransitionGlitter{angle} like \Dissolve but in a band that 

moves across the screen in the direction angle, which can be 0 (left 

to right), 270 (bottom to top), 315 (upper left to lower right). 

We stress again, these declarations are simplifications of the hyperref 

commands, and function without PPower4 post-processing. Once given, 

they remain in effect for all pages until a new transition command is 

declared. 

Adding a background 

We have already seen in Section 15.2.6 how a background image may be 

added with the help of the fancybox package, and how a solid background 

color can be set with the color package. 

These and other effects can also be achieved with the background 

package supplied with PPower4. This package provides the commands 

\hpagecolor[color1]{color2} and 

\vpagecolor[color1]{color2} 

that cause the background color to blend from color1 to color2 horizontally 

or vertically. If the optional color1 is omitted, it is the intensity 

of color2 that changes across the page. Both colors must be defined by 

the color package command \definecolor, or be one of the predefined 

colors. This effect requires post-processing with PPower4!


The package can also add background images, without PPower4 postprocessing, 

if it is loaded with the option bgadd. This then includes the 


\bgadd{image} and \bgaddcenter{image} 

to place image at the upper left corner or centered on the page. The 

image is an \includegraphics command with appropriate scaling to fit 

the page. These commands may be issued more than once to add further 

elements to the background. They may all be removed with \bgclear. 

Note: The addition of background elements to the pages requires that 

the package eso-pic also be installed. 

Building a page successively 

Package: The electronic equivalent to overlays is the successive building of a page 

pause of text. The PPower4 pause package introduces the \pause command 

that places markers in the PDF output which PPower4 then uses to break 

the page at that point, and to start a new page containing the previous 

contents plus whatever comes up to the next \pause or end of page. This 

works for both pdfTEX and dvipdfm. 

For example, we could build a numbered list with 

We proceed as follows: 


\item We write the list source text.\pause 

\item We add the pause markers.\pause 

\item We run pdf\TeX.\pause 

\item We run PPower4.\pause 


And there we have it! 

The first PDF file generated will have a single page with a small colored 

block where each \pause command appears. (This may be suppressed if 

the package is loaded with the nomarkers option.) The resulting PDF file 

after post-processing will have 5 pages in place of this one, each one with 

an additional item. As the presenter steps through the pages, each new 

line appears to be added to a single page, until the full page is complete. 

Building with transitions 

Rather than having the new line simply popping up out of nowhere, one 

can select the type of page transition at any step by giving a modified 

\pause with the name of one of the transition declarations from page 344. 

For example \pauseHBlinds is the same as \pause plus \HBlinds except 

that the transitions between main pages are not affected. Note that


\pauseReplace is the default, causing the new text simply to appear. 

These remain in effect for the following \pause breaks until countermanded. 

There is a \pauseGlitter{angle} command as the equivalent to 

\pageTransitionGlitter{angle}, but all others just attach the declaration 

name to \pause. 

Building with levels 

Even more sophistication can be added by assigning level numbers to the 

chunks of text between the \pause markers. In this way the page can be 

built up in any order, parts can be made to disappear or to be replaced, 

and the footline text at the bottom of the page can be visible from the 

beginning. 

The level number corresponds to the sequence number of the views 

of the built-up page. Thus a chunk of text (that which comes between 

\pause commands) with level n will first appear in view n. By default, the 

level number starts at 1 and increases by 1 for each \pause. To ensure 

that the page footline is at level 1, we can write the last line of the above 

example as 

And there we have it!\pause\pauselevel{=1} 

The \pauselevel command sets the level number for the chunk in 

which it is located. There are several possibilities for its argument. 

{=n} sets the level number absolutely to n; 

{=+n} increases it by n; bear in mind that the preceding \pause has 

already incremented it by 1; 

{=-n} decreases it by n; 

{=n -d} sets level number to n and makes the change in level 

number with \pause to be −d; this would be done with say 

{=10 -1} to make the chunks numbered backwards from 10, 

causing the page to be built from the bottom up; 

{:m} sets the maximum level to m; at higher levels, the text vanishes; 

with {=3 :4} the chunk is visible only at levels 3 and 4; 

the :m may also be set relatively as {:+m} or {:-m}. 

It is also possible to give multiple level specifications, as 

\pause\pauselevel{=2 :2, =5 :6}Text\pause 

to cause ‘Text’ to appear at level 2, to vanish for 3 and 4, and to reappear 

only for 5 and 6. 

The level number may not be less than 1. Zero and negative numbers 

are treated as 1.


With some tricky playing around with level numbers and with the T E X 

overlap commands \rlap and \llap, it is possible to get text or images 

to be replaced. The two must be made to overlap, but at different level 

numbers. For example: 

We now \rlap{alter}change this word 

We now 

\pause\pauselevel{=1 :1}\rlap{alter}\pause 

change\pause\pauselevel{=1} 

this word 

The first line shows the text without the \pause markers, in which ‘alter’ 

overlaps to the right over ‘change’. The next lines show the same text with 

levels inserted: ‘alter’ belongs to level 1, and disappears after 1; ‘change’ 

belongs to 2 (automatic increment), and the remaining text is set back to 

level 1. 

Highlighting instead of building 

An alternative to building the page by pieces is to highlight the section of 

text being discussed. In this mode, the entire page is visible, but printed 

in a dull color, like gray, while the chunk of text at the current level is in 

a bright color. When switching to the next view (level), that text becomes 

dull, and the next chunk brightens. 

To get this to work, one must inform PPower4 what the normal and 

highlight colors are to be, and one must further indicate what text should 

actually change. Clearly certain parts of the page are to remain as they 

always are, at least the head and footlines, and possible titles. This is 

done with 

\pausecolors{textclr}{dullclr}{highclr} 

where textclr is a dummy color that is not otherwise being used. Only text 

indicated in this color will participate in the highlighting. For example 

\pausecolors{cyan}{gray}{red} ... 

This is the \textcolor{cyan}{text to be highlighted} 

by the ... 

The ‘text to be highlighted’ will normally appear in gray, but will turn 

red for that view corresponding to its level number. And only for that 

view! At the next view, it returns to gray. Note that the color cyan never 

appears; it is only a marker for PPower4. (Well, it does appear in the first 

PDF output, before post-processing.) 

The highlighting mode is activated with \pausehighlight which 

makes all text at all levels visible, but with this color switching feature. 

The build mode can be reactivated with \pausebuild. In build mode, the


color switching still takes place, but the text does not appear until the 

right view is reached. It will be highlighted only for this first view. 

One can add the word highlight to the argument of \pauselevel to 

indicate a chunk that should always be visible even in build mode, and 

which participates in the color switching. In highlight mode, this has no 

effect. 

Linking to first view 

Package: Making an internal link with the hyperref commands \hypertarget and 

pp4link \hyperlink (page 247) causes the target to be on the completed built up 

page, on the last view. However, one probably wants to link to the first 

view, before the page is built up. The pp4link package can assist here. It 

defines the commands 

\toplink{name}{text} and \toptarget{name} 

to establish text as a link to the target name. 

This package tries to load hyperref on its own, without any options. If 

you try loading hyperref afterwards, with options, you will get a message 

about conflicting options and the options will be ignored. Therefore, load 

pp4link after hyperref.

16 

Letters 

In addition to the three standard L AT E X document classes, there is a fourth 

one named letter for formatting correspondence. As provided, this 

class is intended for private letters without any frills such as letterheads 

or business reference codes. However, local modifications are possible. 

We first present the standard L AT E X letter class, and then demonstrate 

how a house style can be written using our own institute style as an example. 

As always, it is highly recommended that each of these modifications 

be given its own file name so that letter.cls refers only to the provided 

L AT E X version. 

16.1 The L AT E X letter class 

The letter document class is meant for writing letters. A single input 

file may contain the text for more than one letter and recipient, all from 

the same sender. Address labels may also be printed automatically if 

one wishes. Most of the normal L AT E X commands function as usual within 

the letter class. One exception, however, is the sectioning commands, 

which will lead to the error message ! Undefined control sequence 

when issued. It actually makes little sense to try to divide a letter up into 

chapters, sections, etc. On the other hand, there are a number of special 

commands that may be applied only within this style. 

The input text for a letter begins as for every L AT E X document with 

\documentclass[options]{letter} 

in which all the options listed in Section 3.1 may be given for options 

except for twocolumn and titlepage which hardly make any sense within 

a letter. 

Every letter must contain the name and address of the sender, which 

are set for all the letters in one file by including the commands 

\address{sender address} 

\signature{sender name} or \name{sender name} 

351

352 Chapter 16. Letters 

in the preamble. The sender address normally consists of several lines, 

separated by \\ commands, as in the example 

\address{Max-Planck-Institut f\"ur Aeronomie\\ 

Max-Planck-Str.\ 2\\ 

D--37191 Katlenburg--Lindau\\Germany} 

The entry in the \name command will be used in the return address in the 

letterhead, if such has been programmed. The entry in the \signature 

command will be printed at the end of the letter below the space left 

blank for the writer’s signature. If \signature has not been specified, 

the \name entry is inserted here instead. This allows a more formal 

\name to be used for the return address and a different form, perhaps 

with multiple lines, for the signature, as in 

\name{Prof.\ M.\ Ostmann} 

\signature{Martin Ostmann\\Project Leader} 

When the above commands are issued in the preamble, they remain valid 

for all the letters in the document, except for those letters that contain 

new versions of these commands. Thus one letter might have a different 

\signature from the others. The scope of these entries extends only to 

the end of the environment in which they were called (see Section 8.5.4). 

Two other sender entries are possible in standard L AT E X letter class. 

They are intended to be employed in local modifications for a house style. 

The idea is that if \address is not called, the preprogrammed company 

letterhead which might also contain the sender’s room and/or telephone 

number is generated. Thus the commands 

\location{room number} and \telephone{tel number} 

are provided. In the standard letter class, the entries room number and 

tel number are printed at the bottom of the first page only if \address is 

not issued. 

The preamble may also contain the \pagestyle command with the 

usual entries plain, empty, or headings. The first is the default, putting 

the page number centered at the bottom of all pages after the first. The 

headings page style adds the recipient’s name, the date, and page number 

in a line at the top of all pages after the first. 

After the preamble commands, the actual text begins as in all L AT E X 

files with the command \begin{document}. The text consists of one or 

more letters, each enclosed in a letter environment with the syntax: 

\begin{letter}{recipient} text of letter \end{letter} 

where recipient stands for the name and address of the recipient, divided 

into several lines separated by \\ commands. 

\begin{letter}{Mr Donald J. Burns\\ 

Ontario Institute of Physics\\

16.1. The L AT E X letter class 353 

41 Adelaide St.\\ 

London, Ontario\\Canada N4R 3X5} 

The text of letter normally begins with the command \opening and 

ends with \closing, between which the body of the letter appears, mixed 

with whatever L AT E X commands are desired. The syntax of these two 

commands is 

\opening{dear} 

\closing{regards} 

where dear is the salutation, such as Dear Mr Tibs, and regards stands 

for the terminating text, for example Yours sincerely,. The \opening 

command could also contain other text, for example a subject entry line, 

with the true salutation as part of the following body text. 

L AT E X places the sender’s name and address in the upper right corner 

of the first page with the current date set right justified below it. Then the 

recipient’s name and address appear at the left margin, followed by the 

salutation and the body of the letter. The letter ends with the terminating 

text and the sender’s name or signature left justified from the center of 

the line with sufficient vertical space between them for the handwritten 

signature. 

After the \closing, a number of other commands may appear as part 

of the letter. One is \cc to produce a copy distribution list: 

\cc{name1 \\ name2 \\ ... } 

The text ‘cc:’ (or more correctly, the text defined in \ccname) is printed 

at the left margin, followed by an indented list of names of the copy 

recipients. 

The second additional command is \encl for making a list of enclosures: 

\encl{enclosure1 \\ enclosure2 \\ ... } 

The word ‘encl:’ (actually the text in \enclname) is printed at the left 

margin and then the list of enclosures. 

Finally, the command \ps may be used to add a postscript after the 

signature. The command itself does not generate any text, nor does it 

take an argument. The postscript text is everything located between the 

\ps and \end{letter} commands. 

Normally the letter is dated automatically with the current date. However, 

if it is desired that the letter be back-dated, or that the date be 

otherwise fixed in the text, it may be set with 

\date{date text} 

in which case date text appears where the current date would be placed. 

A letter file may contain any number of letter environments, one per 

letter. As stated already, when \address, \name, and \signature have


A sample letter produced with the standard letter class. 

T E Xproof Ltd 

P. O. Box 123 

9876 Wordtown 

Textland 

Dear Sir; 

Max–Planck–Institut für Aeronomie 

Max–Planck-Str. 2 

D-37191 Katlenburg–Lindau 

Germany 

September 8, 2003 

We are most pleased to be able to answer your request for information about the 

use of L AT E X for general text processing at a scientific institute. 

1. After some initial trepidation on the part of the secretarial staff, which was 

mainly due to the first experience with a non-WYSIWYG text system, the 

system is now fully accepted and appreciated. 

2. Much to the surprise of many secretaries, they find that they are able to set 

the most complicated mathematical formulas in a reasonably short time 

without difficulties. The same applies to the production of detailed tables. 

3. Creating cross-references and keyword indices no longer causes horror, 

even when the author is well known for demanding constant changes. 

4. Finally, the high quality appearance of the output has assisted in winning 

acceptance for L AT E X in our house. 

An additional positive note is the ability to write business letters readily, making 

use of the letter class provided with L AT E X. In our institute, we have designed a 

special version to print our own letterhead, saving the need to have special letter 

paper printed. 

encl: Listing of our mpletter.cls 

Sample output 

cc: H. Kopka 

B. Wand 

Yours truly, 

Patrick W. Daly

16.1. The L AT E X letter class 355 

been specified in the preamble, they remain in force for all the letters in 

the file. It is possible to alter any one or more of these sender entries 

by reissuing the command within one of the letters, before the \opening 

command, but then this change is valid solely for that letter. If both \name 

and \signature have been declared, it is the latter that is printed below 

the signature space. 

The first page contains no page number. Subsequent pages have either 

a centered page number at the bottom (default) or a heading line with 

recipient, date, and page number at the top (page style headings). 

The sample letter on the facing page was generated with the following 

input text. 

\documentclass[a4paper,11pt]{letter} 

\name{Dr P. W. Daly} 

\address{Max--Planck--Institut f\"ur Aeronomie\\ 

Max--Planck--Str.\ 2\\ 

D-37191 Katlenburg--Lindau\\Germany} 

\signature{Patrick W. Daly} 

\date{September 8, 2003} 


\begin{letter}{\TeX proof Ltd\\P.\,O. Box 123\\ 

9876 Wordtown\\Textland} 

\opening{Dear Sir;} 

We are most pleased to be able to answer your request for 

information about the use of \LaTeX{} for general text 

processing at a scientific institute. 


\item 

After some initial trepidation on the part of the secretarial 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

in winning acceptance for \LaTeX{} in our house. 


An additional positive note is the ability to write business 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 

special paper printed. 

\closing{Yours truly,} 

\encl{Listing of our \texttt{mpletter.cls}\\Sample output} 

\cc{H. Kopka\\B. Wand} 

\end{letter} 


By adding the command \makelabels in the preamble, the user can 

print out address stickers after all the letters have been output. The 

entries for the addresses are taken from the recipients’ names and addresses 

in the argument to the letter environment. The standard LAT EX letter class is designed for a page of labels 4 1 

4 × 2 inches, ordered in 

two columns. This could be altered for other formats. An address sticker


without a corresponding letter may be printed by including an empty 

letter environment of the form 

\begin{letter}{recipient}\end{letter} 

16.2 A house letter style 

The sample letter above demonstrates the possibilities of the standard 

L AT E X letter class. The height and width of the text may easily be altered 

with appropriate declarations in the preamble. The use of explicit English 

words and date style is no problem for other languages since they are all 

contained in special commands that can be redefined. 

The letter class has been designed so that if the \address command 

is omitted, a company letterhead will be printed instead. This presupposes 

that the file has been reprogrammed at the local installation for 

this purpose. Each employee using this house letter style will have certain 

personal entries to make, such as his or her room and/or telephone number. 

These commands have been provided for in the standard letter 

class. 

We have such a house letter style at our institute which we will illustrate 

here. However, it was found necessary to add some more personal entries, 

such as ‘Our Ref.’, ‘Your Ref.’, and email addresses. In addition, a command 

to print out ‘Subject:’ has also been added. 

To distinguish our local house style from that of standard L AT E X, we have 

named it mpletter. It contains most of the features of letter, since it in 

fact reads in that class file and then makes its alterations and additions. 

The \address command is not necessary, since all letters are printed with 

the institute letterhead, including its address. The recipient’s name and 

address are taken from the argument of the letter environment and are 

vertically centered in the space provided in the letterhead. 

The writer’s name and telephone extension are entered with 

\name{author} and \telephone{ext number} 

which, if given in the preamble, apply to all the letters in the file. If 

different letters have other authors, these commands must be made in 

the appropriate letter environment before the \opening command. New 

entry commands specific to mpletter are 

\yref{your code} 

\ymail{your date} 

\myref{our code} 

\subject{subj text} 

which produce the words 

Your Ref.:, Your letter of :, Our Ref.:, Subject:

16.2. A house letter style 357 

MAX–PLANCK–INSTITUT FÜR AERONOMIE Max–Planck–Str. 2 

Katlenburg–Lindau 

GERMANY 

MPI für Aeronomie, D–37191 Katlenburg–Lindau 

Mr George Murphy 

35 Waterville Rd. 

Centertown, Middlesex 

United Kingdom 

Dr P. W. Daly 

Tel.: 05556–401–279 

Email: 

daly @ linmpi.mpg.de 

May 10, 2003 

Your Ref.: GFM/sdf Your letter from: April 28, 2003 Our Ref.: PWD/sib 

Subject: L AT E X information 

Dear George, 

Thank you for your inquiry about the latest version of the L AT E X installation and 

additional packages. 

The entire T E X installation, with binaries and of course L AT E X, along with a large 

number of contributed packages, is distributed annually by the T E X Users Group, 

on their T E Xlive CD. 

I am sending you a copy of the current version of this CD, as you requested. 

In a separate directory named bibtex you will find the special bibliography 

formatting package files mentioned in ‘A Guide to L AT E X’. I hope you will find 

these of use. 

Do not hesitate to get in touch with me again if you have any further questions 

about the installation or running of the package. 

encl: 1 CD-ROM with T E Xlive contents 

cc: H. Kopka 

Telephone 05556–401– 1 

Telefax 05556–401– 240 

Telex 9 65 527 aerli 

Bank 

Kreis–Sparkasse Northeim 

41 104 449 (BLZ 262 500 01) 

Regards, 

Patrick W. Daly 

Train Station 

Northeim 

(Han.)


properly positioned under the letterhead together with the corresponding 

text argument. If any of these commands are missing, their text will not 

appear in the letter. 

Since ours is a German institute, there is an option german that translates 

all these words into their German equivalents. The entry commands 

have the same names, however. 

As in the standard letter class, the current date is printed automatically, 

but may be changed to any desired text with the command 

\date{date text} 

if one wishes to back-date a letter or to fix the dating within the letter file 

itself. This is often handy if one only keeps electronic copies of the letters 

for the record. Otherwise, when a hard copy is run off months later, it 

will appear with the new current date and not with the original one. 

The entry author that is given as the argument to the \name command 

appears in the letterhead as the name of the sender. It will also appear 

below the space left for the signature, unless it is overridden by the 

\signature command that specifies an alternative form of the name for 

this purpose. 

The sample letter on the previous page has been generated with our 

house style, using the following input text. 

\documentclass[12pt]{mpletter} 

\name{Dr P. W. Daly} \signature{Patrick W. Daly} 

\myref{PWD/sib} 

\date{May 10, 2003} 

\subject{\LaTeX{} information} 

\telephone{279} \internet{daly} 

\ymail{April 28, 2003} \yref{GFM/sdf} 


\begin{letter}{% 

Mr George Murphy\\35 Waterville Rd.\\ 

Centertown, Middlesex\\United Kingdom} 

\opening{Dear George,} 

Thank you for your inquiry about the latest version 

of the \LaTeX{} installation and additional packages. 

. . . . . . . . . . . . . . . . . . . . . . . . . . 

Do not hesitate to get in touch with me again if you 

have any further questions about the installation or 

running of the package. 

\closing{Regards,} 

\encl{1 CD-ROM with \TeX{}live contents} 

\cc{H. Kopka} 

\end{letter} 

\end{document}

16.3. A model letter customization 359 

The first page of our institute letter appears as shown, without a page 

number. If the text must continue beyond one page, the next page will 

have the heading 

MAX–PLANCK–INSTITUT FÜR AERONOMIE 

To Mr George Murphy May 10, 1998 Page 2 

The recipient’s name that appears in this heading is taken from the first 

line of the recipient argument in the \begin{letter}. This argument 

is split up by the L AT E X processing so that the first line is contained in 

the command \toname, and the rest of the lines in \toaddress. The 

words ‘To’ and ‘Page’ are in the standard commands \headtoname and 

\pagename, and may be changed by an appropriate language adaptation 

option as shown in Section D.4.2. 

16.3 A model letter customization 

Adapting the letter.cls class file to the requirements of a company 

style should present few difficulties to an experienced L AT E X programmer. 

Even a normal user may be able to make the necessary changes with the 

help of the example in this section. 

We present here the class file mpletter.cls that was used to produce 

the sample letter on page 357. It makes heavy use of the L AT E X programming 

features described in Appendix D. In order to understand it, one 

should be familiar with Section D.2. It will not be necessary to make 

any changes to the file letter.cls itself, since all modifications are in a 

separate class file that inputs the original. 

The new class file is to be called mpletter.cls. It begins by specifying 

the T E X format that it requires, and by identifying itself. 

\NeedsTeXFormat{LaTeX2e} 

\ProvidesClass{mpletter} 

It will be necessary to execute conditionals, so we will need the ifthen 

package described in Section 8.3.5. We will want a flag to decide if the 

letter is to be in German or not, determined by an option. Create the flag 

and define the option german to set it. 

\RequirePackage{ifthen} 

\newboolean{@german} 

\setboolean{@german}{false} 

\DeclareOption{german}{\setboolean{@german}{true}} 

All other options that are valid in the standard letter class will also 

be accepted here, so simply pass them on to that class with the default 

option. Then process all options before loading letter itself with the 

a4paper option. We only have A4 paper in our institute.


\DeclareOption*{\PassOptionsToClass{\CurrentOption}{letter}} 

\ProcessOptions 

\LoadClass[a4paper]{letter} 

This completes the preliminaries. So far, we have read in the standard 

letter class, along with the package ifthen, and defined a new option 

that is not present in the original class. Otherwise, all options and 

functions are unchanged, so far. 

We now define the new ‘name’ commands to contain language-sensitive 

text, such as ‘Subject’, which are not provided for in the basic class. The 

actual definitions will be executed by two commands \englishnames and 

\germannames. 

\newcommand{\englishnames}{% 

\newcommand{\yrefname}{\textsl{Your Ref.}} 

\newcommand{\ymailname}{\textsl{Your letter from}} 

\newcommand{\myrefname}{\textsl{Our Ref.}} 

\newcommand{\subjectname}{\textsl{Subject}} 

\newcommand{\telephonename}{Telephone} 

\newcommand{\stationname}{Train Station} 

\newcommand{\germanname}{GERMANY} 

\newcommand{\telcode}{[49]-5556-401} 

\newcommand{\postcode}{D--37191} 

} 

\newcommand{\germannames}{% 

. . . . . . . . . . . . . . . . . . . . . 

\newcommand{\telcode}{(05556) 401} 

\newcommand{\postcode}{37191} 

} 

\ifthenelse{\boolean{@german}} 

{\RequirePackage{german}\germannames}{\englishnames} 

The last lines test whether the flag @german has been set (by the option 

german), and if so, the package german is loaded, and the German names 

are defined, otherwise the English names. The package german already 

translates the standard names commands \toname, \headtoname, and 

\pagename, so they are not included in \germannames. 

Having settled the language problem, we now attack those commands 

for entering extra information in the header. Each of these stores its 

text argument in an internal command for future use. First the internal 

storage commands must be created. 

\newcommand{\@yref}{} \newcommand{\@ymail}{} 

\newcommand{\@myref}{} \newcommand{\@subject}{} 

\newcommand{\@internet}{} 

\newcommand{\yref}[1]{\renewcommand{\@yref}{\yrefname: #1}}


\newcommand{\ymail}[1]{\renewcommand{\@ymail}{\ymailname: #1}} 

\newcommand{\myref}[1]{\renewcommand{\@myref}{\myrefname: #1}} 

\newcommand{\subject}[1]{\renewcommand{\@subject} 

{\subjectname: #1}} 

\newcommand{\internet}[1]{\renewcommand{\@internet}{#1}} 

\newcommand{\INTERNET}{@linmpi.mpg.de} 

Set the dimensions of the text on the page and its margins. These 

numbers are appropriate for A4 paper (which does make the a4paper 

option superfluous). 

\setlength{\textheight}{215mm} \setlength{\textwidth}{160mm} 

\setlength{\oddsidemargin}{0pt} \setlength{\evensidemargin}{0pt} 

\setlength{\topmargin}{-20pt} \setlength{\headheight}{12pt} 

\setlength{\headsep}{35pt} 

The next step is to define some fixed fonts that are needed for the 

letterhead. We refer to the fonts with their explicit NFSS attributes; in 

this case they are all Computer Modern sans serif fonts in various sizes. 

These fonts will not change if totally different families are used for the 

body of the letter. 

\DeclareFixedFont{\xviisf}{OT1}{cmss}{m}{n}{17} 

\DeclareFixedFont{\xsf}{OT1}{cmss}{m}{n}{10} 

\DeclareFixedFont{\viiisf}{OT1}{cmss}{m}{n}{8} 

The letterhead itself is divided into two fields, the left one containing 

the institute name in large letters, the right one the address in a smaller 

font. Below the first horizontal line, the left field displays the name and 

address of the recipient, positioned to fit in the window of an envelope, and 

the right one has the personal data of the letter writer, name, extension, 

computer address. The widths of these fields are established. 

\newlength{\leftfield} \setlength{\leftfield}{117mm} 

\newlength{\rightfield} \setlength{\rightfield}{43mm} 

The total width of these two fields equals 160 mm, which is the same as 

\textwidth. 

Next, we place the institute name and address in several saved boxes. 

\newsavebox{\FIRM} \newsavebox{\firmaddress} 

\newsavebox{\firm} \newsavebox{\firmreturn} 

\sbox{\FIRM} 

{\parbox[t]{\leftfield} 

{\xviisf MAX--PLANCK--INSTITUT F\"UR AERONOMIE}} 

\sbox{\firm} 

{\xsf MAX--PLANCK--INSTITUT F\"UR AERONOMIE} 

\sbox{\firmreturn}


{\viiisf\underline{MPI f\"ur Aeronomie, 

\postcode{} Katlenburg--Lindau}} 

\sbox{\firmaddress} 

{\parbox[t]{\rightfield}{\viiisf\baselineskip10pt 

Max--Planck--Stra{\ss}e 2\\ 

\postcode{} Katlenburg--Lindau\\\germanname}} 

Using these boxes as building blocks, we put together the actual head and 

foot of the letterhead page in two further save boxes. 

\newsavebox{\firmhead} \newsavebox{\firmfoot} 

\sbox{\firmhead} 

{\parbox{\textwidth}{\usebox{\FIRM}\raisebox{6pt} 

{\usebox{\firmaddress}}\\[3pt] \rule{\textwidth}{1pt}}} 

\sbox{\firmfoot} 

{\parbox{\textwidth}{\rule{\textwidth}{0.6pt}\\[5pt] 

\viiisf\setlength{\baselineskip}{12pt}% 

\begin{tabular}[t]{@{}ll} 

\underline{\telephonename} & \telcode-1\\ 

\underline{Telefax} & \telcode-240\\ 

\underline{Telex} & 9\,65\,527 aerli 

\end{tabular}\hfill 

\begin{tabular}[t]{l} 

\underline{Bank}\\ 

Kreis--Sparkasse Northeim\\ 

41\,104\,449 (BLZ 262\,500\,01) 

\end{tabular}\hfill 

\begin{tabular}[t]{l@{}} 

\underline{\stationname}\\ 

Northeim\\ 

(Han.) 

\end{tabular} }} 

The box \firmhead is fairly clear: it is a \parbox of width \textwidth 

containing the boxes \FIRM and \firmaddress side by side, with a line 

below. The foot \firmfoot is also a \parbox of the same width, but 

containing three columns of general institute information, set in tabular 

environments. 

It now remains to have the head and foot boxes placed on the first 

page. In the letter class, there is a special page style named firstpage 

that is always invoked for the first page of a letter. This must be redefined. 

\renewcommand{\ps@firstpage} 

{\setlength{\headheight}{41pt}\setlength{\headsep}{25pt}% 

\renewcommand{\@oddhead}{\usebox{\firmhead}}% 

\renewcommand{\@oddfoot}{\raisebox{-20pt}[0pt] 

{\usebox{\firmfoot}}}%


\renewcommand{\@evenhead}{}\renewcommand{\@evenfoot}{}} 

This page style must (re)define the commands \@oddhead and \@oddfoot, 

which are always inserted at the top and bottom of odd pages, to place 

our special \firmhead and \firmfoot. The even pages are unimportant, 

since the first page is always odd. It is necessary to enlarge \headheight 

and \headsep to allow the big boxes to fit in. 

Subsequent pages are set with the headings or plain page styles. We 

want to modify the former to include the firm address once more. 

\renewcommand{\ps@headings} 

{\setlength{\headheight}{41pt}% 

\renewcommand{\@oddhead} 

{\parbox{\textwidth}{\usebox{\firm}\\[5pt] 

\slshape \headtoname{} \toname\hfill\@date\hfill 

\pagename{} \thepage\\ 

\rule[3pt]{\textwidth}{1pt}}} 

\renewcommand{\@oddfoot}{} 

\renewcommand{\@evenhead}{\@oddhead} 

\renewcommand{\@evenfoot}{\@oddfoot}} 

One small problem remains: the first time one of these page style 

commands is executed, the head and foot commands may not yet exist, 

causing \renewcommand to complain. We ensure that they are there at 

the start by predefining them with \providecommand (Section 8.3.1). 

\providecommand{\@evenhead}{}\providecommand{\@oddhead}{} 

\providecommand{\@evenfoot}{}\providecommand{\@oddfoot}{} 

Now make headings the active page style. 

\pagestyle{headings} 

There is only one last thing to do, and that is to redefine the opening 

command that prints the recipient’s address and the salutation. We add 

a bit more, including the personal data of the sender, as well as the 

reference information. The address goes in the left field, the personal 

data to the right. The references go in a line below the rule, followed by 

the subject line. These entries are tested first, and are only included if 

they are not blank. Several stored entry commands used here are part 

of the standard letter class, such as \toname and \toaddress. The 

\@date entry is either \today or whatever text was stored with \date. 

\renewcommand{\opening}[1]{\thispagestyle{firstpage}% 

\parbox[t]{\leftfield} 

{\usebox{\firmreturn}\\ 

\parbox[b][3.5cm][c]{\leftfield}{\toname\\\toaddress}}% 

\parbox[t]{\rightfield} 

{\fromname 

\ifthenelse{\equal{\telephonenum}{}} 

{}{\\ Tel.: \telcode-\telephonenum}


\ifthenelse{\equal{\@internet}{}} 

{}{\\{\viiisf Email: \@internet\INTERNET}} 

\\[5mm] \@date} 

\par 

\rule{\textwidth}{0.6pt} 

\makebox[\leftfield][l] 

{\ifthenelse{\equal{\@yref}{}} 

{\@ymail}{\@yref\hfill\@ymail\hfill}} 

\@myref\par 

\ifthenelse{\equal{\@subject}{}} 

{}{\@subject\par} 

\vspace{2\parskip} #1 \par\nobreak} 

The result of this formatting can be seen on page 357. It should be 

possible to modify the coding as needed for other organizations without 

too much difficulty. 

The field for the recipient’s name and address has been positioned so 

that it will appear in the address window of an envelope when properly 

folded. The smaller return address will also be visible through this 

window. The printing of extra address stickers is thus superfluous. 

As an exercise, the user should design a document class pletter for writing 

personal letters. With the tips for formatting company letters, it should not be 

difficult to add a letterhead of the form 

Sheila Joan McDonald 

Tel.: 234-9871 

31 Maple Drive 

Willowtown 

Hint: The name has been printed with the font cmdunh10 scaled \magstep4, 

declared with \newfont. Such a pletter class must remain in the personal 

directories of the user at a computer center. It would be rather embarrassing if 

everyone made use of the same personal letterhead.

Appendices

A 

The New Font Selection 

Scheme (NFSS) 

When T E X and L AT E X were invented, the fonts available for them were very 

limited in number. For this reason, the original L AT E X had an inflexible 

system of defining the fonts that were to be used, since it was not obvious 

that one might want to change them. The association between the highlevel 

font commands such as \large and \bf and the external font name 

that is ultimately selected was rigidly fixed internally. 

Today there are many additional fonts available, some of which should 

be used alongside the standard CM fonts, and others that should replace 

them altogether. For example, the Cyrillic fonts of Section 12.4.2 should be 

added parallel to the Latin fonts, but to make them operate automatically 

under the L AT E X size commands was a complex procedure. (We know: 

we have done it!) Similarly, installing PostScript fonts involved calling 

intricate interface macros. 

Another problem was the behavior of the font style and size commands. 

As explained in Section F.2.1, each of the font declarations \rm, 

\bf, \sc, \sl, \it, \sf, and \tt activates a particular font in the current 

size, overriding the previous declaration. One can select either a bold or 

an italic font, but there is no way to select a bold, italic one. Furthermore, 

selecting a new size automatically switches to \rm, an upright, Roman, 

medium font. That is, the attributes cannot be selected independently of 

each other. 

In 1989 Frank Mittelbach and Rainer Schöpf proposed a New Font 

Selection Scheme (NFSS) for L AT E X and a preliminary test package was 

ready in early 1990. A second release (NFSS2) was published mid-1993 

with many substantial changes. With the official release of L AT E X 2ε in 

June 1994, NFSS became firmly entrenched in the new standard. The new 

font declarations and commands are described in Sections 4.1.3 and 4.1.4. 

Here we explain the usage in more detail; in Section A.3 we describe how 

new sets of fonts may be installed. 

367

368 Appendix A. The New Font Selection Scheme (NFSS) 

A.1 Font attributes under NFSS 

According to the NFSS scheme, every character set can be classified by 

five attributes called encoding, family, series, shape, and size, which may 

be selected with the commands 

\fontencoding{encode} \fontfamily{fam} \fontseries{wt wth} 

\fontshape{form} and \fontsize{sz}{line sp} 

The encode attribute was new to the second release of NFSS. It specifies 

the layout of the characters within the font. Possible values for it are listed 

in Table A.1. It is unlikely that one would want to change encoding within 

a document, except to activate Cyrillic fonts perhaps. This feature is more 

for L AT E X programmers to install new fonts on the system. 

Table A.1: The NFSS encoding schemes 

Encoding Description Sample Font Page 

OT1 Original text fonts from Knuth cmr10 492 

OT2 Univ. Washington Cyrillic fonts wncyr10 497 

T1 The Cork (DC/EC) fonts ecrm1000 500 

OML T E X math letter fonts cmmi10 495 

OMS T E X math symbol fonts cmsy10 495 

OMX T E X math extended fonts cmex10 496 

U Unknown coding — — 

The argument fam in \fontfamily denotes a basic set of font properties. 

For the Computer Modern fonts, listed in Section G.2.2, all the serif 

fonts belong to the family cmr. The family cmss includes all the sans 

serif fonts while cmtt contains the typewriter fonts. A number of special 

decorative fonts are the only members of their families. Table A.3 on 

page 370 lists the CM fonts according to the family and other attributes. 

Note: The font attribute family has no relationship to the primitive 

T E X concept of the same name. A T E X family consists of three fonts of 

different sizes for use in math formulas as normal characters, and firstand 

second-level indices. 

The argument wt wth in \fontseries designates the weight (=boldness) 

and width of the characters. These are specified by 1 to 4 letters as 

shown in Table A.2 on the opposite page. 

The argument for \fontseries{wt wth} consists of the letter or letters 

for the weight, followed by those for the width. Thus ebsc indicates 

weight extrabold and width semicondensed while bx means weight bold 

and width expanded. The letter m is omitted when combined with any 

non-normal weight or width; if both are to be normal, it is sufficient to 

give m alone.

A.1. Font attributes under NFSS 369 

Table A.2: The NFSS series attributes 

Weight class Width class 

Ultralight ul Ultracondensed 50% uc 

Extralight el Extracondensed 62.5% ec 

Light l Condensed 75% c 

Semilight sl Semicondensed 87.5% sc 

Medium (normal) m Medium 100% m 

Semibold sb Semiexpanded 112.5% sx 

Bold b Expanded 125% x 

Extrabold eb Extraexpanded 150% ex 

Ultrabold ub Ultraexpanded 200% ux 

In \fontshape, the argument form is one of the letter combinations 

n, it, sl, or sc for selecting normal (or upright), italic, slanted, or small 

caps. 

The \fontsize attribute command takes two arguments, the first sz 

being the point size of the font (without the dimension pt explicitly given) 

and the second line sp being the vertical spacing from one baseline to the 

next. The second argument becomes the new value of \baselineskip 

(Section 3.2.4). For example, \fontsize{12}{15} selects a font size of 

12 pt with interline spacing of 15 pt. (The second argument may be given 

a dimension, such as 15pt, but pt is assumed if no dimension is stated.) 

Once all five attributes have been set, the font itself is selected with 

the command \selectfont. The new feature here is that the various 

attributes are independent of one another. Changing one of them does 

not alter the others. For example, if the selection 

\fontfamily{cmr} \fontseries{bx} \fontshape{n} \fontsize{12}{15} 

has been made for an upright, bold, expanded, Roman font of size 12 pt 

and interline spacing 15 pt, then when \fontfamily{cmss} is later selected 

for a sans serif font, the attributes weight and width bx, form n, 

and size 12(15pt) remain in effect when the next \selectfont is issued. 

Alternatively, all attributes but the size may be specified and the font 

activated immediately with the command 

\usefont{code}{family}{series}{shape} 

The table on the next page (by F. Mittelbach and R. Schöpf) lists the 

classification of the Computer Modern character sets according to the 

attributes \fontfamily, \fontseries, and \fontshape. That there are 

so many attribute combinations without a corresponding CM font may 

appear to be a weakness in the NFSS system, but it must be recalled that 

it is designed for the future. It may also be employed with the PostScript


Table A.3: Attributes of the Computer Modern fonts 

Series Shape(s) Examples of external names 

Computer Modern Roman — (\fontfamily{cmr}) 

m n, it, sl, sc, u cmr10, cmti10, cmsl10, cmcsc10, 

cmu10 

bx n, it, sl cmbx10, cmbxti10, cmbxsl10 

b n cmb10 

Computer Modern Sans Serif — (\fontfamily{cmss}) 

m n, sl cmss10, cmssi10 

bx n cmssbx10 

sbc n cmssdc10 

Computer Modern Typewriter — (\fontfamily{cmtt}) 

m n, it, sl, sc cmtt10, cmitt10, cmsltt10, 

cmtcsc10 

fonts, which are becoming ever more popular, to exploit their complete 

variability. 

Formally it is possible to set any combination of attributes; however, 

there may not exist any font matching all the attributes selected. If that is 

the case, then when \selectfont is called, L AT E X issues a warning stating 

which font has been activated in its place. The font size attribute of 

the \fontsize command may normally take on values of 5, 6, 7, 8, 9, 

10, 10.95, 12, 14.4, 17.28, 20.74, and 24.88, but other values may also 

be added. The second argument, the interline spacing, may take on any 

value since it is not something intrinsic to the font itself. 

With the \begin{document} command, L AT E X sets the five attributes to 

certain preset default values. These are normally standard encoding OT1, 

family cmr, medium series m, normal shape n, and the base size selected. 

The user may change these initial values within the preamble, or they 

might be set differently by a special option, say when a PostScript font 

has been selected. 

A.2 Simplified font selection 

The attribute commands \fontencoding, \fontfamily, \fontseries, 

\fontshape, and \fontsize, together with the command \selectfont, 

are the basic tools in the New Font Selection Scheme. The user need not

A.2. Simplified font selection 371 

employ these commands directly, but rather may make use of the higherlevel 

declarations presented in Sections 4.1.2 and 4.1.3. In fact, a font 

declaration like \itshape is defined as \fontshape{it}\selectfont. 

The high-level commands to select font sizes are: 

\tiny (5pt) \normalsize (10pt) \LARGE (17.28pt) 

\scriptsize (7pt) \large (12pt) \huge (20.74pt) 

\footnotesize (8pt) \Large (14.4pt) \Huge (24.88pt) 

\small (9pt) 

The sizes listed for the commands are those when 10pt (the default) has 

been selected as the basic size option in the \documentclass command; 

for 11pt and 12pt, they will all scale accordingly. 

The family declarations and their standard family attribute values are: 

\rmfamily (cmr) \sffamily (cmss) \ttfamily (cmtt) 

which are the three Computer Modern families, Roman, Sans Serif, and 

Typewriter (Section G.2.2). 

The series declarations and their initial attribute values are: 

\mdseries (m) \bfseries (bx) 

meaning that only a medium and a bold extended series are provided as 

standard. 

Finally, the shape declarations and their attribute values are: 

\upshape (n) \itshape (it) 

\slshape (sl) \scshape (sc) 

to select upright, slanted, italic, and Small Caps. 

Note that there are no high-level declarations for the encoding attributes. 

This is because there is normally no need to change encoding 

within a document. An exception might be to use Cyrillic fonts (coding 

OT2), in which case one could define 

\newcommand{\cyr}{\fontencoding{OT2}\selectfont} 

\newcommand{\lat}{\fontencoding{OT1}\selectfont} 

to be able to switch back and forth more conveniently. 

The family, shape, and series attributes may be reset to their standard 

values at any time with the \normalfont command, which also activates 

that font in the current size. 

For each of the above font attribute declarations there is also a corresponding 

font command (Section 4.1.4) that sets its argument in that 

font. Thus \textit{text} is almost the same as {\itshape text}, the 

only difference being that the command also contains the italic correction 

automatically. The complete list of such commands is:


Family: \textrm \textsf \texttt 

Series: \textmd \textbf 

Shape: \textup \textit \textsl \textsc 

Other: \emph \textnormal 

The \emph command is described in Section 4.1.1; \textnormal sets its 

argument in \normalfont. 

A.3 Installing fonts with NFSS 

In this section, we wish to elaborate on the internal workings of NFSS, 

showing how a set of attributes are associated with a particular font and 

how special symbols are assigned to their proper positions within various 

encoding schemes. 

A.3.1 Default attribute values 

We implied in Section A.2 that the font attribute declarations like \itshape 

are defined as \fontshape{it}\selectfont, whereas in fact they make 

use of certain default attributes. Thus the true definition of \itshape is 

\fontshape{\itdefault}\selectfont 

The default commands available are 

Family: \rmdefault \sfdefault \ttdefault 

Series: \mddefault \bfdefault 

Shape: \updefault \itdefault \sldefault \scdefault 

It is \itdefault that is defined to be the attribute it. 

It is also necessary to define the standard attributes chosen when 

the command \normalfont is issued. These are contained in the four 

defaults 

\encodingdefault \familydefault \seriesdefault 

\shapedefault 

All this may sound like a complicated route linking the high-level 

commands to a particular font. However, it does provide flexibility and 

modularity. The author only needs to know that three families, two series, 

and four shapes are available, and does not care what they really are. A 

programmer defines these with the defaults at a lower level. 

An example of how all the standard fonts may be replaced by PostScript 

ones, by simply redefining the three family defaults, is shown on page 379. 

Such redefinitions are much simpler than trying to alter the font declarations 

themselves, including \normalfont. Those definitions are in fact 

much more complex than implied here, whereas the default commands 

really are as simple as indicated.

A.3.2 Defining font commands 

A.3. Installing fonts with NFSS 373 

There are a number of commands available for defining new font declarations 

! and commands. These are intended mainly for L AT E X package programmers but 

may be used in a normal document as well. 

\DeclareFixedFont{\cmd}{code}{family}{series}{shape}{size} 

defines \cmd to be a declaration that selects the font of the specified attributes. 

It is rigidly fixed in all attributes. This is equivalent to \newfont except that the 

font is determined by attributes and not by name. 

\DeclareTextFontCommand{\cmd}{font specs} 

defines \cmd to be a font command that sets its argument according to font specs. 

This command is used internally to define all the font commands like \textbf, 

which is defined with font specs as \bfseries. 

\DeclareOldFontCommand{\cmd}{text specs}{math specs} 

defines \cmd to be a font declaration that may be used in math mode in the 

manner of L AT E X 2.09: as a declaration, not a command. It is useful for defining 

commands to be compatible with the old version, but should be avoided. For 

example, \it is defined with 

\DeclareOldFontCommand{\it}{\normalfont\itshape}{\mathit} 

A.3.3 Mathematical alphabets 

The font that has been activated for text processing does not influence 

the characters and their fonts in math mode, since special mathematical 

symbol fonts are used for this purpose. If a formula is to appear in 

bold face, the command \boldmath (Section 5.4.9) must be issued, which 

remains in effect until it is countermanded by \unboldmath. Both of 

these declarations must be made outside of math mode. 

These commands may still be employed in the same way under NFSS. 

However, the internal math font selection command is 

\mathversion{vers name} 

in which the argument vers name currently takes on values normal and 

bold. The declarations \unboldmath and \boldmath are defined in terms 

of this command. It is planned that special package files will become 

available to allow additional sets of math symbols. 

On the other hand, mathematical alphabet commands may be issued 

within math mode itself, to set letters as symbols in particular fonts 

(Section 5.4.2): 

\mathrm \mathcal \mathnormal \mathbf \mathsf \mathit \mathtt 

These are all commands operating on arguments rather than declarations, 

unlike the L AT E X 2.09 equivalents. 

New math font alphabets may be defined by the user. For example, to 

define a slanted math font \mathsl, give


\DeclareMathAlphabet{\mathsl}{OT1}{cmr}{m}{sl} 

which means that the new math font command \mathsl selects that font 

in family cmr with weight m and shape sl, which, with the normal font 

definitions, is font cmsl in the appropriate size. However, this font will 

be selected in all math versions, whereas it would be more suitable if a 

bold font were selected when \mathversion{bold} is in effect. This is 

accomplished by adding to the definition of \mathsl 

\SetMathAlphabet{\mathsl}{bold}{OT1}{cmr}{bx}{sl} 

which redefines \mathsl exceptionally for math version bold to be a font 

with weight bx. For the normal font definitions, this is cmbxsl in the 

current size. 

New math versions can be created with 

\DeclareMathVersion{vers name} 

and the fonts belonging to it are determined by issuing \Set... commands 

for each math alphabet or symbol font. 

A.3.4 Mathematical symbol fonts 

Mathematical symbols must be defined in a totally different manner from text 

! characters: they bear a command name (like \alpha), may come from various 

fonts, behave differently depending on type, and can appear in different sizes. 

Under L AT E X 2.09, the symbol names were fixed to the Computer Modern math 

fonts, but NFSS offers more flexibility for additional (or replacement) symbol 

fonts. 

A symbol font name is declared with the command 

\DeclareSymbolFont{sym fnt name}{code}{family}{series}{shape} 

which associates the name sym fnt name with the given set of attributes. This 

name is not a command, but rather an internal designation for use in defining 

symbols. The selected font is valid for all versions, but if a different font is to be 

associated with the same name under other versions, 

\SetSymbolFont{sym fnt name}{version}{code}{family}{series}{shape} 

may be used to redefine sym fnt name for that one version. 

The standard L AT E X setup declares 

\DeclareSymbolFont{operators}{OT1}{cmr}{m}{n} 

\DeclareSymbolFont{letters}{OML}{cmm}{m}{it} 

\DeclareSymbolFont{symbols}{OMS}{cmsy}{m}{n} 

\DeclareSymbolFont{largesymbols}{OMX}{cmex}{m}{n} 

the sequence of which is important for reasons that are built deeply into T E X 

itself. 

Once the symbol font names have been defined, they may be used to construct 

math alphabets and various types of symbols.


\DeclareSymbolFontAlphabet{\math alph}{sym fnt name} 

defines \math alph to be a math alphabet based on the font with the internal name 

sym fnt name. This command is to be preferred over \DeclareMathAlphabet if 

a symbol font exists with the proper attributes for the math alphabet. 

The primary command for defining symbols is 

\DeclareMathSymbol{\symbol}{type}{sym fnt name}{pos} 

which makes \symbol print the symbol in position pos of font sym fnt name. The 

pos is a number, in decimal (10), octal (’12) or hexadecimal ("0A) representation. 

The type specifies the functionality of the symbol and is one of 

\mathord an ordinary symbol 

\mathop a large operator like 

\mathbin a binary operator like × 

\mathrel a relational operator like ≥ 

\mathopen an opening bracket like { 

\mathclose a closing bracket like } 

\mathpunct punctuation 

\mathalpha an alphabetic character 

Math alphabet commands operate only on the symbols of type \mathalpha; 

other types always produce the same symbol, for a given math version, within all 

math alphabets. 

Similarly, math accent commands are established with 

\DeclareMathAccent{\cmd}{type}{sym fnt name}{pos} 

where type is either \mathord or \mathalpha; in the latter case, the symbol 

changes with the math alphabet. 

Math delimiters and radicals can appear in two different sizes. They are set 

up with 

\DeclareMathDelimiter{\cmd}{type}{sym fnt1}{pos1}{sym fnt2}{pos2} 

\DeclareMathRadical {\cmd} {sym fnt1}{pos1}{sym fnt2}{pos2} 

which define \cmd to print the smaller variant from position pos1 of font sym fnt1 

and the larger from position pos2 of sym fnt2. 

The sizes have not been specified in any of the above math declarations. This 

is because there are normally four different sizes available, depending on the 

math style, as explained in Section 5.5.2. However, these sizes must somehow be 

specified. This is done with 

\DeclareMathSizes{text}{math text}{script}{sscript} 

where the four arguments are numbers giving a point size. When the normal 

text font is size text pt, \textstyle will be in math text size, \scriptstyle in 

script, and \scriptscriptstyle in sscript size. For example, 

\DeclareMathSizes{10}{10}{7}{5} 

All the \Declare... and \Set... commands may only be called in the 

preamble.


A.3.5 Addressing the attribute values 

In some programming situations, it is necessary to make use of the current values 

! of the attributes without knowing what they are. These are stored in the internal 


\f@encoding \f@shape \tf@size 

\f@family \f@size \sf@size 

\f@series \f@baselineskip \ssf@size 

The values of these commands should never be changed directly. However, they 

may be used to test if they possess a certain value. Since they all contain the 

character @ in their names, they may only be used in class or package files, and 

not in the main document file. 

A.3.6 Defining fonts under NFSS 

Under NFSS, fonts are specified within a document by giving the attributes 

required and then calling \selectfont. How is the set of font attributes 

then associated with a particular external font name such as those found 

in Appendix G? This is done by means of font definition commands which 

are usually stored in files with extensions.def and.fd. 

First the declaration 

\DeclareFontEncoding{code}{text set}{math set} 

sets up a new encoding attribute named code; whenever a text font of 

this encoding is selected, text set is executed in order to redefine accent 

commands or other things that are coding dependent; similarly math set 

is called for every math alphabet of this encoding. It is possible to define 

default text set and math set with 

\DeclareFontEncodingDefaults{text set}{math set} 

This allows general text and math mode settings to be declared with this 

command, while more specialized ones, which are executed afterwards, 

appear in \DeclareFontEncoding. 

If no font can be found for the specified attributes, 

\DeclareFontSubstitution{code}{family}{series}{shape} 

declares the values of the attributes that should be substituted; substitutions 

are made in order of shape, series, then family; the encoding is 

never substituted. If even this fails to find a valid font, then 

\DeclareErrorFont{code}{family}{series}{shape}{size} 

determines which font is to be used as a last resort. 

A new family with a specified encoding scheme is established with 

\DeclareFontFamily{code}{family}{option}


where option is a set of commands that may be executed every time a font 

of this family and encoding is selected. 

The main font-defining declaration that associates external font names 

with font attributes is 

\DeclareFontShape{code}{family}{series}{shape} 

{font def }{option} 

where option is additional commands that are executed when one of these 

fonts is selected. 

The font def contains a series of size/font associations, each consisting 

of a size part, a function, an optional argument, and a font argument. 

For example, 

\DeclareFontShape{OT1}{cmr}{m}{n} 

{ gen * cmr 

cmr10 

cmr12 

cmr17}{} 

states that the medium series, normal shape members of the cmr family 

are to be represented by external fonts cmr5 . . . cmr12 for sizes 5–12 pt, 

by cmr10 scaled to 10.95 pt for sizes near 11 pt, and so on. If the specified 

size is not present, the nearest size within certain limits is taken instead. 

The size part consists of numbers in angle brackets, representing point 

sizes. The brackets may also contain ranges, as means all sizes 

up to, but excluding, 10 pt, means 10 pt to less than 14 pt, and 

indicates 24 pt and higher. The possible functions are: 

(empty) loads the named font scaled to the requested point size; if an 

optional argument in square brackets precedes the font name, it 

acts as an additional scaling factor; 

[.95] cmr10 would load cmr10 scaled to 95% of 11 pt; 

gen * generates the font name by appending the point size to the font 

argument; 

gen * cmr loads cmr12; 

genb * generates the font name by appending the point size times 100 

to the font argument (for DC and EC fonts, page 501); 

genb * ecss loads ecss1440; 

sub * substitutes a different font whose attributes are given in the font 

argument as fam/ser/shp; 

sub * cmtt/m/n this is best used when there is no font 

of the required attributes; a message is output to the monitor and 

to the transcript file;


subf * is like the empty function, but issues a warning that an explicit 

substitute font has been loaded; 

fixed * loads the specified font at its normal size, ignoring the size part; 

if an optional argument is given, it is the point size to which the font 

is scaled, as in 

fixed * [11] cmr12 which loads cmr12 at 11 pt when 

10 pt is requested. 

All the above functions may be preceded by an s to suppress messages 

to the monitor. Thus silent sub * is ssub *, and silent empty is s *. 

As another example, consider the definition of bold, italic, typewriter, 

for which there is no font in the Computer Modern collection: 

\DeclareFontShape{OT1}{cmtt}{bx}{it}{ 

ssub * cmtt/m/it }{} 

This substitutes (silently) for all sizes () the medium italic typewriter 

attributes. Which fonts those are is determined by a \DeclareFontShape 

command with those attributes. 

The font definition commands may be issued in a package file, or even 

in the document itself. However, the normal procedure is to store each of 

the \DeclareFontEncoding commands in a file named codeenc.def (for 

example, ot1enc.def for the OT1 encoding), and to place the commands 

\DeclareFontFamily and \DeclareFontShape in a file whose name 

consists of the encoding and family designations plus the extension.fd. 

For example, the shape specifications for encoding OT1 and family cmr are 

to be found in ot1cmr.fd. When a coding and family combination that is 

not already defined is selected, LAT EX tries to find the appropriate.fd file 

for input. It is therefore not necessary to input such files explicitly, for 

they are loaded automatically as required. 

Package: However, it is important that the encoding be declared beforehand. If 

fontenc it is not already known in the current format, \DeclareFontEncoding 

must be issued, either explicitly, or by loading the codeenc.def file. One 

way to do this is to invoke the standard package fontenc, as for example 

\usepackage[OT2,T1]{fontenc} 

where the desired codings are listed as options in square brackets, the 

last of which is made current. 

The normal user will never need to worry about such problems. However, 

L AT E X programmers will find things considerably easier for them. 

For example, to install PostScript fonts under NFSS is almost trivial. The 

common PostScript fonts are already defined as separate families in .fd 

files of their own; for example, ot1ptm.fd associates PostScript Times 

fonts to a family named ptm. A package to activate these fonts is supplied 

under the name times.sty containing essentially the lines

\renewcommand{\rmdefault}{ptm} 

\renewcommand{\sfdefault}{phv} 

\renewcommand{\ttdefault}{pcr} 

\renewcommand{\bfdefault}{b} 


This makes Times ptm the default Roman family, invoked with the command 

\rmfamily, Helvetica phv the default sans serif family (called by 

\sffamily), and Courier pcr the default typewriter family (activated by 

\ttfamily). The default attribute for bold face is defined to be b instead 

of the regular bx, since bold extended is not provided by these fonts. 

Another two examples of the usefulness of NFSS are the Cyrillic fonts 

of the University of Washington (Section 12.4.2) and the extended EC fonts 

with the Cork encoding (Section G.4.3). Both of these may be activated 

within a document simply by selecting another encoding, OT2 for Cyrillic, 

T1 for the EC fonts. (These encodings must first be declared, for example 

with the fontenc package as illustrated on the facing page.) 

A.3.7 Encoding commands 

In L AT E X, special characters and accents are addressed by means of commands, 

! such as \O to print the Scandinavian letter Ø. The position of this character in 

the font tables depends on the encoding (character 31 in OT1 and 216 in T1), so 

that it is necessary to redefine all such symbol commands when the encoding is 

altered. This is carried out with the help of certain encoding commands, which 

normally appear in the codeenc.def file along with the \DeclareFontEncoding 


To define a command that functions differently in the various encodings, 

\ProvideTextCommand{\cmd}{code}[narg][opt]{def } 

\DeclareTextCommand{\cmd}{code}[narg][opt]{def } 

are available and behave just like \providecommand and \newcommand except 

\cmd has the definition def only when the encoding code is active. Thus \cmd 

may have different definitions for each encoding. 

\DeclareTextSymbol{\cmd}{code}{pos} 

defines \cmd to print the character in the font position pos when encoding code 

is active. 

\DeclareTextAccent{\cmd}{code}{pos} 

defines \cmd to be an accent command, using the character in font position pos 

as the accent symbol, when encoding code is active. 

\DeclareTextComposite{\cmd}{code}{letter}{pos} 

\DeclareTextCompositeCommand{\cmd}{code}{letter}{def } 

define the action of command \cmd followed by the single letter either to print the 

character in font position pos or to execute the definition def. These declarations 

are most useful with the T1 encoding, where many accented letters are separate


symbols on their own (Section G.4.3). Thus \’{e} in OT1 prints an acute accent 

(character 19) over the letter e, while in T1, it prints the single character in 

position 233. This behavior is achieved with 

\DeclareTextAccent{\’}{OT1}{19} 

\DeclareTextComposite{\’}{T1}{e}{233} 

The command must already have been defined for the encoding, either with 

\DeclareTextAccent or with \DeclareTextCommand; in the latter case, it must 

be defined to take a single argument. 

All of the above definition commands create new commands for a specific 

encoding. If the defined commands are invoked in some other encoding, an 

error message is issued. Default definitions may be provided for all unspecified 

encodings with 

\DeclareTextCommandDefault{\cmd}[narg][opt]{def } 

\ProvideTextCommandDefault{\cmd}[narg][opt]{def } 

\DeclareTextAccentDefault{\cmd}{code} 

\DeclareTextSymbolDefault{\cmd}{code} 

where the first two create a default definition that applies to all unspecified 

encodings, while the second two declare which encoding is to be taken as the 

default.

B 

The LAT E X Clockwork 

In this appendix, we describe how the T E X program and L AT E X files are 

installed, how they are organized, what their roles are. Throughout this 

book we have given examples of contributed packages, but here we list 

those packages and other files that belong to the ‘kernel’, the essential 

installation. In Section B.6, we explain how L AT E X ticks, what happens 

during a processing run, and what all the various file types mean. 

B.1 Installing L AT E X 

We explain first L AT E X installations in general before looking at the particular 

one provided on the enclosed T E XLive CD. 

B.1.1 T E X implementations 

One must have the T E X program and its auxiliaries (METAFONT, font files) 

before L AT E X can be set up on top of it. Installing T E X is a somewhat 

daunting experience, but thankfully it need not be done very often and 

there are many ready-to-run implementations available for practically 

every computer operating system. These can be obtained from CTAN 

under the directory systems (Figure B.4 on page 390) or from the T E X 

Users Group (www.tug.org). 

The system delivered with the T E XLive CD is that developed by Karl 

Berry known generically as Web2c. This name derives from the fact that it 

converts Donald Knuth’s original T E X source files from his Web language 

(no relation to the World Wide Web) into the C programming language for 

subsequent compilation with a C compiler. The teT E X implementation, by 

Thomas Esser, applies Web2c to Unix machines, while Fabrice Popineau’s 

fpT E X is the Win32 (Windows 95, 98, NT, 2000, XP) version. 

Another excellent T E X implementation for Windows is MikT E X by Christian 

Schenk, available from CTAN under systems/win32/miktex/. 

381

382 Appendix B. The L AT E X Clockwork 

The implementation for DOS, emT E X by Eberhardt Mattes is still available 

from CTAN, but is no longer appropriate for modern PCs. We used 

to produce the first two editions of this book and found it superb for its 

time. 

Commercial packages for T E X also exist, offering graphics display of 

the output while you type: for Windows (Y&Y Inc., www.YandY.com/) and 

Textures for Macintosh (Blue Sky Research, www.bluesky.com/). A ‘whatyou-see-is-what-you-get’ 

T E X system is Scientific Workplace, which can be 

purchased from MacKichan Software Research (www.tcisoft.com/). 

It should be stressed that the T E X and pdfT E X programs in all the 

above implementations are identical (for the same version number). They 

provide ready-to-run executable programs from the same source code, 

which are necessarily different for different computer types. There are 

also minor variations as to how the programs are run. 

The L AT E X files are all ascii so they are fully compatible with all systems. 

One uses the already installed T E X or pdfT E X program to generate the L AT E X 

formats (Section B.1.3). Even this task is normally done for you by the 

above implementations. 

A graphics interface to T E X and L AT E X is not absolutely essential, but 

today it is almost unthinkable to do without one. This is an editor program 

for writing and managing the L AT E X source files, and calling the various 

programs (L AT E X, BIBT E X, MakeIndex) and invoking viewers by clicking icons. 

For Unix, the emacs is normally used for this; for Windows, the two editor 

programs mentioned in Section 1.6.2, Winshell and WinEdt can be highly 

recommended. 

B.1.2 The T E XLive CD 

The T E X Users Group, TUG issues a set of CDs each year to its members 

containing the latest T E X and L AT E X installations, with executable files for 

Win32, Linux, Macintosh, and all known flavors of Unix. As of the 7th issue 

in 2002, it was necessary to split the contents over 2 CDs, separating the 

Unix versions from the others. Each CD is otherwise complete, missing 

only the executables of the other one. 

The T E XLive CDs are maintained by Sebastian Rahtz who has kindly 

agreed to allow the CD for Windows, Linux, Mackintosh to be included with 

this book, as it stands in June 2003. It thus corresponds to something 

between the 7th and 8th regular issues. 

The startup window in Figure B.1 appears when the CD is inserted into 

your computer. If the autorun function does not work, you can run the 

autorun.exe program directly from the CD. 

You then proceed with the installation by clicking Install on the upper 

bar. You then follow the instructions on the various pages that then 

appear, making selections as you go. One choice is whether to install 

to run from the CD, or to install to the hard disk. In the former case,

Figure B.1: The T E XLive welcome 

B.1. Installing L AT E X 383 

you have the entire installation available with little lost of storage, but 

will need to have the CD mounted in the player whenever you run L AT E X. 

Processing time will also be slower. 

If you elect to install only part of the system to hard disk, you may still 

add further packages and support programs later by clicking maintenance. 

Or you may run texsetup --maintenance which does the same thing 

bypassing the Welcome window. Running texsetup --help produces a 

list of all the options. 

You may also click Documentation→View TexLive doc to open the 

instruction manual for T E XLive, with details for all systems. (Or you 

can go directly to \texmf\doc\tldoc\english\ and select live.pdf or 

live.html.) 

Documentation for most of the packages and collections are available 

on the CD under \texmf\doc\..., if the author has provided a description. 

We often refer to such manuals throughout this book, indicating 

more precisely where they are to be found. By clicking Documentation→Run 

TeXDocTK, you may open a documentation browser (Figure B.2) 

to assist finding descriptions for any module. 

The T E XLive installation program will do all the housekeeping for you,


Figure B.2: The T E XLive documentation browser 

placing files in the right directories, setting up font mapping files, and 

generating the formats. However, there might still be times when you 

want or have to do some of this yourself. This we explain in the next 

sections. 

B.1.3 Making the L AT E X format 

Adding or updating L AT E X on to an existing T E X installation by hand is 

relatively straight forward. One must obtain the source files, which 

on T E XLive are located at \texmf\source\latex\base\ or on CTAN 

at tex-archive/macros/latex/base/. At CTAN, instruction files with 

extension .txt are available for the various implementations, such as 

miktex.txt and web2ctex.txt, as well as the general instruction file 

install.txt. 

These source files are a set of packed files consisting of a large number 

of .dtx files, the integrated source and documentation files. These are 

unpacked by running initex on the batch job file unpack.ins. This 

generates the necessary.cls,.sty,.clo, and other files. It also constructs 

the fundamental file latex.ltx which is needed to produce the format 

file. On CTAN, the unpacked files are already provided in a parallel 

directory unpacked. 

The next step should be to run initex on the latex.ltx file to produce 

the L AT E X format latex. However, before doing that, there are a number 

of configuration aspects to be considered. At several points during the 

processing, certain files are read in, but if a file of the same name but 

extension .cfg exists, it is loaded instead. It is this device that permits 

one to configure the final format for local conditions or desires. The 

possible configuration files are: 

texsys.cfg offers the possibility of adding some very machine-specific 

adjustments for older versions of T E X or for some peculiarities of 

the T E X installation; information is to be found in the specific .txt 

file, or in ltdirchk.dtx;

B.1. Installing L AT E X 385 

fonttext.cfg is loaded in place of fonttext.ltx if it is present; this 

defines the fonts that are to be available for processing text; details 

can be found in fontdef.dtx; 

fontmath.cfg is loaded in place of fontmath.ltx if it is present; this is 

the equivalent of fonttext.cfg for math fonts; 

preload.cfg is loaded in place of preload.ltx if it is present; this 

determines which fonts are preloaded into the format; a number of 

other preload files may be extracted from preload.dtx, any one 

of which may be renamed to.cfg to be implemented; 

hyphen.cfg is loaded in place of hyphen.ltx if it is present; this specifies 

the hyphenation patterns and their assignments; by default, 

the patterns in hyphen.tex are loaded into language 0; the babel 

system provides such a configuration file (Section 11.1) which must 

be used if babel is to be incorporated into the format, in which case, 

language.dat must be tailored to local requirements. 

The only configuration file that you are likely to want to change is 

hyphen.cfg, especially if you are going to be using L AT E X for languages 

other than American English. 

Once any configuration files have been set up and located where T E X 

can read them, initex may be run on latex.ltx with 

initex latex.ltx or 

tex -ini latex.ltx 

The resulting latex.fmt must be placed where format files are read. 

For fpT E X as installed by T E XLive, this is in texmf-var\web2c\. Other 

unpacked files must be moved to texmf\tex\latex\base\; these are 

latexbug.tex, testpage.tex, lablst.tex, idx.tex, nfssfont.tex, 

small2e.tex, sample2e.tex, and docstrip.tex, 

and all files with extensions.cls,.clo,.sty,.fd,.def, and.cfg. 

Files with extension.ist are moved to texmf\makeindex\base\. 

To make up the L AT E X format with pdfT E X, run instead 

pdfinitex -fmt=pdflatex latex.ltx or 

pdftex -ini -fmt=pdflatex latex.ltx 

This processes latex.ltx base file once more, but names the resulting 

format file pdflatex.fmt rather than latex.fmt. This is necessary 

because the format files may only be used with the program that generated 

them, so the two must be distinguished even though they both 

are essentially the same thing. Move this format too to the format file 

directory. 

The actual commands to invoke T E X or pdfT E X with L AT E X (what we 

normally call ‘running L AT E X or pdfL AT E X’) are


tex &latex 

pdftex &pdflatex 

meaning ‘run the T E X or pdfT E X program with the specified format file’. 

The system should have shortcuts named latex and pdflatex that translate 

to the above. 

The web2c system, for which fpT E X is the Win32 implementation, has 

a different method. It stores all the necessary configuration information 

in a file named texmf.cnf located in the texmf\web2c directory. 

This sets up environments according to the name of the program being 

run, and in particular, adds the format name automatically for a T E X-like 

program. Thus a T E X program named latex will use the latex.fmt format 

file if none is specified. However, there is no real program named 

latex, so on Unix systems, an alias with this name is established to 

point to the tex executable program. On Win32, the actual T E X executable 

code is in a dynamic link library tex.dll and very small files named 

tex.exe, latex.exe exist to use this one library. These .exe files are 

in fact identical, and are really only there to ensure that different formats 

are invoked automatically when they are called. Similarly, there are 

pdftex.dll, pdftex.exe, and pdflatex.exe. If any additional formats 

are ever needed for these programs, one only needs to copy one of these 

.exe files to the new name. 

B.1.4 Updating the database 

T E X works with a very large number of files, for classes, packages, fonts, 

formats, and so on. A set of environments are used to indicate where the 

program should search for these files. With web2c, these are all defined 

in the texmf.cnf configuration file mentioned above. The user normally 

does not have to worry about this. However, to speed up the search, 

a database is prepared giving the exact location of all the files in the 

search paths. This database will be created and updated automatically if 

installation programs are used, such as that for T E XLive. 

If the user should add a new package or fonts collection by hand, 

without going through an installation program, it will be necessary to 

update this database. Under web2c (and fpT E X), this is carried out with 

the command 

mktexlsr 

Under MikT E X, one must run 

configure --update-fndb 

or click ‘refresh now’ in the MikT E X options program. 

It is not necessary to update the database if existing files are being 

replaced. It is only the names and locations of the files that go into the 

database, not their actual versions or sizes.

B.2 Obtaining the Adobe euro fonts 

B.3. T E X directory structure 387 

As pointed out in Section 2.5.8, the type 1 euro currency symbol fonts 

from Adobe Systems Inc. that are used with the europs and eurosans 

packages cannot be distributed on CDs like T E XLive, nor on CTAN, since 

Adobe reserves the exclusive right to distribution, even though free of 

charge. T E XLive and CTAN do provide all other files needed, like the font 

metric.tfm and font definition.fd files. 

You can obtain the Adobe euro.pfb files from www.adobe.com/type/ 

eurofont.html, downloading the appropriate file for your computer type. 

They must be unpacked (on Windows, for example, you just execute the 

downloaded eurofont.exe) and then rename them as follows: 

Windows PostScript T E X 

_1______.PFB EuroSans-Regular zpeurs.pfb 

_1I_____.PFB EuroSans-Italic zpeuris.pfb 

_1B_____.PFB EuroSans-Bold zpeubs.pfb 

_1BI____.PFB EuroSans-BoldItalic zpeubis.pfb 

_3______.PFB EuroSerif-Regular zpeur.pfb 

_3I_____.PFB EuroSerif-Italic zpeuri.pfb 

_3B_____.PFB EuroSerif-Bold zpeub.pfb 

_3BI____.PFB EuroSerif-BoldItalic zpeubi.pfb 

_2______.PFB EuroMono-Regular zpeurt.pfb 

_2I_____.PFB EuroMono-Italic zpeurit.pfb 

_2B_____.PFB EuroMono-Bold zpeubt.pfb 

_2BI____.PFB EuroMono-BoldItalic zpeubit.pfb 

The first column is the name as unpacked on Windows, the second the 

internal PostScript name, and the third, the name used by T E X. The 

.pfb files must be renamed to those in the third column, and placed in 

texmf\fonts\type1\adobe\euro (or better in texmf-local). 

The font map file zpeu.map must also be added to the lists of font maps 

for dvips, dvipdfm, and pdfT E X (Section 10.1.3). The T E XLive installation 

does all this for you; you only need to get and store the.pfb files. 

B.3 T E X directory structure 

The T E X system, of which L AT E X is only one part, albeit a large one, requires 

a very large number of files, for fonts, formats, classes, packages, bibliographies, 

and much more. Section B.6 lists the types of files and their 

extensions, at least for what concerns L AT E X. Other flavors of T E X have 

their own types in addition.


texmf 

bibtex dvipdfm fonts metafont source 

doc dvips makeindex pdftex tex 

pk source tfm type1 vf 

generic latex plain 

babel config dvips hyphen amsfonts base config 

base hyperref natbib psnfss tools 

Figure B.3: The TDS directory tree 

In the early days, anyone setting up a T E X installation would try to 

create order of some sort by distributing the files into different directories 

(also called folders) according to their functions. However, everyone had 

their own ideas how to do this. Thus changing from one installation to 

another required a totally new road map. 

The T E X Users Group, TUG, commissioned a working group to prepare 

a recommended standard, which published its results in 1994. The 

latest version of this paper is on T E XLive at \texmf\doc\tds\ in various 

formats. The T E X Directory Structure (TDS) described there can only be a 

recommendation, but it is fairly universally followed today. The T E XLive 

CD certainly conforms to it. 

Figure B.3 offers an overview of TDS. The top directory is named texmf, 

to demonstrate that it holds the entire T E X and METAFONT systems. 

This may not be the absolute top directory on any installation. For 

example, if you install to a Windows hard disk, you might find it in 

\program files\texlive\texmf. Furthermore, there might be parallel 

directories named texmf-var (where the system writes files it creates, 

like font.pk files) and texmf-local for any private packages you install 

yourself. The idea is to keep the main texmf tree intact, to let the T E XLive 

installation program manage it. These parallel trees should have the same 

structure, but may be incomplete. 

The subdirectories under texmf are the main classification of the 

subsequent files, sorted mainly by program, such as BIBT E X (Section 14.1),

B.4. The CTAN server 389 

dvipdfm (Section 10.2.2), dvips (Section 10.1.1), MakeIndex (Section 9.4.3), 

METAFONT (Section G.3), and pdfT E X (Section 10.2.3). Others are 

doc containing documentation of all sorts; its subdirectories are similarly 

structured to those under tex; most of the documentation files that 

we have referred to in this book are to be found here; 

fonts is subdivided according to the types of files explained in Section 

G.1: pk for the pixel files (these usually go into texmf-var), 

tfm for the font metric files, vf for virtual font files, source for the 

METAFONT.mf sources, type1 for the.pfb or.pfa PostScript outline 

font files; the contents of each of these directories are further sorted 

by supplier (adobe, public, . . . ), and then by font family (times, 

cm, . . . ); 

source contains the delivery files for packages and collections; for L AT E X, 

these are the.dtx and.ins files, found under latex\base; similarly 

for packages, such as latex\geometry containing geometry.dtx 

and geometry.ins; these are not needed any more once installed; 

the subdirectories under source mirror those under tex; 

tex is the main directory for files directly related to T E X; some of its 

subdirectories are 

generic for files that apply to all T E X applications, like hyphenation 

patterns; 

latex for all the L AT E X related files; this is further subdivided into 

directories for individual packages; only a handful are shown in 

the figure; base contains the fundamental L AT E X files, the kernel; 

plain for Plain T E X, which is often just called T E X (Section 1.3.2). 

This directory structure allows a user to locate documentation, source, 

running files for a package, collection, program, on all installations. The 

documentation and source files can readily be removed, or not copied 

at all from the CD, since they play no role during the processing. This 

structure makes it easy to remove such unnecessary branches of the tree. 

B.4 The CTAN server 

The easiest way to obtain the L AT E X software not available on the T E XLive 

CD is via one of the network servers supported by various educational 

institutions. They also provide the most up-to-date standard L AT E X installation, 

T E X extensions, programs for running T E X, collections of drivers for 

different printers, hyphenation patterns for other languages, etc. Such a 

server is a treasure trove for the dedicated T E X and L AT E X user.


tex-archive 

fonts systems support macros 

msdos win32 

amsfonts cm ec latex 

latex2html tex4ht 

doc 

psfonts 

language 

hyphenations 

latex latex209 

required base unpacked contrib 

supported other 

amslatex babel cyrillic graphics psnfss tools 

Figure B.4: Partial directory tree of CTAN servers 

There are three main network servers for T E X and L AT E X located in the 

United States, Great Britain, and Germany, forming the Comprehensive T E X 

Archive Network, CTAN. They are not only kept up to date but also regularly 

exchange contributions among themselves. They are all reachable 

on the Internet. 

Country Internet address 

USA ftp://ctan.tug.org/tex-archive/ 

UK ftp://ftp.tex.ac.uk/tex-archive/ 

Germany ftp://ftp.dante.de/tex-archive/ 

In addition, there are many full and partial mirrors of these sites. See the 

TUG home page, http://www.tug.org, for details. 

The directory structure is the same for all three servers. A sketch of 

the main parts of the directory tree is presented in Figure B.4. 

The latest version of the L AT E X source files may be found under the 

subdirectories base or unpacked, where in the latter the files are already 

unpacked and latex.ltx is ready. The L AT E X extensions and general style 

files that have been contributed by other users are located in the subdirectory 

contrib/supported, with each contribution in a subdirectory of its 

own. Extensions offered by the L AT E X3 Team are found in the required 

subdirectory, which include babel, graphics, tools, and so on.

B.5 Additional standard files 

B.5. Additional standard files 391 

Throughout this book we have pointed out and explained many contributed 

packages that expand L AT E X functionality. However, there are a 

number of service documents, additional classes, and packages that are 

delivered with the basic L AT E X installation, found in texmf\tex\latex\ 

base\. We list them all here, although some have been mentioned in other 

sections. 

B.5.1 Special documents 

The standard installation includes a number of special ‘documents’, files 

with the.tex extension. These are: 

small2e.tex a short sample document; 

sample2e.tex a longer sample document; 

lablst.tex (Section 9.2.1) for printing out all the cross-reference labels, 

with translations and page numbers; it asks interactively the name 

of the document that is to be so listed; 

idx.tex (Section 9.4) lists all the \index entries for a document, ordered 

in two columns, by page; it asks interactively the name of the 

document that is to be so listed; 

testpage.tex to test the positioning of the text on a page; this is more a 

test of the printer driver, to make sure that the margins are correct 

and that the scaling comes out properly; 

nfssfont.tex is a document to print out a font table, sample text, and 

various other tests; it asks for the name of the font interactively; the 

command \help prints out a list of commands that can be given; 

docstrip.tex (Section D.7.1) this is more a program than a document; it 

is the basic tool for unpacking operating files from their documented 

source files; it not only removes comments, but also includes alternative 

coding depending on various options. 

B.5.2 Extra classes 

In addition to the standard classes that have already been discussed (book, 

report, article, letter) the standard installation also contains 

proc for producing camera-ready copy of conference proceedings; it is a 

variant of the two-column article class; it contains one extra command: 

\copyrightspace, which leaves blank space at the bottom 

of the first column for typing in a copyright notice;


minimal a bare-bones class for testing and debugging; 

slides (Section 15.1) the former SLIT E X, for producing presentation material; 

ltxdoc (Section D.7.2) to produce class and package documentation. 

B.5.3 Standard packages 

The standard packages, some of which are described elsewhere, are: 

alltt (Section 4.9.1) enables the alltt environment which is much like 

verbatim except that \ { } behave as normal; this means that 

commands included in the text are executed and not printed literally; 

doc (Section D.7.2) provides many special features for documenting L AT E X 

packages; 

exscale allows the mathematical symbols that appear in different sizes 

to scale with the font size option given in the \documentclass 

command; normally the size of these symbols is fixed, independent 

of the basic size of the text; 

flafter (Section 7.2, page 171) ensures that floating objects (figures and 

tables) do not appear before their position in the text; 

fontenc (Section A.3.6, page 378) declares font encodings by loading 

their.def files; the names of the encodings are listed as options; 

graphpap (Section 13.1.5) provides the command \graphpaper which 

produces coordinate grids in the picture environment; 

ifthen (Section 8.3.5) allows text or command definitions depending on 

the state of various conditionals; one may test the state of a boolean 

switch, the value of a counter or length, or the text definition of a 

command; 

inputenc (Section D.5) allows a L AT E X document file to be prepared with 

a system-dependent input encoding scheme for special characters; 

latexsym loads the 11 special L AT E X symbols in the lasy fonts (Section 

G.2.2), which were part of L AT E X 2.09; this file is automatically 

read in in compatibility mode; 

makeidx (Section 9.4.3) provides the commands that may be used with 

the MakeIndex program to generate a keyword index; 

showidx (Section 9.4) makes the \index entries visible by printing them 

as marginal notes;


shortvrb (Section 4.9.1) provides the commands \MakeShortVerb and 

\DeleteShortVerb, to designate (and remove) a certain character 

as the equivalent of \verb for printing literal text. The character, 

preceded by a backslash, is given as an argument to these commands. 

After \MakeShortVerb{\|} has been issued, |\cmd{x}| is 

equivalent to \verb|\cmd{x}|, producing \cmd{x}; 

syntonly provides the command \syntaxonly which, when given in the 

preamble, suppresses all output to the .dvi file, but continues to 

issue errors and warnings; the job runs about four times faster than 

normal; this is used to check the processing when output is not 

immediately needed; 

t1enc makes the T1 font encoding (Section G.4.3) the standard, for use 

with DC or EC fonts; 

textcomp makes the special symbols in the text companion fonts available 

in text mode (Section G.4.4); 

tracefnt is a diagnostic tool for checking NFSS font selections; the 

options given with \usepackage control its action: 

errorshow suppresses all warnings and information messages to 

the monitor, which are still sent to the transcript file; only error 

messages are printed to the monitor; 

warningshow prints warnings and error messages on the monitor; 

this behavior is much the same as when the package is not used 

at all; 

infoshow (default option) sends font selection information to the 

monitor, which is normally only written to the transcript file; 

debugshow writes considerably more information to the transcript 

file about every change of font; can produce a very large transcript; 

pausing turns all warnings into error messages so that processing 

temporarily stops, awaiting user response; 

loading shows the loading of external fonts. 

B.5.4 The tools packages 

The members of the L AT E X3 Team, as individuals, have written a number 

of packages that they collectively make available; these are to be found in 

the texmf\tex\latex\tools\ directory. Although contributions, they 

are considered to be part of the standard installation. 

Many of these have been described elsewhere in this book, but we list 

them all here.


afterpage (Section 7.1 on page 171) permits commands to be saved 

and executed at the end of the current page. This is useful 

for forcing stubborn floats to be output right away, with 

\afterpage{\clearpage}. If \clearpage were issued directly, 

a new page would be inserted at that point. With \afterpage, the 

command is reserved until the end of the page. 

array (Section 4.8.4 on page 106) is a re-implementation of the tabular 

and array environments with several additional features. 

bm permits individual math symbols to be printed in bold face with 

command \bm{\sym}. Thus $\alpha + \bm{\beta}$ produces 

α + β without recourse to the \boldmath command which would 

set the whole formula in bold face. There is also a command 

\DeclareBoldMathCommand{\name}{\sym} that defines \name to 

be \bm{\sym}. 

calc re-implements the counter and length commands to allow ‘normal’ 

arithmetic with them, such as 

\setlength{\mylen}{3cm + \textwidth} 

\setcounter{page}{\value{section} * 3} 

\parbox{\linewidth / \real{1.6}}{...} 

Note that the arithmetical expressions can be used not only to set 

values but also as arguments to commands and environments. 

dcolumn (Section 4.8.4) requires the array package; it allows decimal 

point alignment in the tabular tables. 

delarray (Section 4.8.4) requires the array package; it permits large 

bracketing symbols to be put around array environments, for making 

matrices of various sorts. 

enumerate (Section 4.3.5, page 74) re-implements the enumerate environment 

so that an optional argument determines the numeration 

style. 

fileerr.dtx when unpacked, produces a set of small files that may 

be used to reply to T E X when a file is not found; this makes the 

responses similar to those for error messages; the files are named 

h.tex, e.tex, s.tex, and x.tex; thus, replying x〈return〉 loads the 

last one, which terminates the input. 

fontsmpl is a package to output a sample text, accents, and special characters 

for a given font; with the accompanying file fontsmpl.tex, a 

family such as cmr can be printed with all attributes and encodings. 

ftnright puts footnotes in the right-hand column in two-column mode.


hhline requires the array package; allows more flexibility in putting 

horizontal rules in tables. 

indentfirst (Section 3.2.4 on page 46) indents the first paragraph of 

sections, which normally are not indented. 

layout (Section 3.2.5 on page 47) defines the command \layout which 

draws a diagram of the current page format, with the values of the 

parameters printed. 

longtable (Section 4.8.4 on page 108) makes tables that extend over 

several pages, with breaks occurring automatically. 

multicol (Section 3.2.8) introduces the environment multicols which 

switches to multi-column output without a page break either before 

or after. It is called with 

\begin{multicols}{num cols}[header text][pre space] 

Text set in num cols columns 

\end{multicols} 

The optional header text is printed in a single column across the top 

of the multi-column part. A \newpage is issued only if the remaining 

space on the current page is less than a certain value, stored in 

the length \premulticols; the optional length argument pre space 

overrides this value if given. Similarly the length \postmulticols 

determines whether a new page is inserted at the end of the multicolumn 

text. Column separation and possible rule are set by the 

standard L AT E X lengths \columnsep and \columnseprule as for the 

\twocolumn command. The text in the multi-column format is 

balanced; that is, the columns are all equally long. 

showkeys prints out all the cross-reference keys defined by \label and 

\bibitem and used by \ref, \pageref, and \cite; these are marginal 

and interlinear notes for a draft only, to check the crossreferencing. 

somedefs allows only selected commands to be defined in a package, 

depending on options in \usepackage. 

tabularx (Section 4.8.4) defines a tabularx environment which is like 

tabular*, a table of desired width, except that it is the column 

widths that expand and not the intercolumn spacing. 

theorem provides extensions to the \newtheorem command for more 

flexible ‘theorem-like’ environments; it is very similar to the A M S 

amsthm package (Section 12.3.1).


varioref (Section 9.2.4) defines \vref and \vpageref commands, analogous 

to \ref and \pageref, which check if the referenced object is 

only one page away, and if so, print text like ‘on the previous page’; 

the actual text printed can alternate between two variants. 

verbatim (Section 4.9.1) is a re-implementation of the verbatim environment 

that prevents memory overflow for long texts; it also defines a 

command \verbatiminput to input a file and to print its contents 

literally, as well as a comment environment for a block comment in 

the input text. 

xr (Section 9.2.3) is a package to allow \ref to cross-reference \label 

commands from other documents. 

xspace (Section 8.3.1 on page 186) contains a device to fix the problem 

of command names swallowing up the blank that follows them; thus 

the command \PS defined as 

\newcommand{\PS}{PostScript\xspace} 

may be used as \PS file without having to terminate the \PS 

command with \ or {}. 

B.6 The various L AT E X files 

A number of different files are used during the L AT E X processing: some 

are read in while others are created to store information for the next run. 

They all consist of two parts: 

root name.extension 

For every L AT E X document, there is a main file, whose root name is used 

when the program L AT E X is called. It normally possesses the extension 

.tex, as do any other files that it might read in with \input or \include 

commands. (If they have a different extension, it must be given explicitly.) 

The other files that are created during a L AT E X processing run normally 

have the same root name as the main file but a variety of different extensions, 

depending on their functions. If the main file contains \include 

commands, there will be additional files created having the same root 

names as the included.tex files but with the extension.aux. 

Some of these files are created with every L AT E X run, whereas others 

appear only when certain L AT E X commands have been issued, such 

as \tableofcontents or \makeindex. The .aux files as well as those 

produced by special L AT E X commands may be suppressed by including 

the command \nofiles in the preamble. This command is useful if a 

document is being constantly corrected and reprocessed, so that the information 

in the special files is not yet finalized and may therefore be

B.6. The various L AT E X files 397 

dispensed with. Files created at the T E X level of processing will always be 

generated. 

Other extensions describe files that contain formatting information, 

or additional instructions (coding), or databases. The root part of these 

file names is not associated with any document file. An example of this 

type is the article.cls file that defines the article class. 

The rest of this section contains a list of the L AT E X file extensions with 

a short description of the role they play in the L AT E X processing. 

.aux This is the auxiliary file written by L AT E X, containing information for 

cross-references as well as some commands necessary for the table 

of contents and other lists. There will be one .aux file created (or 

re-created) for the main file, in addition to one for every file read in 

by an \include command. 

The auxiliary.aux files may be suppressed by issuing \nofiles in 

the preamble. 

.bbl This file is not created by L AT E X, but by the program BIBT E X. It has 

the same root name as the main file. BIBT E X actually only reads the 

.aux file for its information. The.bbl file is read into the next L AT E X 

processing run by the command \bibliography and produces the 

list of literature references. 

.bib Bibliographic databases have the extension.bib. BIBT E X reads them 

to extract the information it needs to generate the .bbl file. The 

root name describes the database and not (necessarily) any text 

file. The database name is included in the text by means of the 

\bibliography command. 

.blg This is the transcript file from a BIBT E X run. The root name is the 

same as that of the input file. 

.bst This is a bibliography style file that is used as input to BIBT E X to 

determine the format of the bibliography. The name of the .bst 

file is included in the text by means of the \bibliographystyle 


.cfg Some classes permit local configurations for paper size or other 

requirements by putting specifications into a .cfg file of the same 

root name as the class. The ltxdoc class (Section D.7.2) is one of 

these. 

.clo This is a class option file, containing the coding for certain options 

that might apply to more than one class. There is no special 

command to input them. 

.cls This is a class file, defining the overall format of the document. It 

is read in by the \documentclass command. A .cls file may also 

be input from another one with the \LoadClass command.


.def Additional definition files of various types carry the .def extension. 

Examples are t1enc.def (to define the T1 encoding) and 

latex209.def (for compatibility mode between L AT E X 2ε and 2.09). 

.dtx These are documented source files for the L AT E X installation. If a.dtx 

file is processed directly, the documentation is printed. This is an 

explanatory manual and/or a detailed description of the coding. The 

class, package,.ltx,.fd or other file(s) are extracted by processing 

the.dtx file with DocStrip (Section D.7.1), usually by running L AT E X 

or T E X on a supplied.ins file. 

.dvi This is the T E X output file, containing the processed text in a format 

that is independent of the output printer, hence a ‘deviceindependent’ 

file. It too is generated by every T E X process run and 

cannot be suppressed by \nofiles. 

The.dvi file must be further processed by a printer driver program 

that converts it into the special commands for the desired output 

device (printer). 

.fd Files containing the NFSS commands associating external font names 

with font attributes (Section A.3.6) are font description files. The 

root names consist of the encoding and family designations, such 

as ot1cmr.fd. When a font of a given family is requested for the 

first time, the corresponding .fd file is automatically read in, if it 

exists. 

.fdd Documented sources for .fd files have the extension .fdd, and are 

nothing more than a special type of.dtx file. They may be processed 

for the documentation, or have the.fd files extracted by DocStrip. 

.fmt This extension belongs to a T E X format file that has been created 

with the special program initex (Section B.1.3). Format files contain 

all the basic instructions in a compact coding for quick loading. 

L AT E X is nothing less than T E X run with the format latex.fmt. (Plain 

T E X uses the format file plain.fmt.) 

.glo This file is only generated when the command \makeglossary 

exists in the preamble. It has the same root name as the main 

file and contains only \glossaryentry commands that have been 

produced by \glossary commands in the input text. This file will 

be suppressed in spite of the presence of \makeglossary if the 

command \nofiles is issued in the preamble. 

.gls This file is the glossary equivalent of the.ind file. It too is generated 

by the MakeIndex program, but with the .glo file as input. The 

input and output extensions must be explicitly given. There are no 

standard L AT E X commands to read in the.gls file. The doc package 

uses it to record a change history.

B.6. The various L AT E X files 399 

.idx This file is only generated when the command \makeindex exists 

in the preamble. It has the same root name as the main file and 

contains only \indexentry commands that have been produced by 

\index commands in the input text. This file will be suppressed in 

spite of the presence of \makeindex if the command \nofiles is 

issued in the preamble. 

.ilg This is the transcript file from a MakeIndex run. The root name is 

the same as that of the input file. 

.ind This file is generated by the MakeIndex program, with the.idx file as 

input. It is read in by the \printindex command which is defined 

by the makeidx package. It contains a theindex environment, built 

up from the entries in the input file. 

.ins To facilitate the DocStrip run on a.dtx file, the necessary commands 

(and options) are put into an installation file. The extraction is 

then accomplished simply by processing (with T E X or L AT E X) the 

appropriate.ins file. 

.ist This is an index style file, containing style settings for MakeIndex. 

The root name reflects the style and has no relationship to any text 

file. For the doc package, the files gind.ist and gglo.ist are 

provided for specialized index formatting. 

.lof This file contains the information for the list of figures. It behaves 

exactly like the .toc file except that it is opened by the command 

\listoffigures instead of by \tableofcontents. 

.log This is the transcript file, containing the protocol of the T E X processing 

run, that is, all the messages that were sent to the monitor 

during the run as well as additional information that can be of use 

to T E X experts when errors occur. This file is always generated, even 

when \nofiles has been issued. 

.lot This file contains the information for the list of tables. It behaves 

exactly like the .toc file except that it is opened by the command 

\listoftables instead of by \tableofcontents. 

.ltx For L AT E X installation, certain files that are inputs to the initex program 

bear the extension .ltx. They form the basis for generating 

the format latex.fmt. The main such file is named latex.ltx. 

.mf METAFONT source files (Sections G.1 and G.3) contain drawing instructions 

for the characters in a font. The METAFONT program 

converts these to bitmap.pk files in a given size and resolution for 

DVI driver programs.


.pk Packed pixel, or bitmap, files (Section G.1) contain images of the 

characters in a font. These are read in by the DVI drivers to print 

the characters either on paper or on the computer monitor. 

.sty This is a package file, to be loaded with \usepackage. A package 

contains additional L AT E X commands to define new features or alter 

existing ones. It should not produce any printable text. It may be 

input by another package with the command \RequirePackage. 

.tex The file containing the user’s input text should have the extension 

.tex. For every L AT E X document there is at least one.tex file. If there 

is only one.tex file, it is also the main file that determines the root 

name of all the others. If there are \input or \include commands 

within the input text, there are other files belonging to the document 

with different root names but all with the extension.tex. The main 

file is the one whose name appears when the program L AT E X is called. 

It contains the highest level \input commands and is the only file 

that may have \include commands. 

.tfm T E X font metric files (Section G.1) contain all the font information 

that T E X needs to process a document: height, width, depth of 

each letter, ligature information, italic correction, and so on. The 

only thing that is missing is what the character looks like. This is 

provided by.pk files. 

.toc This file contains the information for the table of contents. The command 

\tableofcontents reads in the.toc file, if it exists, and then 

outputs the table of contents. At the same time, a new .toc file is 

opened where all the contents information for the current processing 

run is written. This .toc file is closed by the \end{document} 

command, so that if the L AT E X run is interrupted before completion, 

the.toc file is lost. 

A.toc file is only generated if the document contains the command 

\tableofcontents and if the \nofiles command has not been 

issued. It has the same root name as the main file. 

.vf Virtual font files (Section G.1) are read in in place of.pk files for fonts 

that do not really exist independently. For each character, there are 

instructions for the driver program, usually to take character nn 

from real font xx.

C 

Error Messages 

Errors are bound to be made at times during the preparation of a long 

L AT E X document. The mistakes can be of various kinds, from simple typing 

errors for command names to forgetting that some commands must be 

paired or giving an incorrect syntax for a complicated command. 

Errors during the L AT E X processing produce a list of messages on the 

monitor, which appear totally incomprehensible to the beginner. Even the 

advanced user can have difficulty figuring out a particular error message. 

However, these messages contain information about fundamental structures 

that can aid an experienced T E X programmer to see much deeper 

into the heart of the problem. 

In addition, the error messages contain useful information even for 

the beginner. It is the purpose of this chapter to explain some of these 

messages that can be of assistance to non-programmers. 

Error messages are written both to the computer monitor and to the 

processing transcript file, with extension .log. Examine this file if the 

messages on the monitor went by too fast. 

C.1 Basic structure of error messages 

Error messages have two sources: those from L AT E X and those from the 

underlying T E X program. The L AT E X messages are often followed by T E X 

messages since L AT E X operates at a higher level. 

C.1.1 T E X error messages 

We start with a simple error as an example. 



The last words appear in \txetbf{bold face}. 


401

402 Appendix C. Error Messages 

Here the command \textbf has been mistakenly typed as \txetbf. During 

the processing, L AT E X assumes that the user wants to invoke a T E X 

command \txetbf. Since L AT E X does not know the T E X commands, it 

passes this text on to T E X, which then decides that there is no such command 

in its repertoire. The following error message is written to the 

monitor: 

! Undefined control sequence. 

l.3 The last words appear in \txetbf 

{bold face}. 

? 

The program stops at this point and waits for a response from the user. 

This message can be understood even by beginners. It consists of an error 

indicator that starts with an exclamation point !. Here the indicator is: 

! Undefined control sequence, meaning that an unknown command 

name (control sequence) was the cause of the error. Next comes a pair of 

text lines, the first of which is prefixed with l.3, meaning that the error 

occurred in ‘line 3’ of the input text. The error itself was encountered 

at the last symbol printed in this upper line. The lower line shows the 

continuation of the input line being processed when the error was found, 

here the words {bold face}. Before continuing, T E X waits for a reaction 

from the user, as indicated by the question mark in the last line of the 

message. 

Entering another ? and 〈return〉 as response produces the following 

information: 

Type to proceed, S to scroll future error messages, 

R to run without stopping, Q to run quietly, 

I to insert something, E to edit your file, 

1 or ... or 9 to ignore next 1 to 9 tokens of input, 

H for help, X to quit 

? 

This is a list of the possible user responses: 

1. 〈return〉: simply typing the return key tells T E X to continue processing 

after making an attempt to handle the error according to some 

preprogrammed rules. In the case of an unknown command name, 

the error treatment is to ignore it. 

2. S scroll mode: T E X continues the processing, writing further error 

messages to the monitor as they are encountered, but without stopping 

for a user response. This is as though 〈return〉 were pressed 

after all subsequent errors. 

3. R run mode: T E X continues processing as with S, but does not even 

stop as it would in scroll mode if the file named in an \input or 

\include command is missing.

C.1. Basic structure of error messages 403 

4. Q quiet mode: the same as with R except that no further error 

messages are printed to the monitor. They are, however, written to 

the.log file. 

5. I insert: the mistake can be corrected by inserting the proper text. 

T E X puts the line of text entered from the keyboard into the processing 

in place of the error and then continues. Such a correction 

applies only to the processing: the original text in the .tex file is 

unchanged and must be altered with an editor. Typing in I\stop 

brings the program to a halt with the current page in the.dvi file. 

6. 1 ...: entering a number less than 100 will delete that many characters 

and commands from the subsequent text. The program then 

stops to await further response from the user. 

7. H help: an extended account of the problem is printed to the monitor, 

which is more informative than the brief error indicator and which 

may contain tips for relieving the error. 

8. X exit: The T E X processing is halted at this point. The current page 

does not appear in the.dvi file. 

9. E edit: The process is halted as with X and a message is printed 

saying on which line of which file the error occurred. For some 

implementations, the editor may actually be called automatically, 

going directly to the faulty line. 

The above response letters may be typed in either in capitals or in lower 

case. The response does not take effect until after 〈return〉 is pressed. 

Typing H or h for help on the previous sample error message produces 

the text: 

The control sequence at the end of the top line 

of your error message was never \def’ed. If you have 

misspelled it (e.g., ‘\hobx’), type ‘I’ and the correct 

spelling (e.g., ‘I\hbox’). Otherwise just continue 

and I’ll forget about whatever was undefined. 

? 

The error is described in more detail: the command name at the end 

of the upper line is unknown; if this is simply a typing error, enter the 

proper text with the I response, in this case I\textbf. Otherwise, press 

〈return〉 and the faulty command will be ignored. The line of text will 

be processed as though it had been The last word appears in bold 

face. Of course, no bold face can actually appear. 

The basic structure of a T E X error message is summarized as: 

Every error message begins with an error indicator, marked by a ! at the 

beginning of the first line. The indicator text is a brief description of the


problem. Following are one or more lines of input text. The last symbol 

of the first of these lines is the one that has caused T E X to stop and print 

the error message. The last line contains the text or commands that are 

the next to be processed. T E X waits for a response from the user. If 

that response is the typing of an H for help, a more detailed description, 

with possible tips, is printed to the monitor, and T E X waits for a further 

response. 

C.1.2 L AT E X error messages 

As an example of a text with a L AT E X error, we take 



\begin{qoute}\slshape 

Text indented at both ends 

\end{quote} 


Here the call \begin{quote} was incorrectly typed as qoute. The L AT E X 

processing writes the error message: 

! LaTeX Error: Environment qoute undefined. 

See the LaTeX manual or LaTeX Companion for explanation. 

Type H for immediate help. 

... 

l.3 \begin{qoute} 

\slshape 

? 

The first line of this message states that L AT E X itself discovered this error, 

with a brief error indicator, in this case Environment qoute undefined. 

All L AT E X error messages begin with a line like this, followed by a reference 

to the L AT E X manuals for detailed explanation (which is also to be found 

in Section C.3 of this book). The third line of text is a reminder that the 

response H 〈return〉 will also type additional clarification. 

The line of three dots ... indicates that there are more lines of internal 

! coding that have been suppressed. Sometimes it is desirable to look deeper 

behind the error message, especially for practiced T E Xperts who know how to 

interpret them. In this case, L AT E X can be made to output the missing lines of 

coding by setting 

\setcounter{errorcontextlines}{num} 

where num is the number of levels a macro will be decoded on an error. By 

default, num = −1. The knowledgeable user can set it to 99 to get more lines 

than he or she will ever likely to want.


The next pair of lines shows just where the processing was halted 

when the error was detected. As for T E X messages, the current input 

line of text is broken at the error, with the processed part on the first 

line, here \begin{qoute}, and the remainder on the next line. Here this 

remainder is \slshape. The line indicator l.3 (small letter L, not number 

one) shows on which line (3) of the input file the error was found. 

L AT E X now awaits a response from the user. Typing H 〈return〉 yields 

the additional information: 

Your command was ignored. 

Type I to replace it with another command, 

or to continue without it. 

? 

This means the last command in the upper line of the pair beginning with 

l.3 has not been read into the processing. The mistake may thus be 

corrected for the current process run only with I\begin{quote} 〈return〉. 

The misspelling will still be in the text file, though, and needs to be 

removed by an editor run later. 

However, if the processing is continued by simply pressing 〈return〉, 

then \begin{qoute} is ignored, as though this text were not in the 

source file. This leads directly to another error when the command 

\end{quote} is encountered, since now the matching \begin{quote} 

command is missing. The actual text of this second error message is: 

! LaTeX Error: \begin{document} ended by \end{quote}. 



... 

l.5 \end{quote} 

? 

The message again contains the standard L AT E X announcement, with error 

indicator, in the first line, followed by a reference to the manuals, and an 

invitation to type H for help. The error indicator in this case is 

\begin{document} ended by \end{qoute}. 

This arises because when L AT E X encountered \end{quote}, it checked the 

name of the current environment. Since \begin{quote} was missing, 

the matching \begin command is the \begin{document} statement, 

producing a mismatched \begin...\end pair. 

The last pair of lines again shows just how far the processing had 

occurred before the error was detected. In this case, the whole of the 

current line has already been taken in, so the remainder line is blank.


The response H 〈return〉 at this point leads to the same message as 

for the first error. A correction with an I entry will not do any good since 

the missing \begin{quote} can no longer be inserted ahead of the environment 

text. Entering I\begin{quote} will replace the \end{quote} 

with \begin{quote}, but that does not solve the real problem. The best 

response now is to type 〈return〉 so that the command \end{quote} is 

ignored and the processing continues. 

Now both the faulty \begin{qoute} and correct \end{quote} commands 

are removed and the processing takes place as if the quote environment 

had never been in the input text at this point. 

If the typing error qoute had been made in the \end instead of in the 

\begin command, the error message would have been: 

! LaTeX Error: \begin{quote} on input line 3 ended by \end{qoute} 



... 

l.5 \end{qoute} 

? 

The previous explanations should be enough for the user to understand 

what is now being said. The error indicator here is 

\begin{quote} on input line 3 ended by \end{qoute} 

and the last pair of lines indicate that the problem is in line 5 and that the 

troublesome command is \end{qoute}. The obvious thing to do now, and 

the H message encourages this conclusion, is to make a correction with 

I\end{quote}. However, now a new message appears on the monitor: 

! Extra \endgroup. 

\endgroup 

l.5 \end{qoute} 

? 

With the exception of the last pair beginning with l.5, this makes no 

sense at all. The error indicator ! Extra \endgroup is seemingly meaningless. 

That this is a T E X error message and not a L AT E X one is hardly any 

compensation. 

The frustrated user is not really to blame. The response was perfectly 

reasonable even if it was wrong. Only experience could say that the 

best action at that point was to have just pressed 〈return〉. The quote 

environment is closed off anyway, although any special actions associated 

with \end{quote} would be left off. 

The help message at this point is


Things are pretty mixed up, but I think the worst is over. 

This is at least encouraging, and the user should not lose heart. The best 

thing is to just keep on pressing 〈return〉 to get through the run. 

We give here two recommendations for choosing a response, one specific 

and one general. The specific one is: 

If a false environment name appears in a \begin command, the proper 

correction of the error is 

I\begin{right name} 

If the misspelling has occurred in the \end command, the best way 

to handle this mistake is just to press 〈return〉. The environment will 

be closed and any local declarations or definitions will be terminated. 

However, if there are any commands to be executed or text to be printed 

by the \end command, these will be missing. 

The general recommendation is 

If the user knows how to correct the error with the help of the error 

message, this should be done with 

I correction 

Otherwise, one may give 〈return〉 and wait and see what happens. Even 

if more peculiar (T E X) error messages appear, one can keep on pressing 

〈return〉 until the processing is finally at an end. The following printout 

should indicate where the mistake was. 

Instead of continually pressing 〈return〉, one may also enter S, R, or Q 

and 〈return〉 to speed up the error treatment (Section C.1.1). Here, as 

with the simple 〈return〉 entry, the faulty commands are not just ignored. 

T E X attempts to treat them by making assumptions about what the user 

wanted to do at this spot. Only when this is no longer possible will T E X 

ignore the command completely. For example, if the indicator reads 

\begin{environment} ended by \end{environment} 

there was at least a \begin command with a valid environment name 

earlier. It could be assumed that the name of the environment in the \end 

command is wrong. L AT E X thus tries to execute this command using the 

name of the current environment. 

C.1.3 Error messages from T E X macros 

The majority of T E X commands and practically all the L AT E X commands 

are so-called T E X macros. These are combinations of primitive commands 

grouped together under a new command name, by which they may be


called as a unit. T E X macros are structures similar to those that may 

be defined by the L AT E X command \newcommand. Up to nine arguments 

may be passed to them, just as for L AT E X commands. However, the corresponding 

T E X commands for generating macros are more general than 

\newcommand. 

In fact, of the approximately 900 Plain T E X commands available, only 

300 are primitives, or fundamental commands. The remaining 600 are 

macros. If an error occurs within a macro, the other commands within it 

may also be affected. 

Here is an example for clarification. The command \centerline is a 

macro defined as 

\def\centerline#1{\@@line{\hss#1\hss}} 

in which \@@line is itself a macro while \hss is a T E X primitive, a rubber 

length that can be infinitely stretched or shrunk. To avoid leading the 

user into the murky depths of T E X commands, we will simply point out 

that the above macro definition is more or less equivalent to the L AT E X 

command sequence 

\newcommand{\centerline}[1]{\makebox[\textwidth][c]{#1}} 

Now take the following sample input text 



\centerline{This is an \invalid command} 


in which a \ has been placed before the word invalid, making a false 

command named \invalid. During the L AT E X processing, the following 

T E X error message appears: 


This is an \invalid 


l.3 \centerline{This is an \invalid command} 

? 

This T E X message is now a bit more understandable. The error indicator 

is the same as for the example in Section C.1.1: 


The following pair of lines claim that the error was recognized after the 

‘command’ \invalid was processed, and that the next piece of text to be 

read in is the word command. At the same time, the at the 

beginning of the upper line says that this text is the argument of some 

other command. 

The next pair of lines are familiar: the error occurred in line 3 of the 

input text, after the entire line, command and argument, had been read 

in for processing.

C.2 Some sample errors 

C.2.1 Error propagation 

C.2. Some sample errors 409 

The example with the incorrect \begin{qoute} environment showed that 

a simple response with the 〈return〉 key led to a second error message 

in spite of the fact that the \end{quote} command that produced it was 

correct. That one wrongly corrected error leads to further errors is more 

often the rule than the exception. 

Let us look at the input text 



\begin{itemie} 

\item This is the first point in the list 

\item And here comes the second 

\end{itemize} 


The only mistake in this text is the misspelled environment name itemie 

in place of itemize. The L AT E X processing first produces the same error 

message as for the incorrect quote environment earlier: 

! LaTeX Error: Environment itemie undefined. 



... 

l.3 \begin{itemie} 

? 

At this point, entering 〈return〉 leads to a new error messages: 

! LaTeX Error: Lonely \item--perhaps a missing list environment. 



... 

l.4 \item T 

his is the first point in the list 

? 

What has happened here is that without the \begin command, the \item 

has been issued outside of a list environment, where it is rather meaningless. 

(Actually it is defined to print the above error message!) It is now 

too late to try to insert the missing start of the itemize environment. 

Typing H 〈return〉 for help yields


Try typing to proceed. 

If that doesn’t work, type X to quit. 

? 

Following this advice and pressing 〈return〉, one obtains the same error 

message once again, but this time on line 5, for the second \item 

command. Persevering further and pressing 〈return〉 once again, one is 

presented with 

! LaTeX Error: \begin{document} ended by \end{itemize}. 



... 

l.6 \end{itemize} 

? 

The itemize environment has now come to its end, but since it was 

never properly started, L AT E X complains about a mismatch of \begin 

and \end commands. Now at last, typing another 〈return〉 leads to a 

proper continuation of the processing. Of course the itemized text will be 

incorrectly formatted, but the rest of the document will not be affected. 

In this example, a single mistake in the input text generated three 

further error messages. This is by no means unusual. Some L AT E X errors 

can produce hundreds of such successive errors. It is even possible that 

the chain of errors never ceases and that the processing never advances. 

In this case, there is nothing else to do but to terminate the program. 

This should be done with the entry I\stop after the next error message. 

It may be necessary to give it several times before it takes effect. If this 

does not work, that is, if the same message appears every time, then the 

response X 〈return〉 will halt the program immediately. 

It is better to stop the program with I\stop than with X, since then 

the output will include the last page being processed. This can be of 

assistance in trying to deduce what the source of the error was. 

The final lesson of this section is simply: even when faced with a host 

of error messages, don’t panic! Continue to press the 〈return〉 key to 

advance the processing. 

Pressing S 〈return〉 instead will produce the same set of error messages 

on the monitor but without waiting for a user response in between 

(Section C.1.1). 

C.2.2 Typical fatal errors 

Occasionally it may happen that a user forgets to include one of the commands 

\documentclass or \begin{document}, or the entire preamble.


The last can easily occur when a L AT E X file is to be read in with \input or 

\include, but is mistakenly entered directly with the L AT E X program call. 

If a file with the text 

This file has no preamble. 

is put directly into the L AT E X processing, the following message appears 

on the monitor: 

! LaTeX Error: Missing \begin{document}. 



... 

l.1 T 

his file has no preamble. 

? 

with the accompanying help message after typing H 〈return〉: 

You’re in trouble here. Try typing to proceed. 

If that doesn’t work, type X to quit. 

? 

One sees from the error message that L AT E X has discovered a mistake 

on reading the very first letter of the text. The help message is also not 

very encouraging. There is no point trying to continue the processing 

with 〈return〉. Rather one should stop it with X or E right away, since 

nothing useful can be achieved here. 

Even when the above example file contains the environment 


This file has no preamble. 


it is not possible to carry out a proper processing. The T E X error message 

now printed is 

! LaTeX Error: The font size command \normalsize is not defined: 

there is probably something wrong with the class file. 



... 

l.1 \begin{document} 

?


The error indicator is rather peculiar since the command \normalsize 

never appears in the input. Only the last two lines of this message are 

easily understood. They state that the error was recognized on line 1 

after \begin{document} was read in. The fact that this is the text of the 

first line is enough to show that no meaningful processing can take place, 

since the mandatory \documentclass command that must precede it is 

missing. 

Here again there is no choice but to halt the program with X or E, since 

a continuation of the processing would be meaningless. 

(The reason for the strange error indicator is that many formatting 

parameters are initiated by the \begin{document} statement, including 

the standard font. It is here that \normalfont is called internally, and 

that command must be defined in the class file, which is missing. Other 

initiating commands are defined in the L AT E X format itself, and so they do 

not cause this error.) 

If the name of the document class is wrongly typed, say as 

\documentclass{fred} 

the following message appears: 

! LaTeX Error: File ‘fred.cls’ not found. 

Type X to quit or to proceed, 

or enter new name. (Default extension: cls) 

Enter file name: 

This message should be fairly clear: the program is looking for a file with 

the name fred.cls but cannot find it, so would the user be so kind as to 

enter an alternative name. If the new name has the same extension as that 

requested (.cls), it need not be explicitly given. In this case, the name of 

any standard L AT E X class, such as article, report, book, or letter, may 

be typed in, as well as that of any additional classes that you might have 

access to. Of course, it should be the class for which the document was 

written. 

This same error message is printed whenever a file is to be read in that 

cannot be found on the system. The issuing command may be \input, 

\include, or \usepackage. If the file sought really does exist, perhaps 

it is not located where T E X is looking for files. Usually there is a system 

parameter set by the installation to point T E X to its files. If your file is 

somewhere else, you can trying entering its full name, with directory or 

path designations. 

If L AT E X is asking for a file that you know does not exist, or which is 

totally unknown to you, you will want to halt the processing right away 

or tell L AT E X to skip the silly file. At this point, you have a problem. 

L AT E X insists on a valid file name. Typing X 〈return〉 only produces the


message once again, this time stating that it cannot find the file x.ext. 

Some installations provide a dummy file named null.tex, while one of 

the extension packages available (Section B.5.4) offers a set of files named 

after the standard response letters x.tex, e.tex, r.tex, h.tex, s.tex 

to emulate the responses to normal error messages. 

Emergency stop: sometimes one reaches a situation where the program 

cannot be halted after an error, even with I\stop or X. In such a case, one 

has to apply the operating system’s program interrupt, usually CTRL Z. 

C.2.3 Mathematical errors 

Surprisingly few errors occur as a result of incorrect application of mathematical 

formula commands themselves, even for a user with only a little 

experience. More often, the errors in mathematical formulas are fiddly 

ones such as forgetting a closing brace } or neglecting to switch back 

to text mode. Another type of common error is to use a symbol in text 

mode that may only appear in math mode. We will point out a few typical 

examples here. 

One wants to produce the text: ‘The price is $3.50 and the order 

number is type sample’, and types the input text: 

The price is $3.50 and the order number is type_sample. 

This text contains two errors, such that the first cancels the second one: 

the $ sign is the math mode switch for generating formulas within text 

(Section 5.1). The proper way to write a true dollar sign in text is as \$. 

Instead, in the sample text, the $ alone switches to math mode, setting 

everything that follows as a formula. However, the closing switch-back $ 

is missing, something that T E X will first notice at the end of the current 

paragraph. 

! Missing $ inserted. 

 

$ 

l.4 

? 

If 〈return〉 is simply pressed as a response, T E X inserts a $ at this point. 

For our sample text, this would be at the end of the text before the blank 

line. This means that the text from the first $ to the end of the paragraph 

will be set as a text formula: 

The price is 3.50andtheor der number istypesample. 

This output should show the user right away what went wrong. Near 

the end of the line, there is a letter s printed as an index. This is the 

second error in the example. The underbar character _ is only permitted


in math mode and should have been typed here as \_. However, after 

the first $ had switched (incorrectly) into math mode, the _ sign became 

allowed, with the result that the following letter s was lowered. 

If we now correct the text by replacing the $ sign with \$, then the _ 

symbol is no longer in math mode, producing the error message: 

! Missing $ inserted. 

 

$ 

l.5 ...ice is \$3.50 and the order number is type_ 

sample. 

? 

Pressing 〈return〉 here tells T E X to recover from the error by inserting 

the apparently missing $ sign at this point before the math command 

_. The processing continues, and at some place before the end of the 

current paragraph, T E X will notice again that the closing $ is not present. 

The same error message is printed as in the first case. The processing 

goes on with another 〈return〉 response, with the result: 

The price is $3.50 and the order number is typesample. 

In all three cases, asking for more help with H would type to the monitor: 

I’ve inserted a begin-math/end-math symbol since I think 

you left one out. Proceed, with fingers crossed. 

The last sentence above is just what we recommend for mathematical 

error messages: keep on pressing 〈return〉 or S to get through to the end 

of the processing, and then look at the printed output to find the mistake. 

C.2.4 Errors from multi-file texts 

If the document text is split over many files that are to be read in with 

\input or \include commands, the line number in the error message 

refers to the file currently being read. A response with E 〈return〉 will call 

the editor program and open that file, going to the indicated line where 

the error was recognized. With the other responses, the editor has to be 

called separately, with the faulty file explicitly named. 

It is possible to determine which file was being processed at the time 

of the error by examining the processing messages or the.log file. As the 

files are opened for reading, T E X writes an opening parenthesis ( and the 

name of the file to the monitor and to the transcript.log file. When the 

file is closed, a closing parenthesis ) is printed. Output page numbers 

are similarly written to both the monitor and transcript file in square 

brackets. For example, if the monitor shows the processing messages

C.3. List of L AT E X error messages 415 

..(myfile.tex [1] [2] [3] (part1.tex [4] [5]) (part2.tex [6] [7] 

! Undefined control sequence 

l.999 \finish 

? 

this can be interpreted as follows: a file myfile.tex was being read, and 

after pages 1, 2, and 3 had been output, the file part1.tex was read in with 

an \input or \include command inside the file myfile.tex. Pages 4 

and 5 were output and part1.tex was closed, after which another file 

part2.tex was opened for input. An error was discovered in line 999 of 

this file. If this error is corrected from the keyboard, or if the processing 

is otherwise continued, the monitor messages proceed as 

[8] [9]) [10] 

! Too many }’s 

l.217 \em sample} 

The closing parenthesis after page 9 indicates that the file part2.tex has 

been closed. The next error is on page 10 in the main file myfile.tex, 

since there is no closing ) for it. The error was discovered on line 217 of 

this file. 

C.3 List of L AT E X error messages 

The L AT E X error messages are listed here, divided into general, package, or 

font error messages. Within each group, they are ordered alphabetically 

according to the error indicators. A description of the possible causes 

and solutions is also given. 

C.3.1 General L AT E X error messages 

The following error messages are those not involving font selection or 

definition, nor any of the special features for L AT E X programming class 

and package files (Appendix D). 

! LaTeX Error: ... undefined. 

The argument of a \renewcommand or \renewenvironment has not been 

previously defined. The corresponding \new... command should be 

used instead. 

! LaTeX Error: \< in mid line. 

The command \< in a tabbing environment occurred in the middle of 

a line. This command may only appear at the beginning of a line (Section 

4.6.3).


! LaTeX Error: Bad \line or \vector argument. 

The first argument of a \line or \vector command specifies the angle 

of the line or arrow. This message states that the selected angle entries 

are invalid. See pages 293–294. 

! LaTeX Error: Bad math environment delimiter. 

L AT E X has encountered a math switch command in the wrong mode: either 

a \[ or $ command in math mode or the corresponding \] or $ in 

normal text mode. Either the math switches have been improperly paired 

or some braces {...} are incorrect. 

! LaTeX Error: \begin{...} on input line ... ended by \end{...}. 

L AT E X has encountered an \end command without a corresponding \begin 

of the same name. This may be due to a typing mistake in the name of the 

environment, or to the omission of a previous \end command. A good 

way to avoid this error is always to enter the \end command immediately 

after the \begin, inserting the actual environment text with the editor in 

between the two. This is especially useful for long, nested environments. 

It also reduces the risk of typing the environment name incorrectly in the 

\end command. 

! LaTeX Error: Can be used only in preamble. 

Many commands may only be called within the preamble. These include 

\documentclass, \usepackage, \nofiles, \includeonly, \makeindex, 

\makeglossary, and several others. Certain commands that only have 

meaning within class or package files, such as \ProvidesClass and 

\ProvidesPackage (Section D.2.1), as well as many \Declare.. and 

\Set.. commands are also only allowed in the preamble. If one of these 

commands is issued after \begin{document}, this message is printed. 

! LaTeX Error: Command ... invalid in math mode. 

A command has been issued in math mode that only makes sense in 

text mode, such as \item or \circle. The font declarations \itshape, 

\bfseries, and so on, also produce this error in math mode, since the 

math alphabet commands \mathit, \mathbf should be used instead. 

! LaTeX Error: Command ... already defined. 

The user has tried to redefine an existing structure with one 

of \newenvironment, \newcommand, \newtheorem, \newsavebox, 

\newfont, \newlength, \newcounter, or \DeclareMathAlphabet. 

Either a different name must be selected or, in the case of commands 

and environments, the \renew... version must be employed. (Note that 

when an environment named sample is defined, the commands \sample 

and \endsample are also created.)


! LaTeX Error: Command ... undefined in encoding .... 

The specified command has been defined with \DeclareTextCommand 

for a certain NFSS encoding (say OT1) but was executed while another 

encoding (for example T1) for which there is no definition was active. 

! LaTeX Error: Counter too large. 

A counter that is to be printed as a letter contains a value greater than 26. 

! LaTeX Error: Environment ... undefined. 

L AT E X has encountered a \begin command with an unknown environment 

name. This is probably due to a typing mistake. It may be corrected 

during the processing with the response I followed by the correct name. 

(This does not alter the source file, where the error remains.) 

! LaTeX Error: File ‘...’ not found 

Type X to quit or to proceed, 

or enter new name. (Default extension: ...) 

Enter file name:. 

A file is to be loaded with one of the inputting commands, but it cannot 

be found. You may type a new file name, or quit, or proceed without it. 

If the file name has been given correctly, and it does exist, maybe it is 

not located where L AT E X looks for files. Make sure that it is in one of the 

right directories. The user is offered a chance to type in an alternative, 

or correctly typed, name via the keyboard. Type in the file name, with 

optional extension, and press 〈return〉. L AT E X proposes a default extension 

that is the same as that requested. There is no default for the main part 

of the file name. 

! LaTeX Error: Float(s) lost. 

A figure or table environment or a \marginpar command was given 

within a vertical box (\parbox or minipage environment), or these structures 

were issued within a L AT E X command that uses vertical boxes internally, 

such as a footnote. This error is first recognized by L AT E X when a 

page is output so that the actual cause could be many lines earlier in the 

text. A number of tables, figures, or marginal notes may have been lost 

as a result, but certainly not the one that triggered the problem. 

! LaTeX Error: Illegal character in array arg. 

A tabular or array environment contains an unknown column formatting 

character (see Section 4.8.1) or the formatting entry in the second 

argument of a \multicolumn command is wrong. 

! LaTeX Error: \include cannot be nested. 

An attempt has been made to call \include from a file that has already


been read in by an \include command. All \include commands must 

be issued in the main file (Section 9.1.2). 

! LaTeX Error: LaTeX2e command ... in LaTeX 2.09 document. 

This error occurs if a L AT E X 2ε command is used in compatibility mode 

(Appendix F), that is, if the document uses \documentstyle instead of 

\documentclass. These can be \LaTeXe, \usepackage, \ensuremath, 

the lrbox environment, as well as the new syntax for adding an optional 

argument with \newcommand and \newenvironment. 

! LaTeX Error: Lonely \item--perhaps a missing list environment. 

An \item command has been given outside of a list environment (Section 

4.3). Either the environment name has been misspelled in the \begin 

command and not corrected by keyboard entry, or the \begin command 

has been forgotten. 

! LaTeX Error: Missing @-exp in array arg. 

The column formatting argument of a tabular or array environment 

contains an @ symbol without the necessary following text in curly braces 

{ } that must go with it (Section 4.8.1 for @-expressions), or the same 

thing occurs in the second argument of a \multicolumn command. 

! LaTeX Error: Missing begin{document}. 

Either the \begin{document} command has been forgotten or there is 

printable text inside the preamble of the document. In the latter case, 

there could be a declaration with incorrect syntax, such as a command 

argument without curly braces { } or a command name without the 

backslash \ character. 

! LaTeX Error: Missing p-arg in array arg. 

The column formatting argument of a tabular or array environment 

contains a p symbol without the necessary width specification that must 

go with it (Section 4.8.1), or the same thing occurs in the second argument 

of a \multicolumn command. 

! LaTeX Error: No counter ‘...’ defined. 

A \setcounter or \addtocounter command was called that referred 

to a counter name that does not exist. Most likely the name was typed 

incorrectly. If this error occurs while an .aux file is being read and if 

the name of the counter is indeed correct, the defining \newcounter 

command was probably given outside of the preamble. Therefore it is 

highly recommended always to give the \newcounter commands inside 

the preamble. (If the other L AT E X counter commands are used with an 

undefined counter name, a long list of funny T E X error messages appears.)

! LaTeX Error: No \title given. 


The \maketitle command has been given before \title has been declared. 

! LaTeX Error: Not in outer par mode. 

A figure or table environment or a \marginpar command was given 

within math mode or inside a vertical box (\parbox or minipage environment). 

In the first case, the math switch-back command was probably 

forgotten. 

! LaTeX Error: Page height already too large. 

The command \enlargethispage is trying to extend the vertical size of 

the page, which L AT E X already considers too large. 

! LaTeX Error: \pushtabs and \poptabs don’t match. 

The number of \poptabs commands in a tabbing environment does not 

agree with the number of previous \pushtabs given (Section 4.6.4). 

! LaTeX Error: Something’s wrong--perhaps a missing \item. 

The most likely cause of this error is that the text in a list environment 

(list, itemize, enumerate, or description) does not begin with an 

\item command. It will also occur if thebibliography environment is 

given without the argument {sample label} (Section 9.3). 

! LaTeX Error: Suggested extra height (...) dangerously large. 

The command \enlargethispage is trying to extend the vertical size of 

the page more than L AT E X considers reasonable. 

! LaTeX Error: Tab overflow. 

The last \= command in a tabbing environment exceeded the maximum 

number of tab stops allowed by L AT E X. 

! LaTeX Error: There’s no line here to end. 

The command \newline or \\ has been issued after a \par or blank line 

where it makes no sense. If additional vertical space is to be inserted 

here, this should be done with a \vspace command. 

! LaTeX Error: This may be a LaTeX bug. 

This message states that L AT E X is fully confused. This could be the result 

of a previous error after which the user response was to press 〈return〉. In 

this case, the processing should be brought to a halt with I\stop or X or E 

and the earlier error corrected. It is also possible, although unlikely, that 

there is a bug in the L AT E X program itself. If this is the first error message


in the processing, and the text seems otherwise to be satisfactory, the 

file should be saved and submitted to the computing center for further 

investigation. 

! LaTeX Error: Too deeply nested. 

Too many list environments (description, itemize, enumerate, or list) 

have been nested inside one another. The maximum depth of such nesting 

is dependent on the installation but should always be at least four. 

! LaTeX Error: Too many columns in eqnarray environment. 

The eqnarray environment may only have three columns per line. You 

may have forgotten to start a new row with \\, or have placed an extra & 

in the line. 

! LaTeX Error: Too many unprocessed floats. 

This error may occur if there are too many \marginpar commands on one 

page. However, it is more likely that L AT E X is retaining more figure and 

table floats than it can hold. This happens if too many such float objects 

have been given before they can be output (Chapter 7). In this case, the 

last figure or table should be added later in the text. Another cause could 

be that the figure or table cannot be located on a normal text page, but 

rather on a special float page at the end of the text or after a \clearpage 

or \cleardoublepage command. Since the output sequence of figures 

and tables will be the same as that with which they were input, one such 

float can block the entire queue. A \clearpage or \cleardoublepage 

command can free the blockage. 

! LaTeX Error: Undefined tab position. 

A \>, \+, \-, or \< command in a tabbing environment has tried to move 

to a tabulator stop that does not exist (Section 4.6). 

! LaTeX Error: \verb ended by end of line. 

The text of an in-line verbatim command \verb+...+ extends over more 

than one line of text. This is forbidden in order to catch a common error: 

the missing terminating character. Make sure that the entire text between 

the initial and terminating characters is all on one input line. 

! LaTeX Error: \verb illegal in command argument. 

The \verb command may not be used in the argument of any other 

command, except \index and \glossary. It may not be used within 

section titles or footnotes, for example.

C.3.2 L AT E X package errors 


The special programming commands for handling class and package files 

(Appendix D) have their own set of error messages. If one of these features 

issues a serious error, there is often little the user can do about it other 

than report the problem to the author of the file. On the other hand, 

some errors are due to the improper use of the file or of the options that 

it provides. 

Often classes and packages contain their own error or warning messages 

with text and meanings that are peculiar to them. These are indicated 

as such, along with the name of the class or package that issued 

them. For example, the package mypack could print the error 

Package mypack Error: cannot mix options ‘good’ 

(mypack) and ‘bad’. 

A help message should also be available when H is typed. Obviously, 

such error (and warning) messages cannot be explained here, since they 

depend entirely on the package in question. 

! LaTeX Error: \LoadClass in package file. 

A package file has called \LoadClass, which it is not allowed to do. A 

class file can only be loaded from another class file. 

! LaTeX Error: Option clash for package .... 

The specified package has been requested a second time with a different 

set of options. A package file will only be loaded once and a second 

attempt will be ignored. Thus, if two \usepackage or \RequirePackage 

commands load the same package with different options, there is a conflict. 

Try to arrange for a consistent set. Typing H for help after this 

message will print out the two sets of options. 

! LaTeX Error: \RequirePackage or \LoadClass in Options Section. 

These two commands may not appear in the definition of a class or 

package option made with the \DeclareOption command. Instead, the 

option should set some flag or other indicator that is later tested before 

calling the command in question. 

! LaTeX Error: This file needs format ‘...’ but this is ‘...’. 

The \NeedsTeXFormat command specifies a different format from the 

one being used. The format is the prestored set of instructions that 

determine what type of T E X is being run. For L AT E X 2ε, the format name 

specified must be LaTeX2e. This file cannot be processed at all with the 

current format in use.


! LaTeX Error: Two \documentclass or \documentstyle commands. 

A document may contain only one \documentclass or \documentstyle 

command. If you can only see one in the main document file, check that 

other files being loaded do not contain an offending second one. 

! LaTeX Error: Two \LoadClass commands. 

The class file contains more than one \LoadClass command, something 

that is not allowed. The class file has been improperly written. 

! LaTeX Error: Unknown option ‘...’ for package ‘...’. 

An option has been specified with \usepackage but it is not defined 

for that package. Check the instructions for the package or look for a 

misprint. 

! LaTeX Error: \usepackage before \documentclass. 

The \usepackage command may not appear before \documentclass. A 

class file must be loaded before any packages. 

C.3.3 L AT E X font errors 

The following error messages occur when defining or selecting fonts with 

the New Font Selection Scheme (Appendix A). Some of these messages 

indicate that the fonts have not been properly set up, in which case the 

font description files may be corrupted or contain mistakes. In either 

case, the system manager must see that a proper set of files are installed. 

! LaTeX Error: ... allowed only in math mode. 

A math alphabet command such as \mathbf has been used in text mode. 

Perhaps a $ has been forgotten. 

! LaTeX Error: Command ‘...’ not defined as a math alphabet. 

The name of a non-existent math alphabet has been specified in one of 

the \Set.. or \Declare.. commands that take a math alphabet name as 

argument. The math alphabet name must be declared with the command 

\DeclareMathVersion before it may be used. 

! LaTeX Error: Command ... not provided in base LaTeX2e. 

A number of symbols that were part of the basic L AT E X 2.09, but not in T E X, 

are no longer automatically included in L AT E X 2ε. Add one of the packages 

latexsym or amsfonts to include them. 

! LaTeX Error: \DeclareTextComposite used on inappropriate 

command .... 

\DeclareTextComposite (Section A.3.7) is used to redefine an existing


accenting command to print a single character for a certain encoding. If 

the accenting command does not already exist, this message is issued. 

! LaTeX Error: Encoding scheme ‘...’ unknown. 

A font declaration or selection command that refers to a non-existent 

encoding scheme has been issued. There is probably a typing error 

involved here. 

! LaTeX Error: Font family ‘...+...’ unknown. 

A \DeclareFontShape command has been issued for a font encoding 

and family combination that has not been previously declared by means 

of a \DeclareFontFamily command (Section A.3.6). 

! LaTeX Error: Font ... not found. 

The font with the specified attributes could not be found, nor could an 

adequate substitution be made. The font defined by \DeclareErrorFont 

is used instead. 

! LaTeX Error: Math alphabet identifier ... is undefined 

in math version ‘...’. 

A math alphabet (Section A.3.3) that has not been defined for the current 

math version has been invoked. This means that the alphabet has been 

created by \DeclareMathAlphabet with an empty shape attribute, but 

no \SetMathAlphabet declaration has been given to define it for the 

selected math version. 

! LaTeX Error: Math version ‘...’ is not defined. 

The name of a non-existent math version has been specified in one of the 

\Set... or \Declare... commands that take a math version name as argument. 

The version name must be declared with \DeclareMathVersion 

before it may be used. 

! LaTeX Error: Not a command name: ‘...’. 

The first argument of \DeclareMathAccent must be a command name, 

like \acute. Most likely the backslash \ character has been forgotten. 

! LaTeX Error: Symbol font ‘...’ not defined. 

The name of a non-existent symbol font has been specified in one of 

the \Set.. or \Declare.. commands that take a symbol font name as 

argument. The font name must be declared with \DeclareSymbolFont 

before it may be used as a symbol font.


! LaTeX Error: The font size command \normalsize is not defined: 

there is probably something wrong with class file. 

It is necessary for class files to define \normalsize, the basic size of standard 

text in the document. If the class file does not do this, if it only 

defines \@normalsize, then it must be repaired. This message also appears 

if \documentclass is missing, for an empty class file is clearly a 

faulty one. 

! LaTeX Error: This NFSS system isn’t set up properly. 

Something is wrong with the font descriptions in the.fd files, or there is 

no valid font declared by \DeclareErrorFont. This is a serious problem, 

to be reported to the system manager. 

! LaTeX Error: Too many math alphabets used in version .... 

There is a limit of 16 math alphabets possible, a number set by T E X itself. 

Any additional math alphabet definitions are ignored. 

! LaTeX Error: Unknown symbol font ‘...’. 

The name of a non-existent symbol font has been used as argument to 

the command \DeclareSymbolFontAlphabet. The font name must first 

be declared with \DeclareSymbolFont before it may be used. 

C.4 T E X error messages 

This section contains a list of some of the most common T E X error messages, 

ordered alphabetically according to the error indicator, together 

with a brief description of the possible cause. Each one begins with an 

exclamation point only. 

! Counter too large. 

As a T E X error, this message refers to a footnote marked either with letters 

or symbols in which the counter has exceeded a value of 26 or 9. It may 

also occur if there are too many \thanks commands on a title page. 

! Double subscript. 

A mathematical formula contains two subscripts for the same variable, 

for example, x_2_3 or x_{2}_{3}. The proper way to produce x23 is with 

x_{2_3} or x_{2_{3}} (Section 5.2.2). 

! Double superscript. 

A mathematical formula contains two superscripts for the same variable, 

for example, xˆ2ˆ3 or xˆ{2}ˆ{3}. The proper way to produce x23 is with 

xˆ{2ˆ3} or xˆ{2ˆ{3}} (Section 5.2.2).

! Extra alignment tab has been changed to \cr. 

C.4. T E X error messages 425 

A line in a tabular or array environment contains more & commands 

than there are columns defined. The error is probably due to a forgotten 

\\ at the end of the previous line. 

! Extra }, or forgotten . 

In math mode, either an opening brace { has been left off or an extra 

closing brace } has been included by mistake. Another possibility is that 

a math mode switch command such as $, \[, or \( has been forgotten. 

! Font ... not loaded: Not enough room left. 

The text processing requires more character fonts to be loaded than T E X 

can handle owing to memory limitations. If certain parts of the document 

need different fonts, you can try to split it up and to process the parts 

separately. 

! Illegal parameter number in definition of .... 

This error message is probably due to a \newcommand, \newenvironment, 

\renewcommand, or \renewenvironment command in which the substitution 

character # has been applied incorrectly. This character may appear 

within the defining text only in the form #n, where n is a number between 

1 and the number of arguments specified in the command. Otherwise 

the character # may only appear in the definition as \#. This error may 

also arise if the substitution character is applied within the last argument 

{end def } (Section 8.4). 

! Illegal unit of measure (pt inserted). 

If this error message comes right after another error with the message 

! Missing number, treated as zero. 

the problem lies with this previous error (see below). Otherwise, the 

mistake is that T E X expects a length specification at this point but has 

only been given a number without a length unit. This occurs most often 

when a length is to be set to zero and 0 has been typed in instead of 0mm or 

0pt. If this is the case, then responding with 〈return〉 produces the right 

result since for a zero value any unit specification is all right. This error 

may also occur if a length specification has been completely forgotten. 

! Misplaced alignment tab character &. 

The single character command & has been given in normal text outside of 

the tabular and array environments. Possibly the intention was to print 

&, in which case \& should be typed. This may still be achieved during 

the processing by responding with I\& to the error message.


! Missing control sequence inserted. 

This error is most likely caused by a \newcommand, \renewcommand, 

\newlength, or \newsavebox command in which the backslash \ is missing 

from the first argument. Pressing 〈return〉 as response will complete 

the processing correctly since T E X assumes a backslash is missing and 

inserts it. 

! Missing number, treated as zero. 

This error is most likely due to a L AT E X command that expects a number or 

length as argument, which is missing. It may also occur when a command 

that takes an optional argument is followed by text beginning with [. 

Finally another cause can be a \protect command preceding a length or 

\value command. 

! Missing { inserted. 

! Missing } inserted. 

T E X is totally confused when either of these error indicators is written. 

The line number printed will probably not be where the source of the 

error is to be found, which is a missing opening or closing brace. If the 

error is not obvious, continue the processing by pressing 〈return〉 and try 

to deduce where it might be from the printed output. 

! Missing inserted. 

Most likely a symbol or command that may only appear in math mode 

was used in normal text. Recall that all those commands described in 

Chapter 5 are only allowed in math mode unless otherwise stated. If 

an \mbox command is inserted within a math formula, its argument has 

temporarily exited from math to normal text mode. The message may 

also occur if a blank line appears within a math formula, signalling a new 

paragraph and the end of the formula without the necessary $ sign. 

! Not a letter. 

The word list in a \hyphenation command contains a character that is 

not recognized as a letter, such as an accent command like \’{e}. Such 

words can only be hyphenated by explicitly inserting the hyphenation 

possibilities (Sections 2.8.1 and 2.8.2). 

! Paragraph ended before ... was complete. 

The argument of a command contains a blank line or a \par command, 

something that is not allowed. Probably a closing brace } has been 

omitted. 

! TeX capacity exceeded, sorry [...]. 

T E X sets up various storage buffers in the computer memory to carry out

C.4. T E X error messages 427 

the text processing. This message appears when one of these buffers is 

full and can no longer be used. The name of the buffer and its maximum 

size are printed in the square brackets of the error indicator. With this 

message, the T E X processing is terminated. The source of this problem is 

hardly ever due to insufficient memory, no matter how much complicated 

text is being processed, but rather to an error in the text itself. The 

methods described in Section C.6 may be applied to try to detect the true 

error. 

The following descriptions of the various buffers should help to decide 

whether the storage capacity allotted to T E X really is too small and to 

explain what one might do to correct this. 

buffer size The problem here can be that the text in the argument of a sectioning, 

\caption, \addcontentsline, or \addtocontents command is 

too long. The message then normally appears when the \end{document} 

command is processed, but may also arise at one of \tableofcontents, 

\listoffigures, or \listoftables. The way to avoid this is to use the 

optional argument for the short form of the heading text (Sections 3.3.3 

and 7.4). Indeed, such a long entry in the table of contents is a nuisance 

anyway and should be shortened. After the correction has been made in the 

input text, the previous L AT E X.aux file must be deleted before reprocessing. 

This problem can occur on a PC if a word processing program has been 

used to generate the input text instead of a text editor. Some of these 

programs put an entire paragraph into a single line even though the text 

on the monitor is broken up into lines. 

exception dictionary The list of hyphenation exceptions that have been entered 

with the \hyphenation commands has become too large. Words 

used less frequently should be removed and their possible word divisions 

indicated explicitly with the \- command. 

hash size The source file contains too many command definitions or uses too 

many cross-reference markers. This does not mean that the input text 

really needs all these commands, for it may be that the user has developed 

a large collection of private commands that are stored in a single file and 

read into every document, whether they are all applicable or not. 

input stack size An overflow of this buffer is probably due to a mistake in a 

command definition. For example, the command defined with 

\newcommand{\com}{One more \com} 

produces One more {One more {...One more \com}...}} going on for 

ever, since it continually calls itself. Actually, it does not go on for ever, 

but only until this buffer is full. 

main memory size This buffer contains the text for the page currently being 

processed. It also overflows if a recursively defined command has been 

called. However, the more usual reasons are: (1) a large number of very 

complicated commands have been defined on one page; (2) there are too


many \index or \glossary commands on one page; (3) the page itself is 

too complex to fit within the allotted buffer space. 

The solution to the first two situations is clear: reduce the number of 

command definitions and/or \index and \glossary commands on that 

page. In the third case, the cause might be a long tabbing, tabular, 

array, or picture environment, or a stuck float (figure or table) waiting 

for an output command. 

To find out whether the memory overflow is really due to an overly complex 

page, add the command \clearpage just before the spot where the 

overflow occurs. If the error message no longer appears on the next processing 

run, then this page was indeed too complicated for T E X. However, 

if the overflow still persists, the error is probably a mistake in the input 

text. If necessary, it may have to be located with the method explained in 

Section C.6. 

If the page really is too complex for the T E X processing, it must be simplified. 

However, first recall that the entire last paragraph is processed before that 

page is output, even if the page break occurs near the beginning of the 

paragraph. Introducing a \newpage may solve the problem and should 

be tried out before attempting a tedious restructuring of the text. If the 

error is due to a stuck figure or table, it might be alleviated by moving the 

float object to later in the text, or by changing the positioning argument 

(Section 7.1). If the whole text is not yet complete, one can try giving 

\clearpage for now in order to clear the blocked floats and then decide 

on a better ordering when the text is finalized. 

pool size Most likely there are too many command definitions and/or labels, 

or their names are too long. Try shortening the names. This error may 

also occur if a closing right brace } has been forgotten in the argument 

of a counter command such as \setcounter or in a \newenvironment or 

\newtheorem command. 

save size This buffer overflows when commands, environments, and the scope 

of declarations are nested too deeply. For example, if the argument of a 

\multiput command contains a picture environment, which in turn possesses 

a \footnotesize declaration with another \multiput command, 

and so on. Such a nesting must be simplified, unless the real problem 

is a forgotten closing brace } that merely makes the structure appear so 

complex. 

! Text line contains an invalid character. 

The input text contains a strange symbol that T E X does not recognize. 

This could be a problem with the editor itself, that it is inserting extra 

characters. If an examination of the source file does not reveal the strange 

symbol, consult the computing center for help. 


Every T E X user encounters this error message at some point. It is usually

C.5. Warnings 429 

the result of an incorrectly typed command name. It may be amended 

during the processing by responding with I and the proper name of the 

command, plus 〈return〉. This does not alter the source file, which must 

be corrected separately after the L AT E X run. If the command name has 

been entered correctly in a L AT E X command, it may be that it was issued 

in an improper environment where it is not allowed (that is, not defined 

there). 

! Use of ... doesn’t match its definition. 

If ‘...’ is the name of a L AT E X command, it is likely that one of the 

picture commands from Sections 13.1.3 and 13.1.4 has been called with 

the wrong syntax for its arguments. If the name is \@array, there is a 

faulty @-expression (Section 4.8.1) in a tabular or array environment. 

Possibly a fragile command was given in the @-expression without the 

\protect command preceding it. 

! You can’t use ‘macro parameter #’ in ... mode. 

The special symbol # has been used in normal text. Probably there should 

have been a \# in order to print # itself. This can be corrected during the 

processing with the response I\# and 〈return〉. 

C.5 Warnings 

T E X and L AT E X errors both bring the processing run to a temporary stop and 

wait for a reaction from the user, or they may halt the program completely. 

Warnings, on the other hand, merely inform the user that the processed 

output may contain some faults that he or she might want to correct. 

Warnings appear on the monitor along with the page number where they 

occur, without the program coming to a stop. They are also written to 

the .log file where they may be examined after the L AT E X processing and 

possible printing. Warnings may be issued either by L AT E X or by T E X itself. 

C.5.1 General L AT E X warnings 

L AT E X warnings are indicated by the words ‘LaTeX Warning:’ at their start, 

followed by the warning message itself. 

LaTeX Warning: Citation ‘...’ on page ... undefined on 

input line .... 

The key in a \cite or other citation command has not yet been defined 

with a corresponding \bibitem command (Section 9.3). If this message 

does not disappear on the next run, then the bibliography is missing


that defining \bibitem entry, either because BIBT E X has not been run, or 

because that reference is not in any of the specified databases. 

LaTeX Warning: Command ... has changed. 

Check if current package is valid. 

The \CheckCommand statement is used to test if a given command has a 

certain definition. This warning is issued if the test fails. This is used to 

be sure that a given package is doing what the programmer thinks it is 

doing. 

LaTeX Warning: Float too large for page by ..pt on input line... 

A float (figure or table environment) is too big to fit on the page. It will 

be printed anyway, but will extend beyond the normal page margins. 

LaTeX Warning: ‘h’ float specifier changed to ‘ht’. 

The float placement specifier ‘h’ for ‘here’ (Section 7.1) does not really 

mean exactly at the location of the float environment. It will be placed 

‘here’ only if there is enough room remaining on the page for it, otherwise 

it appears at the top of the next available page. This message is a reminder 

of this fact. 

LaTeX Warning: inputting ‘...’ instead of obsolete ‘...’. 

Some packages written for L AT E X 2.09 explicitly input certain files, such 

as article.sty, that have a new equivalent with a different name (in 

this case article.cls). Dummy files with the old names that issue this 

message and input the right file are provided. 

LaTeX Warning: Label ‘...’ multiply defined. 

Two \label or \bibitem commands have defined keys with the same 

name (Sections 9.2.1 and 9.3). Even after the correction has been made, 

this message will be printed once more to the monitor since the information 

that led to it is to be found in the .aux file from the last processing 

run. On the run after that, it should be gone. 

LaTeX Warning: Label(s) may have changed. 

Rerun to get cross-references right. 

The printed outputs from the \ref, \pageref, and \cite commands may 

be incorrect since their values have been altered during the processing. 

L AT E X must be run once more so that the right values are used. 

LaTeX Warning: Marginpar on page ... moved. 

A marginal note on the given page has been shifted downwards to prevent 

it from being too close to another one. This means that it will not be beside 

the line of text where the \marginpar command was actually given.

LaTeX Warning: No \author given. 


The \maketitle command has been issued without a previous \author 

command. Unlike a missing \title command, this is not an error, but 

merely peculiar. The warning is printed just in case you did mean to 

specify an author. 

LaTeX Warning: Optional argument of \twocolumn too tall on page. 

The \twocolumn command (Section 3.2.7) starts a new page and switches 

to two-column formatting. The text in the optional argument is printed 

in one wide column above the double column text. If that text is too large 

to fit onto one page, this warning is printed. 

LaTeX Warning: \oval, \circle, or \line size unavailable on 


The size specified for an \oval, \circle, or slanted \line command in 

a picture environment is too small for L AT E X to print. 

LaTeX Warning: Reference ‘...’ on page ... undefined on 


The marker name in a \ref or \pageref command has not been defined 

by a \label command in the previous processing run (Section 9.2.1). 

If this message does not disappear on the next run, the corresponding 

\label command is missing. 

LaTeX Warning: Text page ... contains only floats. 

This message points out that the float style parameters (Section 7.3) have 

excluded any regular text from appearing on the specified page. This is 

not necessarily bad, but you might want to check that page visually. 

LaTeX Warning: There were multiply-defined labels. 

This message is printed at the end of the L AT E X run if there were any 

\label or \bibitem commands that used the same marker name more 

than once. A warning is also printed near the beginning of the run for 

each repeated marker name. 

LaTeX Warning: There were undefined references. 

This message is printed at the end of the L AT E X run if there were any \ref 

or \pageref commands whose markers had not been defined during a 

previous run. A warning is also printed earlier for each undefined marker 

used. If it does not disappear on the next run, then the corresponding 

\label is missing, maybe due to a typing error in the marker text.


C.5.2 L AT E X package warnings 

The class and package warnings are those that check the names and 

versions of the loaded files against those requested, or the use of options, 

or whether the filecontents environment has written some text to a 

file. The commands involved are described in Section D.2.9. 

Classes and packages can also issue their own warnings which are 

peculiar to them. These cannot be listed here, since they depend entirely 

on the class or package itself. 

LaTeX Warning: File ‘...’ already exists on the system. 

Not generating it from this source. 

The filecontents environment is not extracting text from the main file 

because it has discovered a file of the same name already on the system. 

LaTeX Warning: Unused global option(s):. 

Options specified in the \documentclass statement are global, meaning 

they can apply to the class and/or to any following packages. However, 

if no class or package recognizes any of these options, this warning is 

printed, followed by a list of the unused options. 

LaTeX Warning: Writing file ‘...’. 

The filecontents environment (Section D.2.9) is extracting text out of 

the main file and writing it to a file of the given name. 

LaTeX Warning: You have requested class/package ‘...’, 

but the class/package provides ‘...’. 

The name of the class or package as given by its internal identifying 

command \ProvidesClass or \ProvidesPackage does not agree with 

the file that was read in with \usepackage or \RequirePackage. 

LaTeX Warning: You have requested, on input line ..., version 

‘...’ of class/package ..., 

but only version ‘...’ is available. 

The date of a class or package file, as given by its internal identifying 

command \ProvidesClass or \ProvidesPackage, is earlier than that 

asked for by the \usepackage or \RequirePackage command. The class 

or package may not have all the features that the inputting file expects. 

LaTeX Warning: You have requested release ‘...’ of LaTeX, 

but only release ‘...’ is available. 

The date of your L AT E X version is earlier than that specified for some input 

file in a \NeedsTeXFormat command (Section D.2.1). Your L AT E X may not 

provide all the features needed by that file.

C.5.3 L AT E X font warnings 


Font warnings are those involving the NFSS commands (Appendix A). They 

are indicated by the text: LaTeX Font Warning: plus the warning text. 

LaTeX Font Warning: Command ... invalid in math mode. 

A command that should only appear in text mode has been given in math 

mode. The command is simply ignored. The commands \mathversion, 

\boldmath, \unboldmath, and \em lead to this message. There are other 

commands that produce an error message with the same text. 

LaTeX Font Warning: Command \tracingfonts not provided. 

(Font) Use the ‘tracefnt’ package. 

(Font) Command found: on input line .... 

The font tracing diagnostic tool \tracingfonts may only be used if the 

tracefnt package (page 393) has been loaded. Otherwise, this command 

is ignored. 

LaTeX Font Warning: Encoding ‘...’ has changed to ‘...’ for 

(Font) symbol font ‘...’ in the math version ‘...’. 

To make use of the specified symbol font in the given math version, it was 

necessary to change the font encoding temporarily. 

LaTeX Font Warning: Font shape ‘...’in size not available 

(Font) size substituted. 

No font has been defined for the size and shape requested, so a substitute 

size is used instead. 

LaTeX Font Warning: Font shape ‘...’ undefined 

(Font) using ‘...’ instead. 

The shape attribute that has been requested is unknown, or has not been 

defined, so a substitute shape will be used instead. 

C.5.4 T E X warnings 

A T E X warning is recognized by the fact that it is not an error message 

(not prefixed with !) and that the processing is not halted. The most 

common T E X warnings are: 

Overfull \hbox .... 

T E X could not break this line in a reasonable way so part of it will extend 

into the right margin. The rest of the information in the message can be 

of assistance.


! 

For example, with the complete warning message 

Overfull \hbox (17.2122pt too wide) in paragraph at lines 4--6 

[]\OT1/cmr/m/n/10 If T[]X can-not find an ap-pro-pri-ate 

spot to di-vide a word at the end of the line, as right 

here aaaaaaaaaaaaaaaaaaaaaaaaaa 

one knows that the line is about 17.2 pt (6 mm) too long and extends this amount 

into the right-hand margin. The line is part of the paragraph in lines 4 to 6. The 

font used is designated by its attributes \OT1\cmr/m/n/10. The text of the problem 

line is If ... ... right here aaaaaaaaaaaaaaaaaaaaaaaaaa. Possible 

word divisions are shown with hyphens, such as di-vide. The last word 

cannot be divided, which is the cause of the problem. 

A way around this is to include some suggested hyphenations with the \command, 

such as aaaaaaaaa\-aaaaaaaaaaaaaa. Similarly, a \linebreak command 

before the problem word, or setting the whole paragraph in a sloppypar 

environment, will get rid of this warning message. 

If a badly broken line extends only a tiny bit into the right margin, say 

1 pt or less, in most cases this will hardly be noticed and may be left as 

it is. A sample output should be printed just to verify the appearance. 

Overfull \vbox .... 

This warning occurs very rarely. T E X could not break the page properly 

so that the text extends beyond the bottom of the page. More often T E X 

sets less text on a page than too much. Thus this warning arises only 

when the page contains a very large vertical box, higher than the value of 

\textheight, such as a long table. 

Underfull \hbox .... 

This is the opposite of the Overfull \hbox warning. It appears when 

T E X has filled a line right and left justified, but with so much interword 

spacing that it considers the appearance to be undesirable. This is often 

the result of a sloppypar environment, a \sloppy declaration, or a 

\linebreak command. It may also come about after an inappropriate 

application of a \\ or \newline command, such as two \\ commands 

one after the other. The additional information in the warning message 

contains the text of the badly formatted line plus an evaluation of the 

‘badness’ of the word spacing. 

If in the example of the Overfull \hbox a \linebreak is included in the 

! text at the spot ..., as right\linebreak here, then the warning becomes 

Underfull \hbox (badness 5504) in paragraph at lines 4--6 

[]\OT1/cmr/m/n/10 If T[]X can-not find an ap-pro-pri-ate 

spot to di-vide a word at the end of the line, as right 

which states that the paragraph in lines 4 to 6 contains an output line in which 

the interword spacing may be unacceptably wide. The text of this line reads If 

... ..., as right and is set in the font \OT1/cmr/m/n/10. The evaluation

C.6. Search for subtle errors 435 

badness 5504 is T E X’s estimate of how unacceptable the spacing is: the smaller 

this number the better. 

Underfull \vbox .... 

The page has been broken with head and foot justified, but T E X judges 

the amount of inter-paragraph spacing to be possibly unacceptable. The 

badness number here corresponds to the quantity with the same name 

from the Underfull \hbox warning. 

C.6 Search for subtle errors 

At some point or another you will encounter an error message for which 

you cannot identify any cause, try as you may. For such devious errors, 

we recommend the following search strategy: 

1. Copy the file twice into a previous and a working copy (in addition 

to the original, which remains untouched during this search). 

2. In the working copy, find the outermost environment where the error 

occurred and remove one or more inner environments. If there are 

no inner environments, shorten the remaining text. Process the file 

with L AT E X once more. 

3. If the error still occurs, copy the shortened working copy to the 

previous copy and repeat step 2. If the outer environment in step 

2 is \begin{document} ... \end{document}, the shortening may 

be carried out by simply inserting \end{document} at some earlier 

point. 

4. If the error is no longer in the shortened working copy, copy the 

previous copy back to the working copy. The error is still present in 

this version. Remove less of the text than last time and repeat steps 

2 to 4. 

5. If the error is found to be in the next innermost environment, repeat 

the procedure for this environment with steps 2 to 4. 

With this strategy, the error may be localized to one command or to the 

innermost environment with only a small remaining structure. If the 

error still cannot be identified in spite of its being precisely localized, 

seek help from a more experienced colleague or from the computing 

center. However, it is normally possible to recognize the mistake once the 

position of the error has been found. 

It does happen that even though the error has been corrected, the same 

error message appears on the next L AT E X process run. This is because of 

the internal transfer in information through the L AT E X auxiliary files, which


are always one run behind on the current situation. For example, if there 

is a mistake in one of the sectioning commands, then after it has been 

eliminated, the faulty entry still exists in the .toc file. If the document 

contains the \tableofcontents command, L AT E X will read in that .toc 

file on the next run, and issue the error message once more, since a new 

.toc file is only created after a successful processing. 

In this case, the .toc file should also be edited and the error removed. 

If that is not possible, the file should be erased and the corrected 

source file processed twice with L AT E X. If the error was in \caption, 

\addcontentsline, or \addtocontents, the same applies to the corresponding.lof 

or.lot file. 

Occasionally the .aux file itself must be erased to prevent an error in 

it from being repeated even after the .tex file has been repaired. Here 

one must be careful that the command \nofiles has not been given in 

the preamble, for then a corrected .aux file will not be generated after a 

further L AT E X process run.

D 

LAT E X Programming 

In this appendix, we present the special commands that were designed 

for class and package files, with several examples of useful packages. We 

also explain the system for integrating documentation with coding in a 

single file, the system used for the core L AT E X files and standard packages. 

D.1 Class and package files 

D.1.1 The L AT E X concept, an open system 

The wealth of contributed L AT E X programming was probably never anticipated 

by Leslie Lamport when he released L AT E X. It is now a fact of life, 

and indeed one of the great strengths of the system. The L AT E X Team 

not only accommodates such ‘foreign’ extensions, it actually supports 

and encourages them, as witnessed by the copious presentations of such 

packages in The L AT E X Companion (Goossens et al., 1994) as well as in this 

book. 

And this is the way it should be. The extensions have been written by 

people who needed them, who realized that L AT E X was missing something 

vital for them. On the other hand, to add all of them to the basic L AT E X 

installation would overload it with features that 90% of the users would 

never require. The philosophy now is that L AT E X provides a fundamental 

core, or kernel, which is extended first by the standard class files, and 

then by the myriad of contributed packages and other classes. 

It is the role of the L AT E X Team to establish guidelines for programming, 

to ensure that packages do not clash needlessly with the kernel, or with 

each other, and to provide a basis of stability so that useful packages continue 

to operate through further updates to the kernel and the standard 

classes. These L AT E X features for class and package control, together with 

a set of programming tools, offers an enhanced degree of reliability and 

durability, both among packages and against future updates of the kernel. 

437

438 Appendix D. L AT E X Programming 

D.1.2 Levels of commands 

There are a number of levels of commands with varying degrees of security 

for the future: 

user commands (highest level) described in this and the other manuals, 

consisting of lower case letters, such as \texttt, are part of the 

L AT E X external definition to be supported forever; 

class and package commands with longish names of mixed upper and 

lower cases (like \NeedsTeXFormat) are intended mainly for programmers, 

and are also guaranteed; most are preamble-only commands, 

but there is otherwise no real restriction to class and package 

files; 

internal L AT E X commands containing the character @ in their names can 

only be used in class and package files; they are not guaranteed 

forever, although many of them are indispensable for special effects; 

a programmer makes use of them at the risk that some day his or 

her package may become obsolete; 

low-level T E X commands also have names with lower case letters, and no 

@; they should be safe against future evolution of L AT E X, but even this 

is not absolutely certain; they should be avoided where possible, as 

explained below; 

internal private commands are those used within a contributed class or 

package file; it is recommended that they all be prefixed with some 

upper case letters representing the package name and @ in order to 

avoid clashes with other packages; for example, \SK@cite, from the 

showkeys package. 

A question that confronts L AT E X programmers is to what extent the 

internal L AT E X commands may be used in class and package files. There is 

always a danger that such commands may vanish in later versions, since 

they have never been documented in the official books (Lamport, 1985, 

1994; Goossens et al., 1994). Like the T E X commands discussed in the 

next section, their use cannot be forbidden, but one must be aware that a 

certain degree of risk will accompany them. 

The guidelines issued by the L AT E X Team strongly recommend employing 

the high-level L AT E X commands whenever possible. 

• Use \newcommand and \renewcommand instead of \def; if one of 

the T E X defining commands must be used, because a template is 

required, or because it must be \gdef or \xdef, issue a dummy 

\newcommand beforehand to test for a name clash. If it is unimportant 

whether the command name already exists, issue a dummy

D.1. Class and package files 439 

\providecommand followed by \renewcommand. The ability to define 

commands with one optional argument at the high level removes 

one reason for wanting to reach down to the lower ones. 

• Use \newenvironment and \renewenvironment instead of defining 

\myenv and \endmyenv. 

• Assign values to lengths and glues (rubber lengths) with the 

\setlength command, rather than by simply equating. 

• Avoid the T E X box commands \setbox, \hbox, and \vbox; use instead 

\sbox, \mbox, \parbox and the like. With the extra L AT E X 

optional arguments, the need for the T E X equivalents is greatly diminished, 

and the L AT E X versions are far more transparent. Moreover, 

the L AT E X boxes will function properly with the color package, while 

the others are unpredictable. 

• Issue error and warning messages with \PackageError and 

\PackageWarning rather than with \@latexerr or \@warning; the 

former also inform the user of the source of the message, instead 

of labeling them all as L AT E X messages. 

• There is no suggestion that one should exclusively use the 

\ifthenelse command from the ifthen package (Section 8.3.5) 

in place of the T E X conditionals. It seems that this package is offered 

to simplify employing conditionals, in a manner more consistent 

with L AT E X syntax. Although most of the examples in this book use 

it rather than the T E X versions, we never employ it ourselves in our 

own programming. 

Adhering to these and similar rules will help ensure that a package will 

remain fit through future extensions of the L AT E X kernel. 

D.1.3 T E X commands 

Why should primitive T E X commands be shunned? To define commands 

with \def rather than \newcommand must be just as good, and is often 

unavoidable. Is there really a chance that it might be removed from a 

future L AT E X version? The primitives are the building blocks on which all 

flavors of L AT E X are constructed. Surely they must remain! 

This is not really the point. The primitive T E X commands form the 

bedrock of any format, and anything defined with them will always do 

exactly what the programmer expected. However, the equivalent L AT E X 

tools could actually do more as time goes on. The \newcommand checks 

for name clashes with existing commands, for example. It might even be 

possible that a debugging device that keeps track of all redefinitions could 

be added later; any commands defined with \def would be excluded from


such a scheme. Even now there is something like this to keep track of all 

files input with middle and high-level commands. 

Another example of how low-level programming can go astray is the 

case of robust commands. Many commands are intrinsically fragile, 

meaning that they are prematurely interpreted when used as arguments 

of other commands, but they may be made robust by prefixing them with 

\protect. In L AT E X 2.09, several fragile commands were defined to be 

robust by including the \protect in the definition, as for the L AT E X logo 

command: 

\def\LaTeX{\protect\p@LaTeX} 

\def\p@LaTeX{...} 

The true definition is in the internal \p@LaTeX, not the external \LaTeX. 

Since the original definition for the logo actually possessed some flaws, 

several packages included an improved version. Those that simply redefined 

\LaTeX itself made the command fragile; those that were cleverer 

only redefined \p@LaTeX, with the result that they are totally left behind 

in L AT E X 2ε, where commands are made robust in a completely different 

(and much better) manner. (Incidentally, the internal definition of the 

L AT E X logo today has been greatly improved.) 

In spite of the desirability of employing only official L AT E X commands, 

there are many occasions when either the internal L AT E X commands or the 

T E X primitives just must be used. The risk of future incompatibility must 

be taken in order to have a workable package now. However, one should 

not take this risk lightly, where a high-level equivalent is available. 

D.2 L AT E X programming commands 

All the commands described in this section are new to L AT E X 2ε. They 

are not essential to class and package files, but they do extend their 

usefulness and guarantee that they are employed properly. 

D.2.1 File identification 

Three commands test that the external environment in which the class or 

package has been inserted is correct. The first of these is 

\NeedsTeXFormat{format}[version] 

The first statement in a class or package should be the declaration of 

the T E X format needed. Although there are existing formats with other 

names, only the one named LaTeX2e actually recognizes this statement. 

All others will immediately issue the error message 

! Undefined control sequence.

l.1 \NeedsTeXFormat 

{LaTeX2e} 

D.2. L AT E X programming commands 441 

which all by itself is fairly informative. 

What is perhaps of more use is the optional version argument, which 

must contain the date of issue in the form yyyy/mm/dd. If a package 

makes use of features that were introduced in a certain version, its date 

should be given, so that if it is used with an earlier version of L AT E X 2ε, a 

warning is printed. For example, the command \DeclareRobustCommand 

did not exist in the preliminary test release of L AT E X 2ε, but was first 

introduced with the official release of June 1, 1994. Thus any package 

containing this command should begin with 

\NeedsTeXFormat{LaTeX2e}[1994/06/01] 

The form of the date is important, including the zeros and slashes. 

This declaration is not limited to class and package files: it may also 

be issued at the start of the document itself to ensure that it is processed 

with the right L AT E X. It must, however, be given in the preamble. 

The next two commands identify the class or package file itself: 

\ProvidesClass{class}[version] 

\ProvidesPackage{package}[version] 

In both cases, the version consists of three parts: date, version number, 

and additional information. The date is in the same format as above, 

while the version number can be any designation without blanks, and the 

additional information is text with or without blanks. An example is 

\ProvidesPackage{shortpag}[1995/03/24 v1.4 (F. Barnes)] 

Only the date part is actually checked by L AT E X against the date specified in 

the calling \usepackage command. The version number and additional 

information are printed out if \listfiles has been requested. However, 

the above format is necessary for the \GetFileInfo command in the doc 

package (Section D.7.2). 

Both the \documentclass and \usepackage commands (as well as 

\LoadClass and \RequirePackage) may take an optional argument to 

specify the earliest acceptable release date for the class/package. For 

example, with 

\documentclass[12pt]{article}[1995/01/01] 

if the article class file loaded contains 

\ProvidesClass{article}[1994/07/13 v1.2u 

Standard LaTeX document class] 

a warning message is printed. The same procedure applies to the commands 

\usepackage and \ProvidesPackage.


This system of version checking allows a document to insist that 

suitable versions of the class and package files are loaded. It assumes, 

however, that all later versions are fully compatible with earlier ones. 

There is a further identifying command for general files, those to be 

loaded with \input. 

\ProvidesFile{file name}[version] 

There is no checking of the name or version in this case, but both pieces 

of information will be printed by \listfiles. 

D.2.2 Loading further classes and packages 

In the main document file, classes are read in by means of the initializing 

\documentclass command, and packages with \usepackage. Within 

class and package files, the commands 

\LoadClass[options]{class}[version] 

\RequirePackage[options]{package}[version] 

\LoadClassWithOptions{class}[version] 

\RequirePackageWithOptions{package}[version] 

must be used instead. The first allows one class file to load another, with 

selected options, if desired; the second permits class and package files to 

load other packages. Only one \LoadClass command may appear within 

any class file; it may not be called from a package file. Neither command 

may be invoked in the document file. The packages argument may be a 

list of several package names, separated by commas. 

The ...WithOptions variants load the class or package with all those 

options that were specified for the current one, something that is often 

required. 

How the optional version arguments interact with the corresponding 

\Provides.. command has been explained in the previous section; how 

the options argument is treated is described below. 

D.2.3 Processing options 

Both classes and packages may take options which are defined with 

\DeclareOption{option}{code} 

where option is the name of the option and code is the set of instructions 

that it is to execute. Internally, a command named \ds@option is created. 

Often the code does nothing more than set flags, or input an option 

file. (\RequirePackage may not be used within the option code!) Two 

examples from article.cls are 

\DeclareOption{fleqn}{\input{fleqn.clo}} 

\DeclareOption{twocolumn}{\setboolean{\@twocolumn}{true}}


A default option is defined with \DeclareOption*, which takes no option 

name, specifying the code to be executed for all requested options that 

are undefined. 

There are two special commands that may be used only within the 

code of the default option definition: 

\CurrentOption contains the name of the option being processed; 

\OptionNotUsed declares \CurrentOption to be unprocessed. 

For example, to have a class file emulate L AT E X 2.09 behavior where all 

undefined options load a.sty file of the same name, define 

\DeclareOption*{\InputIfFileExists{\CurrentOption.sty}% 

{}{\OptionNotUsed}} 

which first checks if there is a.sty file of the requested name, and if not 

declares the option to be unused. Requested options that have not been 

used (processed) are listed in a warning message. 

The options are then processed with the commands 

\ExecuteOptions{option list} 

\ProcessOptions 

\ProcessOptions* 

where \ExecuteOptions calls those commands defined for the options 

in option list. This is normally done to establish certain options as being 

present by default. \ProcessOptions executes all the requested options 

in the order in which they were defined and then removes them from the 

list. Options are therefore executed only once by this command. The 

*-version is similar, except that the options are executed in the order 

requested. 

It is also possible to transfer options to a class or package file with 

\PassOptionsToClass{options}{class name} 

\PassOptionsToPackage{options}{package name} 

where options is a list of valid options recognized by the specified class 

or package file. These commands may be used within the definition of 

other options. The class or package named must later be loaded with 

\LoadClass or \RequirePackage. 

If the default options for class and package files have not been altered 

by \DeclareOption*, the standard procedure for handling options that 

have been requested but are undefined is 

• all options requested in the \documentclass statement are designated 

global; they are considered to apply to all subsequent packages, 

but not to classes loaded with \LoadClass; if they are not 

defined in the class, no error or warning is issued;


• all options requested with other commands, including \LoadClass 

and the \PassOptionsTo.., are local; if they are not defined in that 

class or package, an error is issued; 

• if there global options that are defined in neither the class nor any 

of the packages, a warning is issued; 

• options, global and local, are executed in the order in which they are 

defined in the class and packages, unless \ProcessOptions* has 

been called in which case they are executed in the order in which 

they are listed. 

D.2.4 Deferred processing 

Sometimes, to achieve certain special effects or to avoid possible conflicts 

with other packages, it is desirable to have some commands executed 

at the end of the package or class, or at the beginning or end of the 

document. This can be accomplished with 

\AtEndOfClass{cmds} 

\AtEndOfPackage{cmds} 

\AtBeginDocument{cmds} 

\AtEndDocument{cmds} 

The first two store away cmds to be carried out at the end of the class 

or package file. They can be used by local configuration files that are 

read in at the beginning, but contain modifications that should be made 

at the end so that they are not overwritten by the defaults. The last two 

declarations store away the cmds to be executed with \begin{document} 

and \end{document} respectively. All of these may be issued more than 

once, in which case the cmds are processed in the order in which they 

were issued. 

The cmds stored with \AtBeginDocument are inserted into the processing 

stream effectively within the preamble, but after the command 

\begin{document} has done almost everything else that it does. Thus 

the cmds may be considered to be part of the main body but preamble-only 

commands are also allowed. 

D.2.5 Robust commands 

Commands may actually be fragile, meaning that if they are used in the 

arguments of other commands, they could be prematurely interpreted, 

causing unexpected problems. This happens with moving arguments, 

those that appear somewhere else other than where they were given: in 

the table of contents and in running headlines. Complex commands with 

conditionals or redefinitions are likely to be fragile.


Many intrinsic commands in L AT E X 2.09 were fragile, and it was necessary 

to precede them with \protect when they appeared in the argument 

of a \section command, for example. In this case, the command name 

is transferred rather than its translation. Some commands could be made 

robust using the trick shown on page 440 for the \LaTeX command itself. 

In today’s L AT E X, almost all regular commands are robust. However, the 

commands a user may define with \newcommand, \renewcommand, and 

\providecommand (Section 8.3) may very well be fragile. Alternatively, 

one can define the commands with 

\DeclareRobustCommand{\com name}[narg][opt]{def } 

which has the same syntax as the other defining commands. If the command 

to be defined already exists, a message is written to the transcript 

file, and the old definition is overwritten. 

Another command with the same syntax simply checks the current 

definition of \com name: 

\CheckCommand{\com name}[narg][opt]{def } 

and issues a warning message if the actual definition is not the same 

as def, with the same number of arguments, and so on. This is used 

to ensure that the state of the system is as one expects, and that no 

previously loaded packages have altered some important definition. 

Both \DeclareRobustCommand and \CheckCommand may be called at 

any point in the document. 

D.2.6 Commands with ‘short’ arguments 

Normally, the arguments to user-defined commands are allowed to contain 

new paragraphs, with the \par command or with a blank line. In T E X 

jargon, these commands are said to be ‘long’. This is not the standard 

behavior for commands created with the T E X \def command, where the 

arguments must be short in order to act as a test for forgotten closing 

braces. 

With the version from December 1, 1994, L AT E X provides *-forms of all 

the defining commands: 

\newcommand* \renewcommand* 

\newenvironment* \renewenvironment* 

\providecommand* 

\DeclareRobustCommand* \CheckCommand* 

which create user-defined commands with ‘short’ arguments in the same 

way as does \def. 

It is recommended that one should almost always take the *-version 

of these defining commands, unless there is some very good reason to 

expect that the possible arguments may be ‘long’, that is, contain new 

paragraphs. Long arguments should be the exception, not the rule.


D.2.7 Issuing errors and warnings 

Classes and packages may be programmed to issue their own error messages 

and warnings. This is useful to indicate which file is responsible 

for the message. 

Error messages are generated with 

\ClassError{class name}{error text}{help} 

\PackageError{package name}{error text}{help} 

where error text is the message printed to the monitor and to the transcript 

file, and help is additional text printed after the user responds with 

H. If the texts contain command names that are to be printed literally, 

they must be preceded by \protect; spaces are generated with \space, 

and new lines with \MessageBreak. For example, 

\PackageError{ghost}{% 

The \protect\textwidth\space is too large\MessageBreak 

for the paper you have selected} 

{Use a smaller width.} 

produces the error message 

! Package ghost Error: The \textwidth is too large 

(ghost) for the paper you have selected. 

See the ghost package documentation for explanation. 


Typing H 〈return〉 produces 

Use a smaller width. 

after which L AT E X halts again to wait for a response as described in Section 

C.1. 

Warnings may also be issued from classes and packages in a similar 

way. The difference is that there is no help text, and the processing does 

not stop for a response. The line number of the input file where the 

warning occurred may be optionally suppressed. 

\ClassWarning{class name}{warning text} 

\ClassWarningNoLine{class name}{warning text} 

\PackageWarning{package name}{warning text} 

\PackageWarningNoLine{package name}{warning text} 

For example, with the warning 

\PackageWarning{ghost} 

{This text is haunted} 

one obtains the message 

Package ghost Warning: This text is haunted on input line 20.


and the processing continues. Warnings may be split into several lines 

with the \MessageBreak command, just like error texts. 

Two last commands of this type are 

\ClassInfo{class name}{info text} 

\PackageInfo{package name}{info text} 

which write their texts only to the transcript file, and not to the monitor. 

They are otherwise just like the corresponding NoLine warnings. 

D.2.8 Inputting files 

Files other than classes and packages may also be input, in which case it 

is often desirable to make sure that they exist beforehand. Or, alternative 

actions might be taken depending on the existence of a certain file. These 

goals are met with 

\IfFileExists{file name}{true}{false} 

\InputIfFileExists{file name}{true}{false} 

Both these commands test for the presence of the specified file name in 

the area that L AT E X is looking for files, and execute true if it is found, 

otherwise false. In addition, \InputIfFileExists reads in that file after 

executing true. 

These commands are not restricted to the preamble, nor to class or 

package files. In fact, the regular \input command is defined in terms of 

them. 

Many special classes make use of these commands to read in a local 

configuration file. For example, the class ltxdoc contains 

\InputIfFileExists{ltxdoc.cfg} 

{\typeout{Local config file ltxdoc.cfg used}} 

{} 

just before \ProcessOptions is called. This allows one to have a local 

configuration that might specify 

\PassOptionsToClass{a4paper}{article} 

for a European installation, without altering the files that are processed 

with the ltxdoc class. 

D.2.9 Checking files 

Although not really part of programming, two L AT E X features to keep track 

of input files are described here. The first of these, already mentioned in 

Section 9.1.1, is the command 

\listfiles


which may be given in the preamble, even before \documentclass. It 

causes a list of all input files to be printed at the end of the processing, 

along with their version and release data. In this way, one has a record 

of just which files were included, something that may be of use when 

deciding to send a document file to another installation for processing 

there. Since any non-standard files may also have to be included, these 

may be more readily identified from such a listing. 

For example, the simple document file 


\usepackage{ifthen} 

\listfiles 


\input{mymacros} 

This is \te. 


produces the listing 

*File List* 

article.cls 2001/04/21 v1.4e Standard LaTeX document class 

size10.clo 2001/04/21 v1.4e Standard LaTeX file (size option) 

ifthen.sty 2001/05/26 v1.1c Standard LaTeX ifthen package 

mymacros.tex 

*********** 

In this case, the local file mymacros.tex contains no version information 

because it is missing a \ProvidesFile command. 

What should one do if a local file, such as mymacros.tex above, is 

needed for the processing of a document file that is to be sent elsewhere? 

One could send it along with the main file, but that requires giving the 

recipient more instructions on what to do. Or, its contents could simply 

be included in the main file, for shipping purposes. For a package file, 

this is not so easy, since internal commands containing the @ sign would 

cause trouble, and the options would not be handled properly. For this 

purpose, we have the environment 

\begin{filecontents}{file name} 

file contents 

\end{filecontents} 

which may only appear at the very beginning of the document, before the 

\documentclass command. It tests to see if there is a file on the system 

with the name file name, and if not, it writes its contents literally to a file 

of that name. This may be a package file that is subsequently input with 

\usepackage. In this way, the missing non-standard files can be ported 

together with the main document file. 

If we extend the above simple example by including at the very start

\begin{filecontents}{mymacros} 

\newcommand{\te}{the end} 

\end{filecontents} 

the newly written file mymacros.tex contains 


%% LaTeX2e file ‘mymacros’ 

%% generated by the ‘filecontents’ environment 

%% from source ‘mydoc’ on 2003/01/31. 

%% 

\newcommand{\te}{the end} 

Note that the filecontents environment adds some comment lines 

to explain where the new file came from. If this is undesirable, the 

filecontents* environment may be used instead. 

D.2.10 Useful internal commands 

Notwithstanding the guidelines in Section D.1.2 that recommend avoiding internal 

! L AT E X commands, there are a number that are fairly fundamental, and indeed form 

many of the building blocks of the L AT E X kernel and many standard packages. Since 

they are still internal commands, they are not guaranteed for all future updates. 

However, if they were to vanish, many of the interesting extension packages 

provided by the L AT E X Team itself would have to be drastically overhauled. We 

merely present them briefly here for the sake of the bolder user. 

\@namedef{cmd}{def } 

\@nameuse{cmd} 

define and execute a command named \cmd, where the backslash is not included 

in the command name. This name may contain any characters, even those 

normally forbidden in command names. 

\@ifundefined{cmd}{true}{false} 

executes true if the command \cmd does not exist, else false. Again, the backslash 

is not included in cmd, and any characters may appear in the command name. 

This test is often used to define commands conditionally, a task that has been 

taken over by \providecommand. It may also be employed to determine whether 

the main class is article-like or not: \@ifundefined{chapter}{..}{..} tests 

for the existence of the \chapter command. 

\@ifnextcharchar{true}{false} 

tests if the next character is char, and if so, executes true, else false. This 

command is traditionally used to define commands with optional arguments, 

where char is [. The extended syntax of \newcommand offers a high-level means 

of achieving this. 

\@ifstar{true}{false}


tests if the next character is a star *, and if so, executes true, else false. It is used 

to define *-forms of commands and environments, something that still cannot 

be done at the high level. 

\@for \obj := \list \do {cmds} 

where \list is a command that is defined to be a list of elements separated by 

commas, and \obj is successively set equal to each of these elements while the 

code cmds is executed once for each element. For example, 

\newcommand{\set}{start,middle,end} 

\@for \xx:=\set \do {This is the \xx. } 

prints ‘This is the start. This is the middle. This is the end.’ 

D.2.11 Useful T E X commands 

Many of the most sophisticated features of L AT E X and its packages can only be 

! programmed with the help of Plain T E X commands. These are described not only 

in The T E Xbook (Knuth, 1986a), but also in the excellent reference manual by 

Eijkhout (1992), T E X by Topic, a T E Xnician’s Reference. 

We do not intend this book to a manual for T E X; nevertheless, there are a few 

common T E X commands that appear in many packages, and even in the examples 

to follow. A brief description of what they do will aid the understanding of these 

codings. A true T E Xpert or T E Xnician can skip this section altogether. 

\def\cmd#1#2..{definition} 

is the standard defining command in T E X. It is the equivalent of \newcommand* 

except that there is no check for name clashes, and the arguments are specified 

differently. For example, a command \Exp to write scientific notation can be 

defined as 

\def\Exp#1#2{\ensuremath{#1\times10ˆ{#2}}} 

or as \newcommand*{\Exp}[2]{\ensuremath{#1\times10ˆ{#2}}} 

In both cases, \Exp{1.1}{4} produces 1.1 × 10 4 . However, \def can go further: 

it can put the arguments in a template, such as 

\def\Exp#1(#2){\ensuremath{#1\times10ˆ{#2}}} 

to allow the more convenient notation \Exp1.1(4), something that cannot be 

produced with \newcommand. The \def command is often used when a command 

is to be defined without knowing (or caring) whether its name already exists, or 

when a template is needed. 

\gdef \edef \xdef 

are variations on \def: the first makes a global definition, valid even outside the 

current environment or {..} bracketing; the second is an expanded definition, 

such that any commands in it have their meanings and not the command itself 

inserted in the definition; the last is a combination of the other two, expanded 

and global.

\noexpand \expandafter 

D.3. Sample packages 451 

control the expansion of commands in definitions and execution. Any commands 

in the definition part of \edef are expanded (their meanings inserted) unless 

they have \noexpand before them. The opposite is achieved with \expandafter, 

which jumps over the following command, expands the next one, and then 

executes the one skipped. This is very deep T E Xnology, and is best illustrated by 

an example with the \Exp command defined above. 

\newcommand*{\mynums}{1.1(4)} \expandafter\Exp\mynums 

is identical to \Exp1.1(4), whereas \Exp\mynums is not; \mynums is expanded 

to 1.1(4) before \Exp is executed. 

\let\cmd a = \cmd b or \let\cmd a\cmd b 

makes \cmd a take on the current meaning of \cmd b. This is often employed 

to save the current meaning of a command before redefining it, possibly using 

the older meaning too. 

\relax 

does absolutely nothing, but it is often inserted in places where something should 

be but nothing is wanted. 

\ifcond true code \else false code \fi 

is the form of a T E X conditional. There are too many variations on the condition 

cond to explain here, but one common application is the equivalent of the L AT E X 

boolean switch commands: 

\newif\ifflag = \newboolean{flag} 

\flagtrue = \setboolean{flag}{true} 

\flagfalse = \setboolean{flag}{false} 

\ifflag ..\else..\fi = \ifthenelse{\boolean{flag}}{..}{..} 

For those who are used to it, the T E X form is more compact, but does not conform 

to the general L AT E X style of doing things. 

\ifcase num text 0 \or text 1 \or . . . \fi 

executes one of the text num according to the value of num. 

\endinput 

terminates the current file being input. This is not really necessary, but it is 

considered good programming to end all files this way. The main document file 

does not need it since \end{document} has the same effect. 

D.3 Sample packages 

We present here some demonstration packages to illustrate the programming 

commands of the previous section. These packages are not trivial 

and are all useful in their own right, even though their functionality is 

covered by others described in this book.


D.3.1 Modifying the text page size 

The standard L AT E X classes set the text size parameters \textwidth and 

\textheight according to the size option specified in \documentclass, 

such as a4paper or legalpaper. They also adjust the margins so that the 

text is centered both horizontally and vertically. The value of \textwidth 

is restricted, however, so that there are at most 60–70 characters per line, 

this being considered optimal by the rules of typography. 

Often one wants to ignore this limitation and use the paper to its 

maximum extent. The geometry package from Section 3.2.6 does this 

and much more. Here we present a simpler package for this single 

function. 

The fullpage package presented here uses the values of \paperwidth 

and \paperheight, which have been set according to the paper size 

option, to produce a margin of one inch on all sides. It even considers 

what the page style is so that room for head and footlines may be taken 

into account. Optionally, it may set an even narrower margin of 1.5 cm. 

Before proceeding, it might be useful to review the page format parameters 

shown in the figures on pages 48 and 602. 

The package file begins by stating that it needs L AT E X 2ε and by identifying 

itself. The package information contains the date in prescribed 

form, the version number, and additionally the initials of the author. The 

ifthen package is then loaded since conditionals will be required. 


\ProvidesPackage{fullpage}[2002/02/15 1.1 (PWD)] 


Next the options are prepared. These are to be in and cm for margins 

of 1 in and 1.5 cm respectively (1 cm is really too narrow). A special length 

is used to store the margin value. This is a private, internal command, 

conforming to the convention mentioned on page 438. 

\newlength{\FP@margin} 

\DeclareOption{in}{\setlength{\FP@margin}{1in}} 

\DeclareOption{cm}{\setlength{\FP@margin}{1.5cm}} 

Furthermore, the four standard page styles are also to be option names. 

They will set two internal boolean switches and the page style. 

\newboolean{FP@plain} 

\newboolean{FP@empty} 

\DeclareOption{plain}{\setboolean{FP@plain}{true} 

\setboolean{FP@empty}{false} 

\pagestyle{plain}} 

\DeclareOption{empty}{\setboolean{FP@plain}{true} 

\setboolean{FP@empty}{true} 

\pagestyle{empty}} 

\DeclareOption{headings}{\setboolean{FP@plain}{false}



\pagestyle{headings}} 

\DeclareOption{myheadings}{\setboolean{FP@plain}{false} 


\pagestyle{myheadings}} 

Finally, the default set of options is executed, and then the selected 

options are processed, in this case, in the order specified. The reason for 

this is that if more than one page style has been given, the last one should 

dominate. 

\ExecuteOptions{in,plain} 

\ProcessOptions* 

Now the calculation can begin. First, for plain and empty, there are no 

headlines, so set the relevant parameters to zero. This is when FP@plain 

is 〈true〉. Then set the space reserved for the footline to zero for the 

empty page style (FP@empty is 〈true〉). For headings and myheadings, 

this space is maintained since these styles normally have a plain page 

(with footline) on the first page. 

\ifthenelse{\boolean{FP@plain}} 

{\setlength{\headheight}{0pt} 

\setlength{\headsep}{0pt}}{} 

\ifthenelse{\boolean{FP@empty}} 

{\setlength{\footskip}{0pt}}{} 

With all the margins and headline and footline spacings set, the actual 

calculation can be made. Recall that the driver program leaves a 1 inch 

margin at the left and above, which must be subtracted from \topmargin 

and \oddsidemargin. 

\setlength{\textwidth}{\paperwidth} 

\addtolength{\textwidth}{-2\FP@margin} 

\setlength{\oddsidemargin}{\FP@margin} 

\addtolength{\oddsidemargin}{-1in} 

\setlength{\evensidemargin}{\oddsidemargin} 

\setlength{\textheight}{\paperheight} 

\addtolength{\textheight}{-\headheight} 

\addtolength{\textheight}{-\headsep} 

\addtolength{\textheight}{-\footskip} 

\addtolength{\textheight}{-2\FP@margin} 

\setlength{\topmargin}{\FP@margin} 

\addtolength{\topmargin}{-1in} 

Remember that \paperheight and \paperwidth are set to the full dimensions 

of the paper by the paper size option. If this is wrongly specified, 

of course, the final result will be incorrect. 

This completes the coding for the package file fullpage.sty. An 

example of how it might be invoked is


\documentclass[a4paper,12pt]{article} 

\usepackage[headings]{fullpage} 

. . . . . . 

D.3.2 Redesigning the head and footlines 

Something that is often demanded by L AT E X users is the possibility of 

altering the head and footlines that appear on each page. Standard L AT E X 

does provide a limited number of page styles (Section 3.2) but these are 

not always sufficient. The most flexible of them is the myheadings page 

style which allows the author to determine the text of the running heads 

with \markright and \markboth. However, the general format, including 

font style and placement of the page number, is still fixed by L AT E X. 

The fancyhdr package by Piet van Oostrum (Section 3.2.2) solves all 

this very nicely for most cases. Nevertheless we present this example to 

show how it is done internally, and to illustrate how some other special 

effects can be achieved. 

We wish to modify the headline for documentation of large projects 

requiring particular information in the running head. Apart from a short 

version of the title and the section and page numbers, they might demand 

the document serial number, date, and version number. It would be useful 

to have a generalized documentation page style that allows these entries 

to be specified outside of the page style definition itself. 

Since this coding is only feasible for the article class, we will create 

a new class, called dochead, that inputs article.cls and then defines 

a new page style. Actually this class could contain many other special 

features of which our new page style is only one. 

Again we start by declaring the T E X version that we need (release 4 

from December 1, 1995), and by identifying the class file. 


\ProvidesClass{dochead}[2002/04/27 1.5 (PWD)] 

This class has no options of its own, passing all specified ones on to 

the underlying article class by means of the \LoadClassWithOptions 

command, which only became available with release 4. 

\LoadClassWithOptions{article} 

Now come the commands that are used to enter the four pieces of 

information for the headline: short title, date, serial number, and version. 

Each of these commands stores its argument in an internal \DH@ 

command, each of which is initially set to be empty. Finally, the entering 

commands are declared to be allowed only in the preamble, since it would 

be disastrous if they were called after the main text had started. 

\newcommand*{\DocTitle}[1] {\renewcommand*{\DH@title}{#1}} 

\newcommand*{\DocDate}[1] {\renewcommand*{\DH@date}{#1}}


\newcommand*{\DocNumber}[1] 

{\renewcommand*{\DH@number}{\MakeUppercase{#1}}} 

\newcommand*{\DocVersion}[1]{\renewcommand*{\DH@version}{#1}} 

\newcommand*{\DH@title}{} \newcommand*{\DH@date}{} 

\newcommand*{\DH@number}{} \newcommand*{\DH@version}{} 

\@onlypreamble{\DocTitle} \@onlypreamble{\DocDate} 

\@onlypreamble{\DocNumber} \@onlypreamble{\DocVersion} 

The \@onlypreamble command is an internal L AT E X one. 

Note: The commands \MakeUppercase and \MakeLowercase (not used 

here) set the case of their arguments. The former is used here to ensure 

that the document number always appears in upper case letters. 

We now define the new page style, dochead, which means we must create 

a command \ps@dochead, to be executed by \pagestyle{dochead}. 

What this command must do is redefine the four internals \@oddhead, 

\@evenhead, \@oddfoot, and \@evenfoot. We make the footlines empty, 

and the even and odd headlines identical. That headline is to be a minipage, 

the full width of the page, containing a table with the relevant 

document information. 

\newcommand*{\ps@dochead}{% 

\renewcommand*{\@oddhead}{% 

\begin{minipage}{\textwidth}\normalfont 

\begin{tabular*}{\textwidth}{@{}l@{\extracolsep{\fill}}% 

l@{\extracolsep{0pt}:˜}l@{}}% 

\DH@number & Version & \DH@version \\ 

\DH@title & Section & \thesection \\ 

Date:˜\DH@date & Page & \thepage 

\end{tabular*}\vspace{0.5ex} \rule{\textwidth}{0.6pt}% 

\end{minipage}} 

\renewcommand*{\@evenhead}{\@oddhead} 

\renewcommand*{\@oddfoot}{} 

\renewcommand*{\@evenfoot}{} 

} 

\pagestyle{dochead} 

The last line activates the new page style. 

It is also necessary to increase the size of \headheight and \headsep 

because our headline is much higher than normal. We pick a height of 3.5 

times a regular line. 

\setlength{\headheight}{3.5\baselineskip} 

\setlength{\headsep}{3em} 

We now add a further refinement. Such documentation normally wants 

the pages to be numbered within the sections. Therefore, we want to 

redefine the command \thepage which prints the page number from 

the page counter (Section 8.1.4), and we need to modify the \section 

command to start a new page and to reset the page counter. This is done 

by storing the current definition of \section and then redefining it.


\let\DH@section=\section 

\renewcommand*{\thepage}{\thesection-\arabic{page}} 

\renewcommand*{\section}{\newpage\setcounter{page}{1}\DH@section} 

Notice that the new \section command calls the old one now stored in 

our own internal command \DH@section. 

If the above is written to a file named dochead.cls, the following 

source text could be used to invoke it: 

\documentclass[12pt,a4paper]{dochead} 

\DocTitle{Spacecraft Cleanliness} 

\DocNumber{Esa--xy--123} 

\DocDate{2003 Feb 26} 

\DocVersion{4.2} 


. . . . . . . . . . . . . . . . 

The headline on the fourth page of Section 3 would then appear as below. 

ESA–XY–123 Version: 4.2 

Spacecraft Cleanliness Section: 3 

Date: 2003 Feb 26 Page : 3-4 

With this example as a model, it should not be difficult to create similar 

headlines for other applications with different entries. 

D.3.3 Reprogramming the sectioning commands 

Another aspect of L AT E X formatting that one might want to alter is the 

appearance of the sectioning titles. The fonts used, as well as their sizes 

and the spacing around them, are all rigidly set in the class files. We know 

users who feel so frustrated by the predefined sectioning that they type 

in their own by hand, including the numbers, something that completely 

violates the L AT E X concept of logical markup. 

In fact, all the sectioning commands from \section to \subparagraph 

are defined by means of a generalized internal section command with the 

following syntax: 

\@startsection{sec-name}{level}{indent}{pre-skip} {post-skip} 

{style}*[short title]{title} 

where both the * and [short title] are optional. A command like \section 

is defined to be \@startsection with all the arguments up to and including 

style; the remaining optional *, short title, and title are given 

as arguments to \section itself (Section 3.3.3). The meanings of these 

arguments are:


sec-name: the name of the section level, for example subsubsection; 

this name is used to select the appropriate counter and to enter the 

title correctly in the table of contents and page headlines; 

level: is the number in the sectioning hierarchy as described on page 56, 

1 for \section, 2 for \subsection . . . ; 

indent: is the amount of indentation from the left margin; 

pre-skip: is a rubber length, the absolute value of which is the space inserted 

above the title; if this is negative, the first paragraph following 

the section title is not indented; 

post-skip: is a rubber length; if positive, it is the space inserted below 

the title; if negative, its absolute value is the space between the title 

and the subsequent run-in text; 

style: font declarations issued when printing the section title, such as 

\bfseries and \Large; 

*: optional; if present, the section counter is not incremented, the section 

is not numbered, no entry is made in the table of contents; 

short title: optional alternative text for the table of contents and page 

headline; may only be given if the * is absent; 

title: the text printed as the section title; if short title is missing, this text 

is also entered in the table of contents and page headline. 

The font style of the title is determined by the declarations in style; 

these must be declarations, like \bfseries, and not commands with 

arguments like \textbf. However, with the release from June 1, 1996, 

the last item in the style list may indeed be a command with an argument. 

This permits something like \bfseries\MakeUppercase to force the title 

to be bold face capital letters. 

Rubber lengths with generous stretch and shrink should be given for 

pre-skip and post-skip to allow L AT E X more freedom in avoiding bad page 

breaks. 

The section number is written with the command \@seccntformat 

that takes one argument, the name of the sectioning counter (sec-name 

above) that is to be printed. This may be redefined for special effects, as 

illustrated in the example below. 

Since both \@startsection and \@seccntformat contain @, they are 

internal commands that can only be used in a class or package file. We 

give here a brief package, mysects.sty, which redefines all the sectioning 

commands. The values of pre-skip and post-skip are those in the standard 

article.cls.



\ProvidesPackage{mysects}[2003/03/21 v2.1 (PWD)] 

The date of the fifth L AT E X release is specified because we want to use a 

command in the font styles. We start with the \section command which 

is to be centered, \Large and upper case. 

\renewcommand{\section}{\@startsection {section}{1}{0pt}% 

{-3.5ex plus -1ex minus -.2ex}% 

{2.3ex plus.2ex}% 

{\centering\normalfont\Large\MakeUppercase}} 

The subsections and subsubsection titles are in small caps and bold 

slanted fonts respectively. 

\renewcommand{\subsection}{\@startsection{subsection}{2}{20pt}% 


{1.5ex plus .2ex}% 

{\normalfont\large\scshape}} 

\renewcommand{\subsubsection}{\@startsection{subsubsection}{3}% 

{10pt}% 


{1.5ex plus .2ex}% 

{\normalfont\normalsize\bfseries\slshape}} 

The paragraph and subparagraph titles are to be run-in types (negative 

post-skip). 

\renewcommand{\paragraph}{\@startsection{paragraph}{4}{0pt}% 

{3.25ex plus 1ex minus.2ex}% 

{-1em}% 

{\normalfont\normalsize\underline}} 

\renewcommand{\subparagraph}{\@startsection{subparagraph}{5}% 

{\parindent}% 

{3.25ex plus 1ex minus .2ex}% 

{-1em}% 

{\normalfont\normalsize\itshape}} 

Finally, redefine the section numbers to have a period after them. 

\renewcommand{\@seccntformat}[1]{\@nameuse{the#1}.\quad} 

As an example of output with mysects, consider the short document: 


\usepackage{mysects} 

\setcounter{secnumdepth}{3} 


\section{Historical Outline} 

\subsection{Medieval Life} 

\subsubsection{The Role of the Priest} 

\paragraph{In the Church}

The priest fulfills several functions. 

\subparagraph{At the Altar} 

Here the priest was master of ceremonies. 


which produces output: 

1.1. Medieval Life 

1.1.1. The Role of the Priest 

D.4. Changing preprogrammed text 459 

1. HISTORICAL OUTLINE 

In the Church The priest fulfills several functions. 

At the Altar Here the priest was master of ceremonies. 

Because the counter secnumdepth was set to 3, numbering only goes to 

the third level. 

D.4 Changing preprogrammed text 

D.4.1 Changing explicit names 

There are a number of titles that appear automatically in L AT E X, such 

as ‘Contents’, ‘Bibliography’, ‘Chapter’. For works in languages other 

than English, it is necessary that they be replaced by their translations. 

And even in English, an author might prefer say ‘Summary’ in place 

of ‘Abstract’. All such explicit words are to be found in certain name 

commands that may be redefined as one pleases. 

These name commands are not defined in the basic L AT E X format itself, 

but rather in the various class files, as they are needed. Thus 

\chaptername exists only in classes book and report, but not in article. 

This means, if a package were to redefine the word ‘Chapter’ for all classes, 

it must do so with something like 

\providecommand*{\chaptername}{} 

\renewcommand*{\chaptername}{Chapitre} 

The standard set of name commands and their initial values are: 

(defined in book, report, and article classes) 

\contentsname {Contents} 

\listfigurename {List of Figures} 

\listtablename {List of Tables} 

\indexname {Index} 

\figurename {Figure} 

\tablename {Table}


\partname {Part} 

\appendixname {Appendix} 

(defined in book and report classes) 

\chaptername {Chapter} 

\bibname {Bibliography} 

(defined in article class) 

\abstractname {Abstract} 

\refname {References} 

(defined in letter class) 

\ccname {cc} 

\enclname {encl} 

\pagename {Page} 

\headtoname {To} 

(defined in makeidx package) 

\seename {see} 

\alsoname {see also} 

(defined only in certain packages) 

\prefacename {Preface} 

\glossaryname {Glossary} 

\proofname {Proof} 

The redefinition of these name commands is a fundamental part of 

the multilingual babel system (Chapter 11). The commands are not redefined 

directly but rather by means of certain \captionslanguage such 

as \captionsgerman and \captionsenglish that allow for convenient 

switching back and forth between different languages. For example, 

D.4.2 The date 

\newcommand*{\captionsgerman}{% 

\renewcommand*{\contentsname}{Inhaltsverzeichnis} 

... 

\renewcommand*{alsoname}{siehe auch}} 

The \today command for the current date is another one that outputs 

explicit English words. Its standard definition conforms to the American 

style of giving dates, that is ‘January 15, 2004’. If one wants to redefine 

this, either for the British style (15th January 2004), or for another language 

altogether, the best method is to follow the example of the other 

names commands: do not redefine \today directly, but create commands 

that allow one to switch back and forth. 

For example, we can define \dateUSenglish and \dateenglish making 

use of the internal T E X counters \year, \month, \day and the T E X

\ifcase command (page 451). 

D.5. Direct typing of special letters 461 

\newcommand*{\dateUSenglish}{\renewcommand*{\today}{% 

\ifcase\month \or 

January\or February\or March\or April\or May\or June\or 

July\or August\or September\or October\or November\or 

December\fi \space\number\day, \number\year}} 

\newcommand*{\dateenglish}{\renewcommand*{\today}{% 

\number\day \ifcase\day \or 

st\or nd\or rd\or th\or th\or th\or th\or th\or th\or th\or 

th\or th\or th\or th\or th\or th\or th\or th\or th\or th\or 

st\or nd\or rd\or th\or th\or th\or th\or th\or th\or th\or 

st\fi\space \ifcase\month \or 

January\or February\or March\or April\or May\or June\or 

July\or August\or September\or October\or November\or 

December\fi \space\number\year}} 

Definitions for other languages can be modeled after these examples. 

D.5 Direct typing of special letters 

Package: In Section 2.5.9, we explained the inputenc package that permits direct 

inputenc typing of accented and special letters, provided one has an appropriate 

keyboard and display font. For example, on a German keyboard set 

up for Windows, typing the key marked ‘ß’ inserts character 223, which 

is displayed on the author’s monitor as the ‘eszet’ letter for which the 

regular LAT EX input is \ss. The source file however, receives only the 

code number 223, which LAT EX would normally reject as unknown. The 

inputenc package with the ansinew option tells it that character 223 is 

to be treated as \ss. This will have the same result even if the file is 

displayed on a monitor set up for a different coding. 

Table D.1 lists the current set of encoding options for the package. 

Others may be added in the future. For each of these options, there is 

corresponding .def file defining the extended characters with the commands 

\DeclareInputText{pos}{text} or 

\DeclareInputMath{pos}{math} 

which assign text or math coding to the character pos. For example, 

latin1.def contains 

\DeclareInputText{223}{\ss} 

to translate the input character code 223 to the L AT E X command \ss. 

Some T E X installations automatically contain such conversions for their 

own local input coding scheme; files written under such a system without


Table D.1: Input coding schemes for inputenc package 

ascii 7-bit ascii encoding, characters 32–127 only 

latin1 ISO Latin-1 encoding (Western Europe) 

latin2 ISO Latin-2 encoding (Eastern Europe) 

latin3 ISO Latin-3 encoding (Catalan, Esperanto, Galacian, Maltese) 

latin4 ISO Latin-4 encoding (Scandinavian, Greenland Inuit, Lappish) 

latin5 ISO Latin-5 encoding (Turkish) 

latin9 ISO Latin-9 encoding (with euro symbol) 

decmulti DEC Multinational encoding 

cp850 IBM 850 code page (Western Europe) 

cp852 IBM 852 code page (Eastern Europe) 

cp437 IBM 437 code page (North America) 

cp437de Variant on 437, with German ß replacing Greek β in position 225 

cp865 IBM 865 code page (Scandinavia) 

applemac Macintosh encoding 

next NeXt encoding 

ansinew Windows ANSI encoding 

cp1252 Windows 1252 code page (same as ansinew) 

cp1250 Windows 1250 code page, for Central and Eastern Europe 

the inputenc package will be processed correctly on their own system, 

but will generate garbage on another one. Thus this package should 

always be added when special symbols are directly typed in, especially if 

there is any possibility that the file may be processed on another computer 

system. 

D.6 Alternatives for special symbols 

A number of symbols in the text fonts cannot be addressed explicitly, but 

only as ligatures, for example ¿ as ?‘. Other symbols that one might want 

in text are only available in math mode. L AT E X provides some \text.. 

commands, listed in Table D.2 on the opposite page, to print these and 

other characters directly. 

D.7 Managing code and documentation 

This section describes two additional advanced features for maintaining 

L AT E X code and its documentation. There is no need to employ these at 

all, for any home-written packages and classes will function just as well 

without them. They do add a degree of sophistication and security, by 

tracing the history of the code’s development, and by including the current 

documentation within the code itself. Here ‘documentation’ means both 

the user’s manual and the programmer’s description of what each step of

D.7. Managing code and documentation 463 

Table D.2: Alternative commands for special symbols in text mode 

Command Symbol Replaces 

Ligatures 

\textemdash — --- 

\textendash – -- 

\textexclamdown ¡ !‘ 

\textquestiondown ¿ ?‘ 

\textquotedblleft “ ‘‘ 

\textquotedblright ” ’’ 

\textquoteleft ‘ ‘ 

\textquoteright ’ ’ 

Math symbols 

\textbullet • $\bullet$ 

\textperiodcentered · $\cdot$ 

\textbackslash \ $\setminus$ 

\textbar | $|$ 

\textless < $ $>$ 

Miscellaneous 

\textvisiblespace \verb*+ + 

\textasciicircum ˆ \verb+ˆ+ 

\textasciitilde ˜ \verb+˜+ 

\textsuperscript{123} 123 $ˆ{123}$ 

Special symbols 

\textcompwordmark (ligature break) 

\textregistered ® 

\texttrademark 

\textcircled{x} x○ 

the code is doing. 

The L AT E X installation files are written in this way. There are about 40 

such files that are processed with the DocStrip utility (next Section) to 

extract the code from these files and merge it into one file latex.ltx. 

The documentation for each contributory file can be viewed alone, or a 

giant document can be created for the entire set. On the other hand, 

the standard class files, with much common coding, along with their 

option files, are all extracted from the one file classes.dtx, with a single 

documentation. 

More practically, many packages are provided as.dtx files containing 

the package code plus user manual and programmer’s documentation. A 

DocStrip batch job with extension.ins is also given, which when processed 

by L AT E X (or Plain T E X for that matter) extracts the actual .sty file, along 

with a drive.drv file. L AT E Xing the driver file generates the documentation. 

This is usually superfluous since the.dtx file itself is so constructed that


it behaves like a driver file. However, the .drv file can be edited by the 

user to select, say, just the user manual and to suppress the programmer 

notes. 

The next sections describe first the extraction program DocStrip, which 

can be used for many other applications as well (e.g. custom-bib, Section 

14.3), and then the integrated documentation scheme in the following 

section. 

D.7.1 The DocStrip utility 

Since the T E X program is a programming as well as a text formatting 

language, it can be exploited to provide a number of utility ‘programs’ to 

manipulate files. Such programs are immediately portable: if you have 

T E X, you can run them. One example of such a program is the DocStrip 

utility for extracting functioning code out of one or more source files 

containing documentation, as comments. The program was originally 

written by Frank Mittelbach and further developed by Johannes Braams, 

Denys Duchier, Marcin Woliński, and Mark Wooding. 

The basic idea was to copy one file to another leaving off all the 

comment lines. From this simple concept came two additional features 

that extended the application of DocStrip: alternative lines of coding 

could be selectively suppressed or included depending on options chosen 

at processing time; and multiple files, or modules, could be input to form 

a single output file. This means that one source file can be the home 

of several different L AT E X packages, and that a single package file can be 

constructed out of many different components, as for the main L AT E X file 

latex.ltx mentioned above. 

Running interactively 

The simplest way of running DocStrip on a file is to invoke it with T E X (or 

L AT E X, but T E X is faster), as 

tex docstrip 

which produces the response 

********************************************************** 

* This program converts documented macro-files into fast * 

* loadable files by stripping off (nearly) all comments! * 

********************************************************** 

**************************************************** 

* First type the extension of your input file(s): * 

\infileext= 

****************************************************


One replies by entering the input extension (usually .dtx); then one is 

asked in turn for the output extension, the options wanted, and finally, 

for the root name of the file(s) to be processed. Execution follows. 

There are some limitations to this method, since the input and output 

files have the same root name, differing only in the extension, and initial 

and final comments added to the output are fixed. The more flexible 

means of running the utility is with a batch job. 

Running as a batch job 

A DocStrip batch job is a file containing instructions for the utility. For 

example, suppose the fullpage package illustrated in Section D.3.1 is 

put into a documented source file called fullpage.dtx, and the actual 

package file fullpage.sty is to be extracted with the option package, 

while a documentation driver file fullpage.drv is obtained with option 

driver; the batch file, named fullpage.ins, could look like 

\input docstrip 

\preamble 

This is a stripped version of the original file. 

\endpreamble 

\postamble 

This is the end of the stripped file. 

\endpostamble 

\declarepreamble\predriver 

This is a documentation driver file. 

\endpreamble 

\declarepostamble\postdriver 

End of documentation driver file. 

\endpostamble 

\keepsilent 

\askforoverwritefalse 

\generate{\file{fullpage.sty}{\from{fullpage.dtx}{package}} 

\file{fullpage.drv}{\usepreamble\predriver 

\usepostamble\postdriver 

\from{fullpage.dtx}{driver}} 

} 

\endbatchfile 

The first and last lines are vital: first the file docstrip.tex is loaded, 

defining all the special DocStrip commands; then after the instructions 

have been executed, \endbatchfile ensures an orderly termination.


The commands \preamble and \postamble allow one to insert explanatory 

comments at the beginning and end of the extracted file. 

The preamble is often a copyright notice and/or a caveat that the extracted 

file should never be distributed without the source. Additional 

labeled pre- and postambles may be declared with \declarepreamble 

and \declarepostamble, and activated with declarations \usepreamble 

and \usepostamble. 

The instructions \keepsilent and \askforoverwritefalse are optional. 

The former suppresses processing information during the run 

while the latter turns off the warning that an existing file may be overwritten. 

The main command is \generate which specifies the files to be created 

with a series of \file commands, each taking two arguments: the name of 

the new file and a list of instructions for its production. These instructions 

can be declarations like \keepsilent or \usepreamble, but the main one 

is the command \from. This again takes two arguments: the name of the 

input file and a list of the options to be applied. In the above example, 

each generated file has only a single input source, but multiple input files 

are possible with a series of \from commands. 

The created fullpage.sty file now contains 

%% 

%% This is file ‘fullpage.sty’, 

%% generated with the docstrip utility. 

%% 

%% The original source files were: 

%% 

%% fullpage.dtx (with options: ‘package’) 

%% This is a stripped version of the original file. 

%% Copyright (C) 1994 Patrick W. Daly 


. . . . . . . . . . . . . . . . . . . . . . 


%% This is the end of the stripped file. 

%% 

%% End of file ‘fullpage.sty’. 

It is possible to have a master batch job to process individual ones, in 

which case the master must input them with \batchinput, and not with 

\input. 

Rules for removing lines 

The \generate and \file commands cause the input file(s) to be transferred 

to the output file, line by line, according to the following rules: 

1. any lines beginning with a single % sign are removed;


2. any lines beginning with double %% signs are retained; 

3. any line beginning with % (or %) will have the rest of its 

text transferred if opt is one of the selected options; otherwise it is 

removed; 

4. any line beginning with % (or %) will have the rest of its 

text transferred if opt is not one of the selected options; otherwise 

it is removed; 

5. all lines between % and % are retained or removed 

depending on whether opt is a selected option or not. 

Options can be combined logically, negated, and grouped: 

a&b (a and b); 

a|b (a or b); 

!a (not a); 

(a|b)&c (c and one of a or b) 

For more information on DocStrip, see the documentation which can 

be obtained by processing docstrip.dtx with L AT E X. 

D.7.2 Documenting L AT E X coding 

The documenting of software products is extremely important: it provides 

on the one hand a manual for the user, and on the other details about 

the coding for the programmer. The original L AT E X styles, as well as the 

basic latex.tex file, were heavily commented by Leslie Lamport, but only 

with straightforward normal text. It was Frank Mittelbach’s doc package 

that first allowed sophisticated, integrated documentation of the source 

codes. 

By integrated documentation, we mean that the descriptions and the 

coding are to be found merged together in a single source file. Thus 

two processes are necessary, one to extract the actual coding on its own 

(the DocStrip utility of Section D.7.1) and another to print the documentation 

(the doc package). The basic idea behind this package is that the 

comments are in fact regular L AT E X text, with some extra features to allow 

automatic indexing and to record the program’s development. The coding 

itself appears in a special type of verbatim environment. 

The documentation is produced by running a special driver file through 

L AT E X. Such a driver for a source file named fullpage.dtx would be, in 

its simplest form, 


\usepackage{doc} 


\DocInput{fullpage.dtx} 

\end{document}


The \DocInput command reads in the specified file, but it first alters the 

function of the % character from ‘comment’ to ‘do nothing’. This means 

all comment lines in fullpage.dtx become real text to be processed! 

We describe some of the extra features that the doc package makes 

available for the ‘comments’ in the .dtx files, illustrated by an example, 

the source file fullpage.dtx for the fullpage package presented in 

Section D.3.1. As usual, a more complete manual can be acquired by 

processing doc.dtx. 

The description part 

The documentation consists of two parts: the description, which is a 

manual for the end user, and the coding, a detailed explanation of how 

the software works, including the lines of code themselves. It is possible 

to suppress the coding part and to print only the description by issuing 

\OnlyDescription 

in the preamble. 

Special commands for the description part are: 

\DescribeMacro{\macro name} 

\DescribeEnvironment{env name} 

which are placed at the start of the text that illustrates a new high-level 

command (macro) or environment. These commands do two things: they 

place the macro or environment name in a marginal note at that location 

(for easy reference when reading) and they insert an entry in the index. 

Because documentation needs to use the \verb command frequently 

for printing input text, some abbreviations are provided with 

\MakeShortVerb{\c} and \DeleteShortVerb{\c} 

which first turn the character c into shorthand for \verbc, and then restore 

its normal use. For example, after \MakeShortVerb{\|}, |\mycom| 

prints \mycom. (These commands can be made available for any document 

by loading the package shortvrb, which is actually extracted from the 

doc package.) 

If the comment character % has been deactivated, how can one put 

comments into the documentation text? One way is to make use of the 

T E X conditional \iffalse to form a block, or meta comment, as 

% \iffalse 

% These lines are ignored even when the 

% percent character is inactive 

% \fi 

The other method is to use ˆÂ in place of %, a special doc feature. 

The description part is terminated with

\StopEventually{final text} 


where final text is to appear at the very end of the article; if only the 

description part is printed, final text is printed immediately and the 

documentation is ended. 

The coding part 

The coding part should normally contain the more specialized material 

that is of no interest to the everyday user. The special commands that 

may be used here are 

\begin{macro}{\macro name} text and code \end{macro} 

\begin{environment}{env name} text and code \end{environment} 

both of which again insert a marginal note and make an entry in the index. 

They also organize any \changes commands, as explained below. 

The most important environment in the coding part is macrocode, 

which prints its contents as in verbatim, optionally with a code line 

number. The form of this environment is somewhat special: 

\begin{macrocode} 

lines of code 

\end{macrocode} 

The four spaces before the \end{macrocode} are obligatory; those before 

the \begin are not necessary, but it is good practice to insert them for 

symmetry. This environment also counts all the backslashes within it for 

a checksum test, and makes an index entry for every command name that 

it finds. So it is something more than a mere verbatim environment! 

The coding part is brought to an end with 

\Finale 

which carries out the checksum test and prints the stored final text from 

\StopEventually. There may actually be more text following it, which 

is only printed when both description and coding parts are output. 

Index of macros and record of changes 

The doc package makes automatic entries into an index by means of 

the two \Describexxx commands and the two environments presented 

above. As well, all commands that appear in the coding are indexed. 

However, the indexing is turned on only if one of 

\CodelineIndex or \PageIndex 

is given in the preamble. The first references the indexed commands to 

the number of the code line where they appear, the second to the page


number. In the second case, the code lines are not numbered, unless the 

declaration \CodelineNumbered is also issued. 

Since not all commands in the coding really need to be indexed, especially 

those that are part of standard L AT E X and T E X, the command 

\DoNotIndex{list of command names} 

is given, often repeatedly, near the beginning, to exempt the listed commands 

from being indexed. 

The automatic indexing of all the commands in the code slows down 

the processing considerably. Once the index.idx file has been produced, 

future runs do not need to repeat this effort (unless there have been 

changes to the code). The command 

\DisableCrossrefs 

will suppress this indexing, but it may be countermanded by an earlier 

\EnableCrossrefs which neutralizes the disabling command. 

The text of the index is generated from the .idx data file by the 

MakeIndex program (Section 9.4.3), which must be run with the special 

indexing style gind.ist as 

makeindex -s gind.ist filename 

The indexing style file gind.ist can be extracted from doc.dtx. 

To print the index, the command 

\PrintIndex 

is placed where it should appear. Often this is part of the final text in 

\StopEventually. An up-to-date index can only appear after MakeIndex 

has been run between two L AT E X processings. 

A record of changes to the software can be made by inserting 

\changes{version}{date}{text} 

throughout the documentation, in both the description and coding parts. 

To form the change history list, the command 

\RecordChanges 

must be placed in the preamble. This enables the change entries to be 

placed in a glossary file, which is then processed by MakeIndex as 

makeindex -s gglo.ist -o filename.gls filename.glo 

(The indexing style file gglo.ist is also extracted from doc.dtx.) The 

change history is then printed in the documentation where 

\PrintChanges 

is located, again often as part of final text. The texts of the \changes 

commands are ordered, first by version number, then by the name of the 

macro or environment in which they appear.

Integrity tests 


If the source file is to be sent over electronic networks, there is a danger 

that it might be corrupted or truncated. Two tests are possible to check 

for this. By placing 

\CheckSum{num} 

near the start of the documentation (before the coding anyway), all the 

backslashes in the macrocode environments will be added up, and the 

total compared with the number num by the \Finale command. If 

num=0, the true total will be printed on the monitor; otherwise, if the 

sum does not agree with num, an error is printed, with the two values. 

The other test checks that the character set has not been corrupted by 

passing through computer systems with different character codes. 

\CharacterTable 

{Upper-case \A\B\C\D\E\F\G\H. . . 

. . . . . 

. . . . . Tilde \˜} 

The argument must agree exactly with that expected by doc (except for 

inactive % signs and multiple spaces). It should be copied from the 

doc.sty or doc.dtx files. 

Obtaining the file information 

File information may be obtained with the command 

\GetFileInfo{filename} 

which defines \filename, \filedate, \fileversion, and \fileinfo 

from the optional release information to be found in the \Providesxxx 

command identifying the specified file (Section D.2.1). The idea is that 

the .dtx file should contain this information once, and only once, but it 

needs to be known to print it in the title of the article. These \filexxx 

commands may be used for this purpose. The release information must 

conform to the sequence date, blank, version, blank, text. 

The ltxdoc class 

A special class called ltxdoc is provided to assist running the doc package. 

It invokes the article class with the doc package, and then issues 


\AtBeginDocument{\MakeShortVerb{\|}} 

\CodelineNumbered 

\DisableCrossrefs


and defines a number of other useful commands for aiding the documentation. 

See the description by processing ltxdoc.dtx. It also provides for 

local configuration: if ltxdoc.cfg exists, it is read in. This can pass paper 

size or other formatting options to article, issue \OnlyDescription 

by means of \AtBeginDocument, and so on. 

A sample.dtx file 

To illustrate these features, we show part of the source file fullpage.dtx. 

The initial lines ensure that all the files that can be extracted from it (the 

package .sty and the documentation driver .drv) receive their proper 

identifying commands. The date and version information appears only 

once, but is transferred to both extracted files. 

% \iffalse (This is a meta-comment) 

%% Copyright (C) 1994 Patrick W. Daly 


% 

\ProvidesFile {fullpage.dtx} 

% 

%\ProvidesPackage{fullpage} 

%\ProvidesFile{fullpage.drv} 

% \fi 

%\ProvidesFile{fullpage} 

[2002/02/15 1.1 (PWD)] 

It is the last \ProvidesFile{fullpage} that enables the proper functioning 

of \GetFileInfo; the first one is only a dummy to absorb the 

information line when the.dtx file is read directly. 

Next, the driver part is given. This is what L AT E X sees when it processes 

the file directly. 

%\iffalse 

% 

\documentclass[a4paper,11pt]{article} 

\usepackage{doc} 

\EnableCrossrefs 

\RecordChanges 

\CodelineIndex 


\DocInput{fullpage.dtx} 


% 

%\fi 

Now the checksum and list of commands that are not to be indexed 

are given, followed by the start of the article. 

% \CheckSum{73} 

% \DoNotIndex{\addtolength,\boolean,\ExecuteOptions,\ifthenelse}


% \DoNotIndex{\newboolean,\newlength,\ProcessOptions} 

% \DoNotIndex{\RequirePackage,\setboolean,\setlength} 

% \changes{1.0}{1994 Feb 15}{Initial version} 

% \changes{1.1}{2002 Feb 15}{Next revision} 

% \GetFileInfo{fullpage} 

% \title{\bfseries A Package to Set Margins to Full Page} 

% \author{Patrick W. Daly} 

% \date{This paper describes package \texttt{\filename}\\ 

% version \fileversion, from \filedate} 

% \maketitle 

% \MakeShortVerb{\|} 

% 

% \section{Purpose} 

% To set a uniform margin of one inch or 1.5˜cm on all four 

% . . . . . . 

We advance to the end of the description and start of the coding part. 

% \StopEventually{\PrintIndex\PrintChanges} 

% 

% \section{The Coding} 

% The first thing is to read in the \texttt{ifthen} package, 

% if it is not already there. 

% \begin{macrocode} 

% 


% \end{macrocode} 

% 

% \begin{macro}{\DeclareOption} 

% \begin{macro}{\FP@margin} 

% Define the options with help of the length |\FP@margin|. The 

% options |in| and |cm| select the actual margin size. 

% \begin{macrocode} 

\newlength{\FP@margin} 

\DeclareOption{in}{\setlength{\FP@margin}{1in}} 

\DeclareOption{cm}{\setlength{\FP@margin}{1.5cm}} 


% \end{macro} 

The end of the file finishes the coding and calls \Finale. 

. . . . . . . 

\setlength{\topmargin}{\FP@margin} 


% 


% \end{macro}\end{macro} 

% \Finale

E 

LAT E X and World Wide 

Web 

Today it is no longer sufficient to produce professional-looking printed 

output on paper; one has to be able to get it online as well, that is, make 

it available in electronic form of some kind. To some extent, PostScript 

output fulfills this requirement in that it may be considered electronic 

paper. However, as we point out in Section 15.3, true electronic documents 

are something quite different. 

The Internet used to be a nice, quiet neighborhood where academics 

could exchange simple, plain text emails or obtain known files by FTP, until 

it was invaded by the rowdy World Wide Web with its corporate identities, 

slick advertising, energetic yuppies, anarchists, and revolutionaries. In 

other words, it went cosmopolitan. It is here in this glittering marketplace 

that electronic documents have to compete for attention. 

An electronic document is not something that is simply sent to a 

printer or leisurely viewed on a monitor: it is interactive, in that it guides 

the reader to other parts of itself or to other documents located anywhere 

else on the Web. With these links, the viewer can jump to other relevant 

passages with a mere mouse click. Colored illustrations are naturally 

part of such documents, but so are sounds and movies, as well as means 

of sending feedback to the author. This is a totally new medium for 

information exchange, more radical than the Gutenberg revolution from 

handwritten manuscripts to printed books. 

In Section 1.2 we explain the structural similarities between L AT E X and 

the Web languages HTML and XML. Here we point out the possibilities for 

conversion from L AT E X to the others, and then how T E X can be used as an 

engine to render XML documents in L AT E X formats. Detailed descriptions of 

the conversion programs are to be found in the supplied documentation, 

and in the L AT E X Web Companion (Goossens and Rahtz, 1999). This topic 

is far too extensive to be dealt with in this book. 

475

476 Appendix E. L AT E X and World Wide Web 

E.1 Converting to HTML 

E.1.1 The L AT E X2HTML program 

The L AT E X2HTML translator, written by Nikos Drakos with additions by a 

large number of contributors, is a comprehensive Perl script that converts 

a L AT E X source file into HTML with the help of several other programs, 

notably L AT E X itself, dvips, Ghostscript, and the netpbm library of graphics 

utilities. 

When L AT E X2HTML processes a L AT E X file, it creates a new subdirectory 

with the same name as that file, to which it writes the resulting HTML 

output and any generated images as .gif files. The program interprets 

the L AT E X input text in the same way as L AT E X itself does, but instead of 

producing typesetting instructions in the .dvi file, it writes appropriate 

HTML code to an .html file. For example, \section{Introduction} is 

interpreted as Introduction for the output. 

This means that L AT E X2HTML is essentially duplicating the L AT E X processing, 

an enormous undertaking. Since the document classes contain 

varying commands or have common ones behaving differently, there must 

be Perl scripts for each one (article.perl, and so on) to program these 

commands properly. Additional packages loaded with \usepackage can 

also define new commands, or alter the functionality of existing ones; 

the translator must be informed about these by means of corresponding 

Perl scripts. For example, the natbib package described in Section 9.3.4 

defines citation commands \citet and \citep, which need to be made 

available to L AT E X2HTML in a file natbib.perl. Most of the tools packages 

of Section B.5.4 are included as Perl scripts, as are many other popular 

contributed packages. In other words, the entire L AT E X2HTML installation 

must mirror the L AT E X one. 

This process is somewhat simplified by the fact that both formats 

are markup languages written in pure typewriter text. However, this 

similarity soon reaches its limits with the many L AT E X features not available 

in HTML, such as complex math, included figures, cross-referencing. In 

the other direction, HTML exhibits hyperlinks, both internally and to 

external documents, something not provided by normal L AT E X. 

L AT E X2HTML attempts to reproduce math with the limited HTML possibilities, 

but failing that, formulas are handled the same as figures and 

other environments that cannot be directly rendered: they are converted 

to GIF image files along the route.tex →.dvi →.ps →.pbm →.gif, which 

are then included in the HTML file as in-line images. 

Cross-references and citations are automatically provided with hyperlinks 

to their targets; the keyword index also links back to the text. Other 

features of HTML can be included by means of special commands defined 

in the html.sty package. For example: 

• explicit hyperlinks, internal and external

E.1. Converting to HTML 477 

• conditional text for the HTML or L AT E X versions only 

• raw HTML code 

• segmentation of the HTML output 

• finer image control 

• . . . 

This package thus provides the L AT E X writer with a more convenient interface 

for producing hypertext documents; furthermore, the same document 

can still be processed with L AT E X to generate the ‘paper’ version at 

any time. The conditional text mentioned above allows the electronic and 

paper versions to exhibit some differences. 

L AT E X2HTML is available on the CTAN servers (Figure B.4 on page 390) 

under the support directory. It can also be downloaded from the 

L AT E X2HTML home page http://cbl.leeds.ac.uk/nikos/tex2html/ 

doc/latex2html/latex2html.html, where more information as well as 

a manual can be obtained. 

E.1.2 The T E X4ht program 

An alternative program is that by Eitan M. Gurari, called T E X4ht. It produces 

much the same results as L AT E X2HTML, but by a different route. 

Rather than trying to interpret the L AT E X input text itself, it processes 

instead the.dvi file to extract the HTML output. It is therefore applicable 

both to L AT E X and Plain T E X, as its name implies. 

In reality, it is not as simple as that. Since the DVI output is normally 

only a list of typesetting instructions, containing no logical markup information 

at all, it is necessary to process the L AT E X input with the matching 

TeX4ht.sty package that redefines all the regular L AT E X commands to 

write appropriate code to the.dvi file. This is accomplished by means of 

the \special command, a basic T E X command for writing instructions to 

a particular DVI driver. In this case the driver is the T E X4ht program; no 

other driver will be able to process this.dvi file. 

The T E X4ht processor writes the HTML output according to various 

options specified in the L AT E X input. For example, the HTML output can be 

segmented by chapters or sections, tables of contents may be added to 

each segment, links are automatically established for cross-references. As 

with L AT E X2HTML, manual links to external documents or internal points 

can be made, raw HTML coding included, explicit images included. It 

is also possible to turn off the HTML conversion in order to produce a 

normal .dvi file; conditional input for whether HTML is on or off is also 

available. 

Any symbols, tables, equations, or imported figures that cannot be 

directly rendered in HTML are written to a new DVI file, with extension


.idv, one item per page. Finally, a log file is written containing conversion 

instructions for each such object; this file is so constructed as to act as 

a script, or batch job, for creating GIF images. Just how this conversion 

is realized and how the instructions are formatted depend on the local 

installation and can be configured by the user. However, the usual route is 

something like.idv →.ps (dvips) →.ppm (Ghostscript) →.gif (pbm). An 

example of T E X4ht output with a GIF representation of a math equation 

can be seen in Figure E.1 on page 482. 

Since T E X4ht needs to redefine all the L AT E X markup commands, one 

could argue that it too is duplicating L AT E X just like L AT E X2HTML. To some 

extent this is true, although there is a very important difference: it is only 

the markup, not the formatting commands, that needs to be massaged. 

Most classes and additional packages will need no extra configuring to 

be processed by T E X4ht. A package like natbib which adds new citation 

commands \citet and \citep will totally confuse L AT E X2HTML without 

the additional Perl script natbib.perl, whereas under T E X4ht these commands 

will produce the correct output text in the HTML file without any 

additional definitions. (However, the automatic links between the citations 

and list of references will be missing, since this is a markup feature, 

but the printed text will be all right.) 

T E X4ht is available from the T E XLive CD, or from http://www.cis. 

ohio-state.edu/˜gurari/TeX4ht/mn.html. 

L AT E X2HTML and T E X4ht are only two examples of procedures for creating 

HTML output from L AT E X. In both cases, one should not try to convert 

arbitrary L AT E X files, but should view them as means of producing normal 

L AT E X and HTML output from a common source file. This source file must 

be constructed accordingly, possibly with conditional texts for the two 

outputs, with manual hyperlinks, formatting options for HTML, and so 

on. 

The use of GIF bitmap images for rendering items not readily available 

in HTML, while producing the visual features on screen, does not allow 

these objects to be searched for or otherwise further processed. Such 

images also do not participate in automatic font size changes and other 

Web manipulations. 

E.2 The Extensible Markup Language: XML 

It has long been recognized that HTML has many severe limitations. Its 

simplicity may have been responsible for its rapid rise in popularity, but 

that simplicity also restricts it to the type of documents for which it was 

originally intended. Today the Web is offering much more than at the 

beginning, with databases, search engines, data verification. HTML is not 

programmable, so that browser suppliers have found it necessary to add 

additional features of their own, which only function with their browser.

E.2. The Extensible Markup Language: XML 479 

Standardization is thus destroyed. 

What is needed is a means of adding new features to an HTML file 

so that the browser can be informed of what they mean. This would be 

like adding new features to a L AT E X file by means of a package. It is to 

this end that the Web Consortium began the development in 1996 of the 

Extensible Markup Language, or XML. 

However, XML is not so much a language as a specification for defining 

languages in a standard manner. An XML interpreter must be able 

to determine whether an XML document conforms to the general rules 

(mainly that all tags are properly closed), in which case it is considered 

to be well-formed, and to the specific syntax of the tags that it uses (as 

defined in a DTD, Document Type Definition), in which case it is also valid. 

There remains the question of how those tags are to be rendered (how 

are sections formatted, what to do with cross references, how to include 

graphics), which is relegated to style sheets documented in the Extensible 

Stylesheet Language, XSL. Here we have true separation between content 

and format, between logical and typographical markup. 

If all this sounds overwhelming, it is! Furthermore, the DTDs are 

considered to be too cumbersome and various alternatives have been 

developed, themselves conforming to XML (which the DTDs do not). This 

whole area is in transition, but the rewards will be high when a reliable 

standard for electronic data exchange is achieved. Data in this sense is 

very generic, and includes textual documents. We recommend the books 

by (Marchal, 2000) and (Williamson, 2001); or one can check the Web 

Consortium home page http://www.w3.org/. 

E.2.1 XML and L AT E X 

What does all this have to do with L AT E X? Just as the similarity between 

L AT E X and HTML allow relative easy conversion, so does the essential 

similarity with XML. The T E X4ht system has already done the real work of 

converting L AT E X structure into HTML tags; it is therefore simply a matter 

of changing the output for each of those tags, something that is put into 

corresponding configuration files. 

Since XML itself is not a language with defined tags, how do we know 

what those tags are to be? Here we rely on existing DTDs for textual 

documents. These can be the DocBook DTD developed for software user 

manuals and other computer documentation, or the TEI DTD, the Text 

Encoding Initiative meant essentially for general text, including images 

and sounds. 

The T E X4ht collection on the T E XLive CD includes configurations for 

producing XML output from L AT E X for both of these DTDs. The preamble 

to the L AT E X source file should then include 

\usepackage[html]{tex4ht} for HTML output


\usepackage[xhtml,docbook]{tex4ht} for XML, DocBook output 

\usepackage[xhtml,tei]{tex4ht} for XML, TEI output 

One might ask why does the author not write an XML file directly as 

the source document? First, one might very well want to convert older 

documents recorded in L AT E X to XML. Secondly, one might find it simpler 

to write in L AT E X (not everyone agrees) since no one would write an XML 

file by hand (everyone agrees) but would rely on some application to do it 

behind the scenes. Certainly L AT E X source text is easier to read, even when 

intersperse with commands, than an XML file. 

E.2.2 MathML and L AT E X 

One of the strengths of T E X and thus L AT E X over other text processing 

systems is the ability to handle mathematics in a way that is acceptable to 

mathematicians. HTML has never been able to even begin to compete in 

this area. The only way conversion programs like L AT E X2HTML and T E X4ht 

can render equations like that in Figure E.1 is to convert it to an image 

file. This is hardly a satisfactory solution. 

The MathML project attempts to define an XML language to encode 

mathematics. We refer to chapter 8 of the L AT E X Web Companion (Goossens 

and Rahtz, 1999) for a description. T E X4ht can produce output in MathML 

by specifying 

\usepackage[xhtml,mathml]{tex4ht} 

Symbols that do not exists in the browser fonts will still be converted to 

image files, but instructions are inserted to carry out the placement for 

fractions and superscripts and subscripts. Unfortunately, most browsers 

do not recognize MathML (yet). Again, once this becomes established 

there will be a standard for recording and exchanging documents with 

complex mathematics. 

E.2.3 Rendering XML with T E X/L AT E X 

Another role that L AT E X, or more properly the T E X program, can play in the 

XML world, is to translate it to a form that can be viewed or printed as a 

finished document. This is known as rendering. 

David Carlisle has written macros in a file xmltex.tex to act as an 

XML parser for output in T E X or L AT E X. Sebastian Rahtz has written an 

additional macro package PassiveT E X, to incorporate an XSL style sheet 

for the TEI DTD into xmltex. With this combination, one can produce DVI 

or PDF output from a TEI-coded XML file. 

There are a number of ways of going about this. The simplest is to 

write a ‘wrapper’ file for processing document.xml as

E.3. The techexplorer Hypermedia Browser 481 

\def\xmlfile{document.xml} 

\input xmltex.tex 

and then to process this with either L AT E X or pdfL AT E X. The PassiveT E X 

macros will be found automatically, if they are on the system. 

Another means is to generate an xmltex format, as described in Section 

B.1.3. This would be done with 

initex &latex xmltex.tex or 

tex -ini &latex xmltex.tex 

to generate a format named xmltex.fmt. One can then invoke it with 

tex &xmltex document.xml 

or simply with xmltex document.xml if an alias or batch file for the 

command xmltex has been defined to equate to the real command. 

Instead of using the T E X program, one can also make use of pdfT E X 

instead, to generate PDF instead of DVI output. 

E.3 The techexplorer Hypermedia Browser 

A completely different approach to Web viewing of T E X and L AT E X documents 

is offered by the IBM techexplorer Hypermedia Browser, a plug-in 

for Netscape Navigator and Microsoft Internet Explorer under Windows. 

When properly installed, it is launched automatically within the parent 

browser window when one opens a file with one of the extensions .tex, 

.bbl,.ltx,.latex, or.tcx. 

The techexplorer interprets the T E X and L AT E X commands directly 

and displays the results on the monitor. The original page height and 

line width are ignored, the paragraphs being fitted to the size of the window 

as is normal for any browser. Fonts, tables, environments, but most 

importantly mathematics are reproduced very well. Figure E.1 on the 

next page demonstrates a single formula represented by both T E X4ht and 

techexplorer. Even user commands defined with \def and \newcommand 

are recognized, provided they occur within the same file. However, no 

external packages may be loaded, nor are the class files accepted. (Both 

\usepackage and \documentclass are simply ignored.) 

Hypertext features can be included with additional T E X-like commands 

that establish internal and external links, pop-up menus, multiple-file 

documents (segmentation) with automatic links between them, and inclusion 

of images. In other words, a techexplorer document can have 

the full range of hypertext features of HTML. Here one must make the 

same comment as for L AT E X2HTML and T E X4ht: it is not primarily intended 

for viewing arbitrary L AT E X documents, but rather for preparing hypertext 

Web pages by means of T E X or L AT E X. The only problem here is that most


Figure E.1: Example of a math formula represented by T E X4ht (as a GIF 

image of the original L AT E X output) and by techexplorer rendering. 

Internet users will probably not have techexplorer installed, so that 

they will only be presented with the input text, and not its rendering. 

One additional application of techexplorer which is not available 

to the other programs is to embed T E X commands, especially mathematics, 

inside a regular HTML file. If the commands are stored in a file 

formula.tex, they could be embedded with 

 

or they can be entered directly as 

 

The result, shown in the lower part of Figure E.1, is a very acceptable 

display of the math formula in the middle of an HTML file. (This file was 

generated by T E X4ht from a L AT E X file, with the above techexplorer text 

included as raw HTML code.) 

The techexplorer is obtainable from http://www.software.ibm. 

com/enetwork/techexplorer. It comes in a free introductory version, 

as well as in a professional version for a nominal charge.

F 

Obsolete LAT E X 

Since 1994, L AT E X 2ε has replaced L AT E X 2.09 as the official version. Although 

the newer version is backward compatible with the previous one, 

this is only to allow older source files written for L AT E X 2.09 to be processed 

under L AT E X 2ε; it is not intended that newer files should make use of the 

obsolete syntax. 

Human nature being conservative as it is, many authors continued to 

use the commands they were familiar with, even today. The purpose of 

this chapter is to explain those commands for users who may come across 

such legacy documents, and to indicate to traditional L AT E X users which 

commands they should no longer be employing. It is not meant to be a 

guide for using L AT E X 2.09! 

At first sight, the differences seem very slight indeed. In fact, the 

real changes are mainly internal, allowing packages to be handled more 

systematically and fonts to be dealt with in a more flexible manner. The 

advantages of L AT E X 2ε for class and package writers are considerable, as 

explained in Appendix D. Regular users benefit indirectly by being able 

to employ these extensions, and by being able to activate fonts other than 

the original Computer Modern families. 

F.1 The 2.09 preamble 

As for L AT E X 2ε, the document preamble in L AT E X 2.09 contains overall 

specifications for the entire document, including general layout. 

F.1.1 Style instead of class 

L AT E X 2.09 works with style rather than class files, so the first command in 

the source file is 

\documentstyle[options]{style} 

483

484 Appendix F. Obsolete L AT E X 

rather than \documentclass. Possible values for style are article, 

report, book, or any other main style files that might exist locally. These 

styles may have various options associated with them, like 11pt, twoside, 

and so on, which can be listed in the set of options. Not all the options 

available with the L AT E X 2ε class files are recognized by the style files. 

The options list has an additional function in L AT E X 2.09: for any option 

in the list that is not recognized by the style file, a file with that name and 

extension.sty is loaded, if it exists. The original idea of this mechanism 

was to allow certain options applicable to all styles to have the common 

coding stored in an extra file, rather than repeating it within each main 

style file. It was this that later led to the concept of packages. 

There is no \usepackage command in L AT E X 2.09; packages can only 

be loaded with the options list, and no options may be specified for the 

packages. 

To start a document with the article style and 12pt option, with the 

parskip and makeidx packages, one gives 

\documentstyle[12pt,parskip,makeidx]{article} 

F.1.2 Compatibility mode 

L AT E X 2ε is designed to be able to process older L AT E X documents with 

exactly the same output as with L AT E X 2.09. To do this, it recognizes the 

\documentstyle command which switches it into compatibility mode, 

changing the functionality of many L AT E X 2ε features. For example, the 

\usepackage command becomes inoperable, issuing an error message. 

With a true L AT E X 2.09 installation, the above \documentstyle example 

would load the files article.sty, parskip.sty, and makeidx.sty. 

Under L AT E X 2ε, in compatibility mode, it first tries to load the main style 

as a class file with the extension .cls, in this case article.cls, but if 

that fails, it then looks for the extension.sty. This is to handle local main 

style files that have no class file equivalent. The packages parskip.sty 

and makeidx.sty are loaded as well. However, they too may recognize 

the compatibility mode and behave differently as they would with regular 

L AT E X 2ε. 

F.2 Font selection 

The other major difference between the two versions is font selection. 

The New Font Selection Scheme (NFSS) and font attributes described in 

Section 4.1.3 are missing in L AT E X 2.09, as are the font commands \textbf, 

\texttt, and so on, and the font attribute declarations such as \bfseries 

and \ttfamily. (They are still available even in L AT E X 2ε compatibility 

mode.)

F.2.1 Old font declarations 

F.3. Obsolete means obsolete 485 

In L AT E X 2.09, only the two-letter font declarations that originally came 

from Plain T E X are possible. 

\rm Roman \it Italic \sc Small Caps 

\bf Bold face \sl Slanted \sf Sans Serif 

\tt Typewriter 

These are all declarations, changing the font until another font declaration 

is given, or until the current environment is ended. They are normally 

used as {\bf bold face} to yield bold face. 

The emphasizing declaration \em also belongs to this group, but it has 

been taken over as part of L AT E X 2ε proper. 

These old declarations have a different behavior from that of their 

NFSS counterparts: they rigidly select a particular font instead of altering 

only one attribute, but retaining the current size. Compare: 

{\sl slanted {\bf bold}} ⇒ slanted bold 

{\slshape slanted {\bfseries bold}} ⇒ slanted bold 

It is not possible to obtain slanted or italic bold face with L AT E X 2.09 even 

though such fonts are available. 

These two-letter declarations are also retained in the standard L AT E X 2ε 

classes, and not just in compatibility mode. Note carefully that wording: 

they are not part of L AT E X 2ε itself, but are only included as part of the 

standard class files. This means other class files might not provide them, 

or that they might even be removed in some later version (although this 

is most unlikely). Therefore, their use is not to be encouraged at all. 

F.2.2 Font size declarations 

In L AT E X 2.09, the font size declarations also reset all the other font attributes 

(Section 4.1.3) to their defaults, that is, Roman, upright, medium 

weight. By contrast, in L AT E X 2ε, these attributes remain unchanged. Compare 

the results of {\sl slanted {\Large larger}}: 

L AT E X 2.09: slanted larger 

L AT E X 2ε: slanted larger 

In L AT E X 2ε compatibility mode, the size declarations behave as they do in 

L AT E X 2.09. 

F.3 Obsolete means obsolete 

Many features have been added to L AT E X 2ε, and a number of commands 

have been given extended syntax, in the form of additional optional arguments. 

We have no intention of indicating them anymore in this book as

486 Appendix F. Obsolete L AT E X 

this is a manual for the newer version only. (Previous editions did make 

a distinction.) 

Apart from the restrictions pointed out in the above sections, all of 

the L AT E X 2ε features will work in compatibility mode. Nevertheless, we 

stress once more, compatibility mode is only intended for the processing 

of older source files. Every effort has been undertaken to ensure that the 

results are identical to those of a processing with the true L AT E X 2.09 itself. 

Compatibility mode (\documentstyle in place of \documentclass) 

issues a very strong warning: 

Entering LaTeX 2.09 COMPATIBILITY MODE 

************************************************************* 

!!WARNING!! !!WARNING!! !!WARNING!! !!WARNING!! 

This mode attempts to provide an emulation of the LaTeX 2.09 

author environment so that OLD documents can be successfully 

processed. It should NOT be used for NEW documents! 

New documents should use Standard LaTeX conventions and start 

with the \documentclass command. 

Compatibility mode is UNLIKELY TO WORK with LaTeX 2.09 style 

files that change any internal macros, especially not with 

those that change the FONT SELECTION or OUTPUT ROUTINES. 

Therefore such style files MUST BE UPDATED to use 

Current Standard LaTeX: LaTeX2e. 

If you suspect that you may be using such a style file, which 

is probably very, very old by now, then you should attempt to 

get it updated by sending a copy of this error message to the 

author of that file. 

************************************************************* 

Take this message seriously, and stick to \documentclass.

G 

T E X Fonts 

Computers work exclusively with numbers, or more precisely, only with 

bits which may be interpreted as numbers. They do not know the difference 

between the letter A and an apple, or that either even exists. For text 

processing, all symbols, both input and output, need to be represented 

as numbers somehow. The association between number and symbol is 

called the encoding or layout. The latter term derives from the common 

method of illustrating the encoding in the form of a table. 

Encoding tables are by no means standard. The standard ascii scheme 

is just one of several, and it is limited to 7 bits, or 128 characters. There 

are 8-bit (256 characters) versions as well, in fact, a large number exist, 

tailored to different computer systems and languages. The question of 

coding the input for L AT E X documents with more than 7 bits are addressed 

in Sections 2.5.9 and D.5. 

In this appendix, we look at how the underlying T E X program deals 

with output fonts, their nomenclature, and their encoding tables. 

G.1 Font metrics and bitmaps 

When T E X decides to output character nn in a particular font, all it needs 

to know is how much room to leave for it. T E X does not care what the 

character looks like, for that is the task of the DVI driver afterwards. 

Information about the characters in each font are stored in various 

files, all bearing the root name of the font but with different extensions. 

This information is divided into files that contain only the sizes of each 

symbol, and those with the actual drawing or image data. 

.tfm T E X font metric files are the only font files read in by T E X itself. 

They contain the sizes of the characters, such as width, height, and 

depth. For slanted fonts, they also possess the ‘italic correction’ 

for each letter. Furthermore, they specify for which letter combinations 

a different spacing is required, such as AV instead of AV, 

487

488 Appendix G. T E X Fonts 

or for which a ligature is available. Finally, the .tfm files provide 

information about the slope of the characters (zero for an unslanted 

font), the standard word spacing and its stretch and shrinkage, the 

width of the em and quad spacings, and the spacings at the end 

of a sentence. Mathematical and symbol fonts require even more 

information which is also included in their.tfm files. 

.pk Compressed pixel files contain bitmaps (images) of the symbols in 

each font, in one size and one resolution. The DVI driver program 

uses these to send the output to a printer or previewer. These files 

are normally generated automatically by the METAFONT program 

(Section G.3) as needed. 

Since these files might exist several times for a given font, in difference 

sizes or resolutions, many installations distinguish them 

by adding the resolution to the extension, as cmr10.300pk or 

cmr10.600pk for 300 and 600 dpi versions, respectively. On systems 

that cannot support extensions with more than three characters, 

both files are named cmr10.pk but are stored in directories 

dpi300 and dpi600. 

Font magnification is achieved by selecting an appropriate resolution. 

If cmr10 is to be printed on a 300 dpi printer at double its 

normal size, the 600 dpi version is used instead. 

.mf METAFONT source files contain the drawing instructions for each 

of the fonts. These are not read by the DVI driver but rather by the 

METAFONT program which converts them into .pk bitmap files at 

the required size and resolution. 

.vf Virtual font files are an alternative to .pk files. Instead of having 

bitmaps for each character, they contain instructions that refer to 

characters in different, real fonts, or that tell the driver to draw a 

black box and issue a warning message (that the character does not 

exist, for example). 

Virtual fonts are normally associated with PostScript fonts and are 

therefore described in Section G.5. However, there is no reason for 

them not to be used with bitmap fonts as well. 

G.2 Computer Modern fonts 

When Donald E. Knuth invented the T E X program, he also provided it with 

an extensive set of character fonts, which he named Computer Modern, 

rather than relying on the fonts available on any given printer. At that 

time, the printer fonts were not so good, and certainly were not uniform.

G.2. Computer Modern fonts 489 

With the supplied fonts, T E X could produce identical, high-quality results 

on all printers. 

L AT E X, of course, has inherited these fonts, so that they have almost 

become a trademark for documents produced by T E X or L AT E X. This is not 

really necessary, for L AT E X need not be married to any particular set of 

fonts, especially with the New Font Selection Scheme (Appendix A) which 

simplifies font installation enormously. The main fonts used in this book, 

for example, are Lucida Bright, Lucida Sans, and Lucida Sans Typewriter, 

designed by Bigelow & Holmes and distributed by Y&Y Inc. 

G.2.1 Font families 

Typography is the study and classification of typefaces, something that 

goes back to Gutenberg’s invention of movable type (not of the printing 

press, which was invented by the Chinese) five and a half centuries ago. 

Since that time, many families of fonts have been created, bearing classical 

names like Baskerville, Garamond, Univers, etc. Each member of such a 

family has the same overall design, or basic look, but vary by being slanted, 

italic, bold, or thin; and of course, they come in different sizes. 

Font families are classified according to certain criteria that often 

determine to what use they will be put. 

Serif fonts: are those that have little horizontal lines, or serifs, at their 

edges, to guide the eye better. Experience has shown these to be 

best for general reading, and so they are regularly used for the main 

body of text. The NFSS terminology refers to these as Roman fonts. 

Sans serif fonts: are those that are lacking any serifs. Such fonts with 

their starker appearance are often employed for titling or headlining. 

Compare sans serif with regular text. 

Fixed fonts: are those with a uniform letter width, something that has 

evolved from the typewriter and has been carried over to computer 

listings. Classically, such fonts have no business in book printing, 

where proportional fonts (the letter i is narrower than m) have always 

dominated. 

Decorative fonts: are ones that stand out because of some unusual characteristic. 

They are intended to catch attention and to attract the 

eye. Such families are not complete, with a full range of shapes and 

widths, and are employed mostly by advertising. 

Mathematical fonts: are collections of special symbols needed for mathematical 

works. Their further classification cannot be compared to 

that of text fonts at all.


A book designer must decide what kinds of typeface families are 

needed. He or she might decide on a Roman (serifed) font for the main 

body, a sans serif one for headlines, and then, if it is a book containing 

computer code, select a fixed font for setting those parts. Finally, symbol 

fonts will be needed if the work contains mathematical sections. 

The Computer Modern set of fonts provides all these classes of families. 

However, since they were produced before the NFSS attribute system 

of classification was established, the CM font nomenclature does not conform 

perfectly with this scheme. The NFSS system frees the user from 

having to think about the CM font names, for specifying the attributes is 

sufficient. Of course, the NFSS installation must provide the font definition 

.fd files which translate any set of attributes into a real font name, 

or into some acceptable substitute. 

G.2.2 Classification of CM fonts 

All the T E X font names begin with the letters cm, which stands for ‘Computer 

Modern’, followed by one to four letters describing the style of font, 

and finally one or two digits specifying the design size in points. This is 

the root name of the font which must be given in the \newfont command 

(Section 4.1.5) if it is to be activated directly. 

The CM fonts can be classified as text, math, decorative, or other 

special symbol fonts, as described in the rest of this section. Each type of 

font has an encoding scheme shown in the accompanying layout tables. 

Text fonts 

The Computer Modern fonts appropriate for straightforward text can be 

classified into the three families Roman, sans serif, and typewriter. Within 

each of these families there are upright, slanted, italic, small caps, and 

bold variants. 

Table G.1 presents the names of these fonts together with the NFSS 

family/series/shape assignments from Appendix A. They all have the 

same coding (OT1). The * in each name is the design size specification 

which takes on values of 5, 6, 7, 8, 9, 10, 12, 17. Only cmr* is available 

in all these sizes, while some are only to be found in size 10. (This table 

contains much the same information as in Table A.3 on page 370 but in a 

different form.) 

One sees several apparent inconsistencies in these font names, such 

as cmti* for the italic upright font. Why is this not cmri*? The answer 

is that this is a text italic font, as opposed to the math italic font cmmi* 

described below. These inconsistences have arisen because the CM fonts 

were not created with the NFSS classification system in mind, since that 

was established a decade later.


Table G.1: Computer Modern text fonts 

Roman Sans serif Typewriter 

Family= cmr cmss cmtt 

Style Series/shape 

Upright m/n cmr* cmss* cmtt* 

” slanted m/sl cmsl* cmssi* cmsltt* 

” italic m/it cmti* — cmitt* 

” small caps m/sc cmcsc* — cmtcsc* 

Bold b/m cmb* — — 

Bold extended bx/m cmbx* cmssbx* — 

” ” slanted bx/sl cmbxsl* — — 

” ” italic bx/it cmbxti* — — 

Further inconsistencies exist in the encoding schemes for the CM text 

fonts. They are all nominally OT1, but there are slight deviations among 

them. The proper OT1 encoding is displayed in Layout 1 on the next page 

for font cmr10. The text italic fonts are identical to the upright ones, 

except that the dollar sign in position 36 is replaced by the pound sign 

(Layout 3 for cmti10). The slanted fonts, on the other hand, are exactly 

the same as the upright ones. Other deviations are to be found in the 

small caps (Layout 2) and all the typewriter fonts (Layout 4). 

Math fonts 

The Computer Modern fonts provide three types of mathematical fonts, 

each with its own encoding scheme. 

The mathematical italic fonts cmmi* contain Latin and Greek letters, 

in upper and lower case, plus a number of extra symbols. Since variable 

names in formulas are set in italics, these are basically italic fonts. The 

encoding scheme is designated OML, for old math letters, and is displayed 

in Layout 5. It is available in design sizes 5–12 pt in normal weight, but 

only in 10 pt size in bold face, cmmib10. These fonts are used for the 

letters math alphabet of Section A.3.4. 

Symbol fonts cmsy* provide the rest of the math symbols for formulas, 

except for those that appear in variable sizes. The encoding scheme is 

named OMS for old math symbols and is shown in Layout 6. It comes in 

sizes 5–10 pt in normal weight, and in 10 pt size in bold face, cmbsy10. 

The symbols math alphabet uses these fonts. 

Variable sized symbols are to be found in the font cmex10, with encoding 

OMX, for old math extension. Layout 7 illustrates this set. There is no 

bold version of this font. It belongs to the largesymbols math alphabet 

of Section A.3.4.


0 1 2 3 4 5 6 7 

’00x Γ 0 ∆ 1 Θ 2 Λ 3 Ξ 4 Π 5 Σ 6 Υ 7 

’01x Φ 8 Ψ 9 Ω 10 ff 11 fi 12 fl 13 ffi 14 ffl 15 

’02x ı 16 j 17 ` 18 ´ 19 ˇ 20 ˘ 21 ¯ 22 ˚ 23 

’03x ¸ 24 ß 25 æ 26 œ 27 ø 28 Æ 29 Œ 30 Ø 31 

’04x ̷ 32 ! 33 ” 34 # 35 $ 36 % 37 & 38 ’ 39 

’05x ( 40 ) 41 * 42 + 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 ¡ 60 = 61 ¿ 62 ? 63 

’10x @ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 “ 92 ] 93 ˆ 94 ˙ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 – 123 —124 ˝ 125 ˜ 126 ¨ 127 

Font Layout 1: The character font cmr10. This is the standard character 

assignment for the OT1 encoding scheme. 

0 1 2 3 4 5 6 7 

’00x Γ 0 ∆ 1 Θ 2 Λ 3 Ξ 4 Π 5 Σ 6 Υ 7 

’01x Φ 8 Ψ 9 Ω 10 ↑ 11 ↓ 12 ' 13 ¡ 14 ¿ 15 

’02x ı 16 j 17 ` 18 ´ 19 ˇ 20 ˘ 21 ¯ 22 ˚ 23 

’03x ¸ 24 ß 25 æ 26 œ 27 ø 28 Æ 29 Œ 30 Ø 31 

’04x ̷ 32 ! 33 ” 34 # 35 $ 36 % 37 & 38 ’ 39 

’05x ( 40 ) 41 * 42 + 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 < 60 = 61 > 62 ? 63 

’10x @ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 “ 92 ] 93 ˆ 94 ˙ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 – 123 —124 ˝ 125 ˜ 126 ¨ 127 

Font Layout 2: The character font cmcsc10. The differences from Layout 

1 are symbols 11–15, 25, 60, and 62. The ligatures that normally 

appear in 11–15 have been replaced by some extra symbols.


0 1 2 3 4 5 6 7 

’00x Γ 0 ∆ 1 Θ 2 Λ 3 Ξ 4 Π 5 Σ 6 Υ 7 

’01x Φ 8 Ψ 9 Ω 10 ff 11 fi 12 fl 13 ffi 14 ffl 15 

’02x ı 16 j 17 ` 18 ´ 19 ˇ 20 ˘ 21 ¯ 22 ˚ 23 

’03x ¸ 24 ß 25 æ 26 œ 27 ø 28 Æ 29 Œ 30 Ø 31 

’04x ̷ 32 ! 33 ” 34 # 35 £ 36 % 37 & 38 ’ 39 

’05x ( 40 ) 41 * 42 + 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 ¡ 60 = 61 ¿ 62 ? 63 

’10x @ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 “ 92 ] 93 ˆ 94 ˙ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 – 123 —124 ˝ 125 ˜ 126 ¨ 127 

Font Layout 3: The character font cmti10. The only difference from 

Layout 1 is symbol 36 (£ instead of $). All other text italic fonts 

have this same pattern. 

0 1 2 3 4 5 6 7 

’00x Γ 0 ∆ 1 Θ 2 Λ 3 Ξ 4 Π 5 Σ 6 Υ 7 

’01x Φ 8 Ψ 9 Ω 10 ↑ 11 ↓ 12 ' 13 ¡ 14 ¿ 15 

’02x ı 16 j 17 ` 18 ´ 19 ˇ 20 ˘ 21 ¯ 22 ˚ 23 

’03x ¸ 24 ß 25 æ 26 œ 27 ø 28 Æ 29 Œ 30 Ø 31 

’04x 32 ! 33 " 34 # 35 $ 36 % 37 & 38 ’ 39 

’05x ( 40 ) 41 * 42 + 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 < 60 = 61 > 62 ? 63 

’10x @ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 \ 92 ] 93 ^ 94 _ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 { 123 | 124 } 125 ~ 126 ¨ 127 

Font Layout 4: The character font cmtt10. All tt fonts are set up with 

the same pattern. The differences from Layout 1 lie in symbols 11– 

15, 60, 62, 92, 123, 124, and 125. Furthermore, the italic typewriter 

fonts also have £ in place of $.


Package: 

latexsym 

amsfonts 

The term old in these encoding names refers to the original encoding 

as introduced by Donald Knuth, in the same way that the text encoding 

OT1 indicated the old, or original, text encoding with 128 characters per 

font. The new encoding schemes contain 256 characters. For text, the 

encoding is named T1 (Section G.4.3); such extended math fonts do not 

yet exist. 

The extra A M S fonts in Section 12.4.1 add additional bold face sizes 

for cmmib* and cmbsy* as well as some more sizes for cmex*. 

Decorative fonts 

Three families of decorative or special fonts are available in a single design 

size and limited attributes. They all exhibit OT1 encoding. 

cmfr Funny Roman family, consists of two fonts, cmff10 leaning to the left 

and cmfi10 which leans to the right. Both are series m with shapes n 

and it respectively. 

cmfib Fibonacci family, contains one font cmfib8 derived from the Fibonacci 

series of numbers. It has series m and shape n. 

cmdunh Dunhill family, with one font cmdunh10 with series m and shape 

n. This font is demonstrated in the sample letterhead on page 364. 

Logo fonts 

The fonts logo8, logo9, logo10, logosl10, and logobf10 contain only 

the nine letters A, E, F, M, N, O, P, S, and T, for generating the logos 

METAFONT METAFONT METAFONT 

METAPOST METAPOST METAPOST 

The L AT E X lasy fonts 

As an extension of the cmsy* fonts for mathematical symbols, L AT E X pro- 

vides some additional symbols with the font lasy* in design sizes 5–10 pt. 

It contains the 15 symbols: ✭ ✮ ✯ ✰ ✁ ✂ ✄ ☎ ✵ ✶ ✷ ✸ ❀ ❁ ❂. 

These symbols are not defined unless one of the packages latexsym 

or amsfonts has been loaded. 

Fonts for making pictures 

The special picture elements for use in the L AT E X picture environment are 

stored in fonts named line10, lcircle10, linew10, and lcirclew10. 

The first two are used for lines, ovals, and circles when \thinlines is 

in effect. The sloping lines and arrow heads are to be found in line10


0 1 2 3 4 5 6 7 

’00x Γ 0 ∆ 1 Θ 2 Λ 3 Ξ 4 Π 5 Σ 6 Υ 7 

’01x Φ 8 Ψ 9 Ω 10 α 11 β 12 γ 13 δ 14 ɛ 15 

’02x ζ 16 η 17 θ 18 ι 19 κ 20 λ 21 µ 22 ν 23 

’03x ξ 24 π 25 ρ 26 σ 27 τ 28 υ 29 φ 30 χ 31 

’04x ψ 32 ω 33 ε 34 ϑ 35 ϖ 36 ϱ 37 ς 38 ϕ 39 

’05x ↼ 40 ↽ 41 ⇀ 42 ⇁ 43 ↩ 44 ↪ 45 ⊲ 46 ⊳ 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 . 58 , 59 < 60 / 61 > 62 ⋆ 63 

’10x ∂ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 ♭ 91 ♮ 92 ♯ 93 ⌣ 94 ⌢ 95 

’14x ℓ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 ı 123 j 124 ℘ 125 126 127 

Font Layout 5: The font cmmi10. This corresponds to the OML encoding 

scheme, as does cmmib10. It contains lower case Greek letters in 

positions 11–39 and math symbols in 40–47, 60–64, and 123–127. 

0 1 2 3 4 5 6 7 

’00x − 0 · 1 × 2 ∗ 3 ÷ 4 ⋄ 5 ± 6 ∓ 7 

’01x ⊕ 8 ⊖ 9 ⊗ 10 ⊘ 11 ⊙ 12 ○ 13 ◦ 14 • 15 

’02x ≍ 16 ≡ 17 ⊆ 18 ⊇ 19 ≤ 20 ≥ 21 22 23 

’03x ∼ 24 ≈ 25 ⊂ 26 ⊃ 27 ≪ 28 ≫ 29 ≺ 30 ≻ 31 

’04x ← 32 → 33 ↑ 34 ↓ 35 ↔ 36 ↗ 37 ↘ 38 39 

’05x ⇐ 40 ⇒ 41 ⇑ 42 ⇓ 43 ⇔ 44 ↖ 45 ↙ 46 ∝ 47 

’06x ′ 48 ∞ 49 ∈ 50 ∋ 51 △ 52 ▽ 53 54 ↦ 55 

’07x ∀ 56 ∃ 57 ¬ 58 ∅ 59 ℜ 60 ℑ 61 ⊤ 62 ⊥ 63 

’10x ℵ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 ∪ 91 ∩ 92 ⊎ 93 ∧ 94 ∨ 95 

’14x ⊢ 96 ⊣ 97 ⌊ 98 ⌋ 99 ⌈ 100 ⌉ 101 { 102 } 103 

’15x 〈 104 〉 105 | 106 107 ↕ 108 ⇕ 109 \ 110 ≀ 111 

’16x √ 112 ∐ 113 ∇114 ∫ 115 ⊔ 116 ⊓ 117 ⊑ 118 ⊒ 119 

’17x § 120 † 121 ‡ 122 123 ♣ 124 ♦ 125 ♥ 126 ♠ 127 

Font Layout 6: The font cmsy10. The cmsy fonts, adhering to the OMS 

encoding scheme, contain many math symbols as well as the calligraphic 

letters A . . . Z in positions 65–90. The bold face version 

cmbsy10 is set up on the same pattern.


0 1 2 3 4 5 6 7 

’00x 0 1 2 3 4 5 6 7 

’01x 8 9 10 11 12 13 14 15 

’02x 16 17 18 19 20 21 22 23 

’03x 24 25 26 27 28 29 30 31 

’04x 32 33 34 35 36 37 38 39 

’05x 40 41 42 43 44 45 46 47 

’06x ⎛ 48 ⎞ 49 ⎡ 50 ⎤ 51 ⎣ 52 ⎦ 53 ⎢ 54 ⎥ 55 

’07x ⎧ 56 ⎫ 57 ⎩ 58 ⎭ 59 ⎨ 60 ⎬ 61 ⎪ 62 ⏐ 63 

’10x ⎝ 64 ⎠ 65 ⎜ 66 ⎟ 67 68 69 70 71 

’11x 72 73 74 75 76 77 78 79 

’12x 80 81 82 83 84 85 86 87 

’13x 88 89 90 91 92 93 94 95 

’14x 96 97 98 99 100 101 102 103 

’15x 104 105 106 107 108 109 110 111 

’16x 112 113 114 115 116 117 118 119 

’17x 120 121 122 123 124 125 126 127 

Font Layout 7: The font cmex10, containing the mathematical symbols 

that appear in varying sizes. This is the OMX encoding scheme.

G.3. The METAFONT program 497 

0 1 2 3 4 5 6 7 

’00x ǋ 0 ǈ 1 2 З 3 4 5 6 7 

’01x ǌ 8 ǉ 9 10 з 11 12 13 14 15 

’02x 16 Ж 17 18 19 20 21 22 23 

’03x 24 ж 25 26 27 28 29 30 31 

’04x ¨ 32 ! 33 ” 34 35 36 % 37 ´ 38 ’ 39 

’05x ( 40 ) 41 * 42 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 « 60 ı 61 » 62 ? 63 

’10x ˘ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 “ 92 ] 93 Ь 94 Ъ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 – 123 —124 125 ь 126 ъ 127 

Font Layout 8: The character font wncyr10, one of the Cyrillic fonts from 

the University of Washington. Its coding scheme is designated OT2. 

while the circles and oval segments (Section 13.1.4) are in lcircle10. 

The second pair of fonts with the added w contain thicker lines, for use 

with \thicklines. 

G.2.3 A M S Cyrillic fonts 

The Cyrillic fonts that are part of the amsfonts collection are described in 

Section 12.4.2. They conform to the font encoding scheme OT2, displayed 

in Layout 8. They are available as upright (wncyr*), bold (wncyb*), small 

caps (wncysc*), and upright sans serif (wncyss*) fonts. As for the CM 

fonts, * represents the design size in points. 

The NFSS system assigns these fonts to families cmr and cmss; that is, 

they are treated as members of the Computer Modern set. This is not as 

absurd as it sounds since they are intended to be used with the CM fonts. 

G.3 The METAFONT program 

METAFONT is a program for designing and developing character fonts, 

written by Donald E. Knuth as a companion to his T E X program (Knuth, 

1986c, 1986d, 1986e). It is this program that made the uniform high 

quality typographic output possible for T E X at a time when most printers 

had their own built-in fonts of varying standards.


The printer driver programs have the task of converting the DVI output 

from T E X into instruction code for a particular printer. To do this, they 

need to know how to print each of the symbols used in the document. 

These symbols are represented as a set of black and white dots (pixels) 

that is adjusted to the output resolution. Clearly to produce the letter A 

in a given size at 600 dpi (dots per inch) requires many more dots than at 

150 dpi (16 times). This information is contained in the corresponding 

.pk files, in compressed format. 

However, designing fonts as a set of dots, and that for any number of 

resolutions, and taking into account idiosyncrasies of individual printers, 

is an impossible task. Rather, the fonts are defined with.mf files, containing 

instructions on how to draw the symbols with a pen of a given shape. 

This is the most general definition, describing the ideal form. METAFONT 

then translates this ideal into the practical realization, the pixels, for any 

specified resolution. Aspects of the various printers, such as relative pixel 

size and shape, may also be included, so that the resulting patterns are 

both resolution and printer dependent. The font metric.tfm files are also 

produced in this process. 

Today it is hardly necessary for most users to know anything more 

about METAFONT other than that it exists. The drivers and previewers 

are now so constructed that when they notice that the required .pk file 

is missing, they invoke METAFONT themselves to generate it. In this way 

the collection of pixel files grows as needed and includes only those that 

are actually used. (Originally huge sets of pixel files had to be supplied 

to cover every remote possibility.) 

Another reason for the decline in the awareness of METAFONT is 

the increasing use of the type 1 fonts for PostScript and PDF output 

(Chapter 10). 

Donald Knuth has announced that he wishes to withdraw from any 

further development of either of these programs, being prepared merely to 

correct any definite errors they may contain. To emphasize this decision, 

he will from now on give version numbers to T E X that converge to π 

(3.14159. . . ) and to METAFONT that approach the value of e (2.71828. . . ). 

At present, T E X is at version 3.14159 and METAFONT at 2.718. As a 

consequence of this decision, any further major developments to these 

two programs, such as undertaken by user groups, will be under new 

names, since Knuth has kept the copyright to the existing names. 

G.4 Extended Computer fonts 

The Computer Modern fonts were developed in the early days of T E X when 

it could only handle 128 characters per font. The modern T E X program 

can deal with 256 characters in each font, requiring a new set of standard 

fonts to exploit this feature.

G.4. Extended Computer fonts 499 

G.4.1 Limitations and deficiencies of the CM fonts 

Most of the diacritical marks (accents) used in European languages written 

with the Latin alphabet are contained in, or may be generated by, T E X’s 

Computer Modern (CM) fonts. A basic set of naked accents is available 

for combination with other letters, such as the acute accent ´ with the 

letter e to make é. Other combinations may be constructed for diacritical 

marks that are not predefined in T E X or L AT E X. 

Fashioning diacritical characters as a combination of letters and special 

symbols has one great disadvantage for the T E X processing: words containing 

such characters cannot take part in the automatic word division 

since the hyphenation patterns include only pure letters. The accented 

letters, such as those in German and French and most other languages, 

must be treated as single characters in the hyphenation patterns, and 

must appear as single letters in the character set. 

In addition to diacritical characters, a number of special letters are 

employed in some European languages, such as ß, Æ, æ, Œ, œ, Ø, and 

ø, which are provided in standard T E X with the CM fonts (Section 2.5.6). 

However, other special letters, such as Ŋ, ŋ, Þ, þ, and ð, are missing 

completely and cannot be easily constructed from existing ones. 

G.4.2 The Cork proposal 

At the 1990 International T E X Conference in Cork, Ireland, an extension of 

the Latin alphabet and its assignments within the 256 character positions 

was proposed and accepted. This extension includes the majority of 

special and diacritical letters as single characters for many languages 

written with the Latin alphabet. Hyphenation patterns for such languages 

may include the special and diacritical letters as single letters for optimal 

word division by T E X and L AT E X. 

Character fonts conforming to the Cork scheme are to bear the identifying 

letters ec in their names for ‘Extended Computer’ in place of the 

cm for ‘Computer Modern’. 

G.4.3 The realization of EC fonts 

The Cork proposal for extending the T E X fonts to 256 characters was first 

implemented by Norbert Schwarz, who produced an initial set of META- 

FONT source files. He also selected the designation dc to emphasize that 

this was a preliminary realization of the EC fonts. Some work was still 

needed to fine-tune the design of several symbols. 

After issuing versions 1.2 and 1.3 of the DC fonts in 1995 and 1996, 

Jörg Knappen released the first set of true EC fonts in January, 1997. Font 

Layout 9 presents his font ecrm1000, the extended version of cmr10. The 

EC fonts are now considered to be stable in that neither their encoding


0 1 2 3 4 5 6 7 

’00x ` 0 ´ 1 ˆ 2 ˜ 3 ¨ 4 ˝ 5 ˚ 6 ˇ 7 

’01x ˘ 8 ¯ 9 ˙ 10 ¸ 11 ˛ 12 ‚ 13 ‹ 14 › 15 

’02x “ 16 ” 17 „ 18 « 19 » 20 – 21 — 22 23 

’03x ‰ 24 ı 25 j 26 ff 27 fi 28 fl 29 ffi 30 ffl 31 

’04x 32 ! 33 " 34 # 35 $ 36 % 37 & 38 ’ 39 

’05x ( 40 ) 41 * 42 + 43 , 44 - 45 . 46 / 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 : 58 ; 59 < 60 = 61 > 62 ? 63 

’10x @ 64 A 65 B 66 C 67 D 68 E 69 F 70 G 71 

’11x H 72 I 73 J 74 K 75 L 76 M 77 N 78 O 79 

’12x P 80 Q 81 R 82 S 83 T 84 U 85 V 86 W 87 

’13x X 88 Y 89 Z 90 [ 91 \ 92 ] 93 ^ 94 _ 95 

’14x ‘ 96 a 97 b 98 c 99 d 100 e 101 f 102 g 103 

’15x h 104 i 105 j 106 k 107 l 108 m 109 n 110 o 111 

’16x p 112 q 113 r 114 s 115 t 116 u 117 v 118 w 119 

’17x x 120 y 121 z 122 { 123 | 124 } 125 ~ 126 127 

’20x Ă 128 Ą129 Ć130 Č 131 Ď 132 Ě 133 Ę 134 Ğ 135 

’21x Ĺ 136 Ľ 137 Ł 138 Ń 139 Ň 140 Ŋ 141 Ő 142 Ŕ 143 

’22x Ř 144 Ś 145 Š 146 Ş 147 Ť 148 Ţ 149 Ű 150 Ů 151 

’23x Ÿ 152 Ź 153 Ž 154 Ż 155 Ĳ156 İ 157 đ 158 § 159 

’24x ă 160 ą 161 ć 162 č 163 ď 164 ě 165 ę 166 ğ 167 

’25x ĺ 168 ľ 169 ł 170 ń 171 ň 172 ŋ 173 ő 174 ŕ 175 

’26x ř 176 ś 177 š 178 ş 179 ť 180 ţ 181 ű 182 ů 183 

’27x ÿ 184 ź 185 ž 186 ż 187 ĳ 188 ¡ 189 ¿ 190 £ 191 

’30x À 192 Á193 Â194 Ã 195 Ä 196 Å 197 Æ 198 Ç 199 

’31x È 200 É 201 Ê 202 Ë 203 Ì 204 Í 205 Î 206 Ï 207 

’32x Ð 208 Ñ209 Ò210 Ó211 Ô 212 Õ 213 Ö 214 Œ215 

’33x Ø 216 Ù217 Ú218 Û 219 Ü 220 Ý 221 Þ 222 ß223 

’34x à 224 á 225 â 226 ã 227 ä 228 å 229 æ 230 ç 231 

’35x è 232 é 233 ê 234 ë 235 ì 236 í 237 î 238 ï 239 

’36x ð 240 ñ 241 ò 242 ó 243 ô 244 õ 245 ö 246 œ 247 

’37x ø 248 ù 249 ú 250 û 251 ü 252 ý 253 þ 254 ß 255 

Font Layout 9: The extended font ecrm1000 with T1 encoding.

G.4. Extended Computer fonts 501 

nor their metrics (the .tfm files) will be changed in future. Thus their 

behavior as far as T E X and L AT E X are concerned is finalized. The actual 

printed characters might be modified slightly in later updates. 

The EC font names are of the form ecxxnnnn, where xx represents two 

letters specifying the font characteristics, and nnnn the design size in 

hundredths of points. Thus ecrm1000 is the upright Roman font in size 

10 points. 

The METAFONT source files are available for the following extended 

fonts (without the size specification): 

ecrm 

ecbx 

ecbl 

ecrb 

ecrb 

ecti 

ecui 

ecbi 

eccc 

ecxc 

ecsc 

ecoc 

ecci 

ectt 

ecit 

ecvt 

ecvi 

ecvi 

ectc 

ecst 

ecss 

ecsi 

ecsx 

ecso 

ecdh 

Comparing these with the root names of the CM fonts from Table G.1 on 

page 491, one may easily recognize the correspondence. For example, 

ecbx* is the extended bold font corresponding to the CM font cmbx*. 

It is intended that the EC fonts should exist in most design sizes. 

The present distribution contains almost all the fonts in sizes from 5 to 

35.83 pt, that is, with size specifications: 

0500 0600 0700 0800 0900 1000 1095 

1200 1440 1728 2074 2488 2986 3583 

In contrast to the CM text fonts which exhibit differences in the symbol 

assignments among them, as illustrated in Layouts 1–4, the EC fonts all 

have exactly the same encoding scheme, as presented in Layout 9. 

The EC fonts are now part of the standard L AT E X installations. 

A parallel set of fonts called text companion, or TC, fonts is also provided. 

These contain special symbols for text that are normally found in 

the CM math fonts, if at all, such as currency symbols and degree signs. 

These fonts are still somewhat experimental so that the symbol assignments 

are not yet stable. The current contents are shown in Layout 10. 

G.4.4 Invoking the EC and TC fonts 

The EC fonts correspond to the NFSS encoding scheme T1 (Section A.1). 

The simplest way to activate them is to place 

\usepackage[T1]{fontenc} 

in the preamble of the document. All this really does is to make T1 

the standard encoding by redefining \encodingdefault to be T1, and 

it loads the file t1enc.def which redefines the accent and special letter 

commands by means of the encoding commands of Section A.3.7.


0 1 2 3 4 5 6 7 

’00x ` 0 ´ 1 2 3 ¨ 4 ˝ 5 6 ˇ 7 

’01x 8 ¯ 9 10 ¸ 11 ˛ 12 ‚ 13 14 15 

’02x 16 17 „ 18 19 20 21 — 22 ‌ 23 

’03x ← 24 → 25 26 27 28 29 30 ‌ 31 

’04x ␢ 32 33 34 35 $ 36 37 38 ' 39 

’05x 40 41 ∗ 42 43 , 44 45 . 46 ⁄ 47 

’06x 0 48 1 49 2 50 3 51 4 52 5 53 6 54 7 55 

’07x 8 56 9 57 58 59 〈 60 − 61 〉 62 63 

’10x 64 65 66 67 68 69 70 71 

’11x 72 73 74 75 76 ℧ 77 78 ○ 79 

’12x 80 81 82 83 84 85 86 Ω 87 

’13x 88 89 90 〚 91 92 〛 93 ↑ 94 ↓ 95 

’14x ` 96 97 98 99 100 101 102 103 

’15x 104 105 106 107 108 109 ♪ 110 111 

’16x 112 113 114 115 116 117 118 119 

’17x 120 121 122 123 124 125 126 127 

’20x ˘ 128 ˇ 129 ˝ 130 ̏ 131 † 132 ‡ 133 ‖ 134 ‰135 

’21x • 136 ℃ 137 $ 138 ¢ 139 ƒ 140 ₡ 141 ₩ 142 ₦ 143 

’22x 144 145 ₤ 146 147 ‽ 148 149 ₫ 150 151 

’23x ‱152 153 ฿154 № 155 156 ℮ 157 ◦ 158 ℠ 159 

’24x ⁅ 160 ⁆ 161 ¢ 162 £ 163 ¤ 164 ¥ 165 ¦ 166 § 167 

’25x ¨ 168 ©169 ª 170 171 ¬ 172 ℗173 ®174 ¯ 175 

’26x ° 176 ± 177 ² 178 ³ 179 ´ 180 µ 181 182 · 183 

’27x ※ 184 ¹ 185 º 186 √ 187 ¼ 188 ½ 189 ¾ 190 € 191 

’30x 192 193 194 195 196 197 198 199 

’31x 200 201 202 203 204 205 206 207 

’32x 208 209 210 211 212 213 × 214 215 

’33x 216 217 218 219 220 221 222 223 

’34x 224 225 226 227 228 229 230 231 

’35x 232 233 234 235 236 237 238 239 

’36x 240 241 242 243 244 245 ÷ 246 247 

’37x 248 249 250 251 252 253 254 255 

Font Layout 10: The text companion font tcrm1000 with the TS1 encoding 

scheme.

G.5. PostScript fonts 503 

Package: To obtain access to the symbols in the TC fonts, one can load the 

textcomp textcomp package, which not only redefines several existing symbol commands, 

it also adds many new ones. For example, \copyright, which is 

normally defined to be \textcircle{c}, is changed to print character 

169 from an appropriate TC font. Character 191 is the symbol for the European 

currency unit, the euro, printed with \texteuro; however, better 

ways of producing it are presented in Section 2.5.8. 

G.4.5 Special character commands 

Inspecting Font Layout 9, one notices that the EC fonts contain not only 

many single characters that are formed out of two CM symbols (like Ä = 

A + ¨) but also several characters that have no correspondence in the CM 

font layout at all. The first type is accommodated by internally redefining 

the action of accent and special character commands. The second set 

requires new commands that are recognized only when the T1 encoding 

is active. These are 

the ogonek accent \k{o}: o˛ 

special letters \DH = Ð \DJ = Ð \NG = Ŋ \TH = Þ 

\dh = ð \dj = đ \ng = ŋ \th = þ 

special symbols \guillemotleft = « \guillemotright = » 

\guilsinglleft = ‹ \guilsinglright = › 

\quotedblbase = „ \quotesinglbase = ‚ 

\textquotedbl = " 

When issued in OT1 encoding, these commands print an error message. 

Note: The \guillemotleft and \guillemotright are not misprints 

even though the proper word for the French quotations marks is guillemet. 

The PostScript fonts contain these erroneous names for these symbols 

and this mistake has propagated to such an extent that it can never be 

removed from all the software that includes it. A guillemot is in fact an 

Arctic bird, not a French quotation mark. 

G.5 PostScript fonts 

PostScript fonts, also known as outline or type 1 fonts, are treated exactly 

the same way as the METAFONT fonts as far as L AT E X is concerned: during 

the processing, a.tfm file is read in for each font specifying the character 

sizes and other properties. This is all that L AT E X needs to know, for it is 

the task of the driver program to print the actual character that fills the 

reserved space. 

The driver makes use of the virtual font mechanism, which means that 

what L AT E X sees are actually artificial fonts that do not really exist on their


own. There are.tfm files for these fonts, so the L AT E X processing proceeds 

as normal. What the driver then does is to read a .vf file instead of the 

pixel.pk files. The instructions in the virtual font file tell the driver how 

to create each character: they may be drawn, taken from other fonts, 

or distorted. This is how PostScript slanted and small caps fonts are 

emulated, for such fonts do not exist in the ‘raw’ form. See Section 10.1.4 

for example of how this works. 

Even the font layout can be redesigned with virtual fonts. The raw 

PostScript fonts have an encoding scheme that conforms to neither OT1 

nor T1, but which is used as a pool of symbols for constructing virtual 

fonts that do conform to these schemes. In particular, the Computer 

Modern fonts contain upper case Greek letters in the first 11 slots (Layout 1 

on page 492) which are to be found in the PostScript symbol font only. 

The virtual font ptmr7t conforms to this by taking its characters from 

both the raw Times-Roman and Symbol fonts. 

G.5.1 Naming scheme for PostScript fonts 

In order to be compatible with all possible operating systems, it is necessary 

to reduce the names of the PostScript font files to a maximum of 

eight characters. This makes for an extremely abbreviated and cryptic 

nomenclature. 

The most commonly used scheme is that of Karl Berry. Here, the first 

letter of the name specifies the supplier of the font, for example p for 

Adobe (stands for PostScript), or h for Bigelow & Holmes, who designed 

the Lucida fonts used in this book, or m for Monotype, or l for Linotype, 

and so on. 

Table G.2: Root names of the 35 standard PostScript fonts 

pagd AvantGarde-Demi phvrrn Helvetica-Narrow 

pagdo AvantGarde-DemiOblique phvron Helvetica-Narrow-Oblique 

pagk AvantGarde-Book pncb NewCenturySchlbk-Bold 

pagko AvantGarde-BookOblique pncbi NewCenturySchlbk-BoldItalic 

pbkd Bookman-Demi pncr NewCenturySchlbk-Roman 

pbkdi Bookman-DemiItalic pncri NewCenturySchlbk-Italic 

pbkl Bookman-Light pplb Palatino-Bold 

pbkli Bookman-LightItalic pplbi Palatino-BoldItalic 

pcrb Courier-Bold pplr Palatino-Roman 

pcrbo Courier-BoldOblique pplri Palatino-Italic 

pcrr Courier psyr Symbol 

pcrro Courier-Oblique ptmb Times-Bold 

phvb Helvetica-Bold ptmbi Times-BoldItalic 

phvbo Helvetica-BoldOblique ptmr Times-Roman 

phvbrn Helvetica-Narrow-Bold ptmri Times-Italic 

phvbon Helvetica-Narrow-BoldOblique pzcmi ZapfChancery-MediumItalic 

phvr Helvetica pzdr ZapfDingbats 

phvro Helvetica-Oblique

G.6. Computer Modern as PostScript fonts 505 

Table G.3: A selection of the encoding suffixes in the Berry nomenclature 

Suffix Encoding NFSS designation Page 

7t 7-bit T E X text encoding OT1 492 

7m T E X math italics encoding OML 495 

7v T E X math extension encoding OMX 496 

7y T E X math symbols encoding OMS 495 

8t 8-bit Cork encoding T1 500 

8a Adobe standard encoding 

8r TeXBase1Encoding 

A two-letter typeface code follows the supplier letter, such as tm for 

Times-Roman. Next come various letters to specify weight, for example 

r for regular (upright Roman) or b for bold, and variant, like i for italic, 

o for oblique (slanted). Next come a number plus letter to indicate the 

encoding scheme, followed by possible width code letters. The Berry 

names for the 35 standard PostScript fonts that should be loaded in every 

printer are listed in Table G.2 on the facing page, without the encoding 

suffixes. 

The most important encoding suffixes for our purposes are listed in 

Table G.3. The 7-bit encodings have already been explained in Section G.2.2 

and shown in Layouts 1, 5–7, while the 8t or T1 scheme is to be found in 

Layout 9 in Section G.4.3. 

The other 8-bit schemes listed in Table G.3 are the raw encoding, 8r, 

and the Adobe standard encoding, 8a, which is the default for most type 1 

fonts if no re-encoding is specified. This scheme plays absolutely no role 

in the NFSS installation for PostScript fonts. 

Virtual fonts ptmr7t and ptmr8t are both constructed from the same 

raw font ptmr8r, differing only in the character assignments. There are 

also raw fonts that are modifications of the basic fonts. For example, 

ptmro8t is an oblique Times-Roman virtual font, based on ptmro8r, the 

raw version. However, there is no such font in the repertoire of Table G.2. 

This pseudo font is generated by applying a PostScript slanting operation 

to Times-Roman, as demonstrated in Section 10.1.4. 

G.6 Computer Modern as PostScript fonts 

The traditional pixel fonts produced by METAFONT have served us well 

in the days when the emphasis was on printed paper output. Even 

for PostScript output, the pixel fonts are acceptable for online viewing. 

However, they become very tedious when included in PDF output for


viewing: they appear very fuzzy and require a long time to draw. For 

this reason, PDF output, whether produced with pdfT E X, dvipdfm, or 

converted from PostScript, should use type 1 fonts exclusively. To this 

end, the Computer Modern fonts themselves have been converted to type 1 

fonts. 

The first such undertaking was carried out by Basil Malyshev whose 

set of font files are known as bakoma. However, the set that is more 

often used today is the result of a joint project of Y&Y Inc. and Blue 

Sky Research. Originally part of their commercial T E X installations, these 

fonts are now available free of charge. They are included on the T E XLive 

CD under texmf\fonts\type1\bluesky\ in several subdirectories as 

.pfb files. They bear the same names and have the same font metrics 

as the corresponding pixel fonts (Appendix G) and therefore do not need 

any additional .tfm files. However, one must also get the mapping file 

bsr.map from texmf\dvips\bluesky\ and add it to the list of mapping 

files in config.ps. 

The Bluesky fonts are incomplete in that certain less often used 

fonts are missing in some sizes. One can either include the map file 

bsr-interpolated.map, which substitutes the missing sizes with the 

nearest approximation, or employ the bakoma fonts that fill the gaps, 

with the map file bakomaextra.map. The T E XLive CD does not include the 

complete bakoma set, but only those fonts that are missing in the Bluesky 

ensemble. 

Together Bluesky and bakoma provide PostScript-encoded versions for 

all the Computer Modern fonts described in Section G.2, for text, math, 

symbols, decoration, pictures, along with the additional A M S fonts for 

additional symbols and Cyrillic. However, these are only for the OT1 

encoding with 128 characters per font. 

A more recent effort to apply PostScript to the Extended Computer (EC) 

fonts with the T1 encoding has been successfully completed by Vladimir 

Volovich with his cm-super collection. Each of the basic.pfb files contains 

from 468 to 585 characters, which by means of encoding vectors and 

mapping files can be used to emulate all the fonts in Section G.4, including 

the text companion fonts with TS1 encoding, and many other additional 

encoding schemes. They were generated from the original METAFONT.mf 

source files and should therefore produce exactly the same results as the 

corresponding pixel fonts. 

To use them, one must install the cm-super package from the T E XLive 

CD, which copies the .pfb files from \texmf\fonts\type1\cm-super\ 

and the mapping and encoding files from \texmf\dvips\cm-super\. 

The metric.tfm files are the same as those for the pixel files and so need 

not be provided extra.

H 

Command Summary 

This appendix contains a brief description of all the L AT E X commands, 

in alphabetical (ascii) order, neglecting the leading backslash \\, along 

with some T E X commands that have been explained in this book. In the 

following section, the commands are presented in their logical grouping 

in a number of tables and figures. 

H.1 Brief description of the L AT E X commands 

For each command in the following summary, the section and page number 

is given where it is introduced and described in detail. The numbers are 

shown in the form: ‘(Section) – Page’: for example (2.5.1) – 13 means 

‘Section 2.5.1, page 13’. If these numbers are missing, then the command 

has not been presented in the book but is only mentioned here. 

The following notations may be added to the commands: 

[m] those permitted in math mode only; 

[a] those belonging to A M S-L AT E X; 

[p] those allowed only in the preamble. 

\ . . . . . . . . . . . . . . . . . . . . . . . (2.1), (2.7.1) – 19, 29 

Normal space between words after a command without arguments 

or after a period that is not the end of a sentence. 

! . . . . . . . . . . . . . . . . . . . . . . . . . . . (9.4.2) – 226 

Field separation character within the \index command. For example: 

with \index{command!fragile} one produces an index sub-entry 

‘fragile’ under the main entry ‘command’. 

!‘ produces ¡ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\! [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.5.1) – 145 

In math mode, a negative space of −1/6 quad: xx\!x = xxx. 

507

508 Appendix H. Command Summary 

" . . . . . . . . . . . . . . . . (2.5.2), (9.4.2), (14.2) – 23, 227, 311 

1. In normal text, this produces the double closing quote ”. 

2. Literal sign for MakeIndex, in order to print one of the special 

characters !, @, |, or ". Example: \index{"!} to enter character 

! without it being interpreted as a separation character. 

3. Delimiter for a text field in BIBT E X. Example: 

AUTHOR = "Donald E. Knuth". 

\" . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces an umlaut accent: \"{a} = ä. 

# . . . . . . . . . . . . . . . . . . . . . . (8.3.2), (8.4.2) – 187, 198 

Argument replacement character in a user-defined command or environment. 

## . . . . . . . . . . . . . . . . . . . . . . . . . . (8.5.6) – 204 

Replacement character for an internal argument within a nested userdefined 

command or environment. 

\# . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.4) – 23 

Command to produce a hash symbol: \# = #. 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . (5.1) – 119 

Toggle character for switching between text and in-line math modes. 

On the first appearance (text to math) it behaves the same as $ 

or \begin{math}, while the second call (math to text) is as $ or 

\end{math}. 

\$ . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.4) – 23 

Command to produce a dollar sign: \$ = $. 

% . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.11) – 118 

Comment character. The rest of the line of text following % is ignored 

by the T E X processing. 

\% . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.4) – 23 

Command to produce a percent sign: \% = %. 

& . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.8.1) – 97 

Indicates a new column in array and tabular environments. 

\& . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.4) – 23 

Command to produce an ampersand symbol: \& = &. 

\’ . . . . . . . . . . . . . . . . . . . . . . (2.5.7), (4.6.4) – 24, 83 

1. Command to produce an acute accent: \’a = á.

H.1. Brief description of the L AT E X commands 509 

2. Within the tabbing environment, a command to jump to the 

end of the current column. 

( ) . . . . . . . . . . . . . . . . . . . . (13.1.2), (14.2) – 289, 312 

For a picture command in picture environment, specifies a coordinate 

pair. In BIBT E X, an alternative form for the outermost grouping 

of the entry type. 

$ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5.1) – 119 

Switches from text to in-line math mode to produce formulas within 

a line of text. It functions the same as \begin{math} and as a $ sign 

in text mode. 

$ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5.1) – 119 

Switches back from in-line math mode to text mode. It functions the 

same as \end{math} and as a $ sign in math mode. 

\+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.3) – 82 

Within the tabbing environment, increments the left margin by one 

tab stop (moves it to the right). 

\, . . . . . . . . . . . . . . . . . . . . . . (2.7.1), (5.5.1) – 29, 145 

Small space, the size of 1/6 quad, for use in text and math mode: 

xx\,x = xx x. 

- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.3) – 23 

As -, produces the hyphen - for compound words and word division, 

as --, the en dash –, and as ---, the em dash —. 

\- . . . . . . . . . . . . . . . . . . . . . . (2.8.1), (4.6.3) – 35, 82 

1. Denotes possible word division. If a word contains at least one 

\- the normal word division rules are suspended for that word 

and division may occur only at those locations. 

2. Within the tabbing environment, decrements the left margin 

by one tap stop (moves it to the left). 

\. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Command to produce a dot accent: \.o = ˙o. 

\/ . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.10) – 26 

Command to break up ligatures: shelf\/ful = shelfful. 

\: [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.5.1) – 145 

In math mode, a medium space, the size of 2/9 quad: xx\:x = xx x. 

\; [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.5.1) – 145 

In math mode, a large space, the size of 5/18 quad: xx\;x = xx x.


\< . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.3) – 82 

Within the tabbing environment, moves to the left by one tab stop. 

\= . . . . . . . . . . . . . . . . . . . . . . (2.5.7), (4.6.1) – 24, 81 

1. Command to produce a macron accent: \=o = ō. 

2. Within the tabbing environment, sets a tab stop at the current 

position within the line. 

\> . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.1) – 81 

Within the tabbing environment, moves right to the next tab stop. 

?‘ produces ¿ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

@ . . . . . . . . . . . . . . . . . . . . . . (9.4.2), (14.2) – 226, 311 

1. In MakeIndex, separates an entry in an \index command into 

a lexical (for alphabetization) and printing part. Example: 

\index{sum@$\sum$} means that the entry appears in the index 

at the location of the word ‘sum’ but what is printed is the 

summation sign . 

2. In BIBT E X, denotes the entry type. Example: @BOOK indicates that 

the following literature entries correspond to those of a book. 

\@ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

Extra space at the end of a sentence ending with a capital letter. 

[ ] . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.1) – 18 

After commands or environment calls, specifies an optional argument. 

\[ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5.1) – 120 

Switches from text mode to displayed math mode for putting 

a formula on a line by itself. Has the same effect as 

\begin{displaymath} or $$ in text mode. 

\\[space] . . . . . . . . . . . . . . . . . . . . . . . . (2.7.2) – 31 

Ends the current line (without right justifying it). The optional argument 

[space] inserts additional vertical spacing of length space 

before the next line. 

\\*[space] . . . . . . . . . . . . . . . . . . . . . . . . (2.7.2) – 31 

The same as \\ but prevents a page break from occurring between 

the current and next line. 

\] . . . . . . . . . . . . . . . . . . . . . . . . . . . . (5.1) – 120 

Switches back from displayed math mode to text mode. Has the 

same effect as \end{displaymath} or $$ in math mode.


ˆ [m] . . . . . . . . . . . . . . . . . . . . . . . . . . (5.2.2) – 121 

Exponents and superscripts in equations: xˆ2 = x 2 , xˆ{-2n} = 

x −2n . 

\ˆ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Command to produce a circumflex accent: \ô = ô. 

_ [m] . . . . . . . . . . . . . . . . . . . . . . . . . . (5.2.2) – 121 

Subscripts in equations: a_n = an, a_{i,j,k} = ai,j,k. 

\_ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.4) – 23 

Command to produce the underbar sign: t\_v = t v. 

\‘ . . . . . . . . . . . . . . . . . . . . . . (2.5.7), (4.6.4) – 24, 83 

1. Command to produce a grave accent: \‘o = ò. 

2. Within the tabbing environment, pushes the following text up 

against the right margin of the line. 

{ } . . . . . . . . . . . . . . . . . (2.1), (2.2), (14.2) – 18, 20, 311 

1. After a command or environment call, specifies a mandatory 

argument. 

2. Grouping a section of text to create a nameless environment. 

3. In BIBT E X, delimiting the name of an entry type, as well as an 

alternative delimiter for the text field. 

\{ produces a left curly brace: \{ = { . . . . . . . . . . (2.5.4) – 23 

| [m] produces | . . . . . . . . . . . . . . (5.3.4), (5.4.1) – 126, 132 

| . . . . . . . . . . . . . . . . . . . . . . . . . . . . (9.4.2) – 227 

In MakeIndex, the command character within a \index command. 

1. After the command \newcommand{\ii}[1]{\textit{#1}} 

has been defined, \index{entry|ii} produces the page number 

for ‘entry’ in the index in italic type. 

2. The cross-reference command \see from makeidx.sty can be 

invoked with \index{bison|see{buffalo}} to produce crossreferences 

within the index. 

\| [m] produces . . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\} produces a right curly brace: \} = } . . . . . . . . . . (2.5.4) – 23 

˜ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

A normal space between words, but without the possibility that the 

line will be broken there. Example: Prof.˜Jones ensures that ‘Prof.’ 

and ‘Jones’ both remain on the same line. 

\˜ . . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Command to produce a tilde accent: \ñ = ñ.


\a= . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Produces a macron accent within tabbing environment: \a=o = ō. 

\a’ . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Produces an acute accent within tabbing environment: \a’o = ó. 

\a‘ . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Produces a grave accent within tabbing environment: \a‘o = ò. 

\AA produces ˚A . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\aa produces ˚a . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\abovedisplayskip [m] . . . . . . . . . . . . . . . . (5.5.4) – 149 

Vertical space between a long displayed equation and the preceding 

line of text. A new value may be assigned with the \setlength 


\setlength{\abovedisplayskip}{10pt plus2pt minus5pt} 

\abovedisplayshortskip [m] . . . . . . . . . . . . . (5.5.4) – 149 

Vertical space between a short displayed equation and the preceding 

line of text. A new value may be assigned with the \setlength 

command as in the above example. 

\abstractname . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the heading for the abstract. In English, this is 

‘Abstract’ but may be altered for adaptation to other languages. 

\acute{x} [m] . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Acute accent over math variable x: \acute{a} = á 

\Acute{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \acute, but with multiple 

A M S-L AT E X math accents the positioning will be correct. 

\addcontentsline{file}{format}{entry} . . . (3.4.3), (3.4.4) – 59, 60 

Manual addition of entry into the list file.toc,.lof, or.lot, according 

to the value of file, to be formatted as the heading of a sectioning 

command, as given by format, for example 

\addcontentsline{toc}{section}{References} 

\address{sender} . . . . . . . . . . . . . . . . . . . (16.1) – 351 

In the letter document class, enters the sender’s address. Multiple 

lines in sender are separated by \\. 

\addtime{secs} . . . . . . . . . . . . . . . . . . . (15.1.3) – 327 

In the slides class, if the option clock has been selected, a time 

marker, in minutes, appears at the bottom of the notes. This command 

adds the specified number of seconds to the marker. See also 

\settime.


\addtocontents{file}{entry} . . . . . . . . (3.4.3), (3.4.4) – 59, 60 

Manual addition of entry into the list file.toc,.lof, or.lot, according 

to the value of file. Example: 

\addtocontents{lof}{\protect\newpage} 

\addtocounter{counter}{number} . . . . . . . . . . (8.1.3) – 182 

Adds number to the current value of the number stored in counter. 

\addtolength{length name}{length} . . . . . . . . . . (8.2) – 184 

Adds the quantity length to the current value of the length command 

\length name. 

\addvspace{length} . . . . . . . . . . . . . . . . . . . (8.2) – 185 

Inserts vertical spacing of amount length between paragraphs at the 

point where the command is given. If other vertical spacing exists, 

the total will not exceed length. 

\AE produces Æ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\ae produces æ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\aleph [m] produces ℵ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\allowdisplaybreaks[num] [p][a] . . . . . . . . . . (12.2.7) – 278 

With the amsmath package, allows automatic page breaks to occur 

within multiline math formulas. If num is present, it takes a value 

of 0–4 to increase the ease with which page breaks occur. Without 

this command, a manual page break can be made at the end of any 

formula line with \displaybreak. 

\Alph{counter} . . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of counter as a capital letter. 

\alph{counter} . . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of counter as a lower case letter. 

\alpha [m] produces α . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\alsoname . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command for use in the makeidx package. It prints the text for a 

command \seealso. In English, this is ‘see also’ but may be altered 

for adaptation to other languages. 

\amalg [m] produces ∐ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\and . . . . . . . . . . . . . . . . . . . . . . . . . . . (3.3.1) – 53 

Used to separate author names within the \author command for 

generating a title page with \maketitle. 

\angle [m] produces ∠ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\appendixname . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the heading for the appendix. In English, this 

is ‘Appendix’ but may be altered for adaptation to other languages.


\approx [m] produces ≈ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\arabic{counter} . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of counter as an Arabic number. 

\arccos [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘arccos’ in equations. 

\arcsin [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘arcsin’ in equations. 

\arctan [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘arctan’ in equations. 

\arg [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘arg’ in equations. 

\arraycolsep . . . . . . . . . . . . . . . . . . . . . . (4.8.2) – 98 

Half the width of the intercolumn spacing in the array environment. 

Value is assigned with the L AT E X command \setlength: 

\setlength{\arraycolsep}{3mm} 

\arrayrulewidth . . . . . . . . . . . . . . . . . . . . (4.8.2) – 98 

The thickness of vertical and horizontal lines in the array and 

tabular environments. Its value is assigned to a length with 

\setlength: 

\setlength{\arrayrulewidth}{0.5mm} 

\arraystretch . . . . . . . . . . . . . . . . . . . . . (4.8.2) – 98 

Factor to change the spacing between lines in a table, normal value 

being 1. Spacing is multiplied by this factor, which is set to a new 

value with \renewcommand{\arraystretch}{factor}. 

\ast [m] produces ∗ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\asymp [m] produces ≍ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\AtBeginDocument{code} [p] . . . . . . . . . . . . . . (D.2.4) – 444 

Stores the code to be inserted into the processing stream when 

\begin{document} is executed. Commands that are only allowed in 

the preamble may be part of code. A package might include coding 

in this way to ensure that it is not overwritten by another package. 

\AtEndDocument{code} [p] . . . . . . . . . . . . . . . (D.2.4) – 444 

Stores the code to be inserted into the processing stream when 

\end{document} is executed. A package might include coding in 

this way to have additional features printed automatically at the end 

of the document.


\AtEndOfClass{code} [p] . . . . . . . . . . . . . . . (D.2.4) – 444 

Stores the code to be inserted into the processing stream when the 

current class file has finished being read. May only be given in a class 

file, or in another file that is read by a class file. May be used by a 

local configuration file to overwrite defaults in the class file itself. 

\AtEndOfPackage{code} [p] . . . . . . . . . . . . . . (D.2.4) – 444 

Stores the code to be inserted into the processing stream when the 

current package file has finished being read. May only be given in a 

package file, or in another file that is read by a package file. May be 

used by a local configuration file to overwrite defaults in the package 

file itself. 

\author{name} . . . . . . . . . . . . . . . . . . . . . (3.3.1) – 53 

Enters the author name(s) for a title page produced by the 

\maketitle command. 

\b{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Command to produce an underbar accent: \b{o} = ō. 

\backmatter . . . . . . . . . . . . . . . . . . . . . . . (3.3.5) – 57 

In the book class, introduces the material that comes at the end 

(bibliography, index) by turning off the chapter numbering of the 

\chapter command. 

\backslash [m] produces \ . . . . . . . . . . . . . . . (5.3.6) – 127 

\bar{x} [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Macron accent over the math variable x: \bar{a} = ā. 

\Bar{x} [m][a] . . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \bar, but with multiple 


\baselineskip . . . . . . . . . . . . . . . . . . . . . (3.2.4) – 46 

Interline spacing within a paragraph. Every font has its own internal 

line spacing. A new value (a rubber length) may be assigned with 

\setlength: 

\setlength{\baselineskip}{12pt plus2pt minus1pt} 

\baselinestretch . . . . . . . . . . . . . . (3.2.4), (4.1.2) – 46, 63 

A factor with the normal value of 1 by which the internal length 

\baselineskip is multiplied to produce the actual interline spacing. 

May be changed with: 


The new value takes effect after the next change in font size!


\begin{envrnmnt} . . . . . . . . . . . . . . . . . . . . (2.2) – 19 

Start of an environment with the name envrnmnt. This command 

must be coupled with \end{envrnmt} to terminate the environment. 

The environment name in both these commands must be identical. 

\begin{abstract} . . . . . . . . . . . . . . . . . . . . (3.3.2) – 55 

Start of the environment abstract to produce an abstract. With 

document class article, font size \small and the quotation environment 

are selected. With report, the abstract appears on a 

separate page with normal font size and line width. In both cases, 

the heading Abstract is centered above the text. 

\begin{align} [a] . . . . . . . . . . . . . . . . . . (12.2.6) – 274 

With the amsmath package, switches to displayed math mode to produce 

a set of aligned equations. Line are terminated by \\ commands. 

The lines are split into columns aligned on the first, third, fifth . . . & 

characters. Each line receives an equation number unless the *-form 

of the environment has been selected. 

\begin{alignat}{num} [a] . . . . . . . . . . . . . (12.2.6) – 275 

Is the same as the align environment except that no spacing is 

inserted automatically between the column pairs. The argument num 

is the number of column pairs = (1 + n&)/2 where n& is the number 

of & signs in one row. Explicit spacing may be placed between column 

pairs, especially if the left part of that pair is otherwise empty. 

\begin{aligned}[pos] [m][a] . . . . . . . . . . . . (12.2.6) – 276 

With the amsmath package, is like the align environment but is 

used as an element within math mode. The optional argument pos 

determines the vertical positioning relative to neighboring elements: 

t or b for top or bottom, no argument for centering. 

\begin{appendix} . . . . . . . . . . . . . . . . . . . . (3.3.4) – 57 

Start of the environment appendix to produce an appendix. The 

main section counter is reset to zero and its numbering appears as 

capital letters. 

\begin{array}[pos]{col} [m] . . . . . . . . (4.8.1), (5.4.3) – 95, 134 

Start of the environment array to produce matrices and arrays in 

math mode. The column definition col contains a formatting character 

for each column. Thus \begin{array}{lcr} produces an array 

with three columns: one left justified, one centered, and one right 

justified. The optional parameter pos determines how the array is 

aligned vertically with text outside it on the same line: t with the top 

line, b with the bottom line, while the default is with the center. See 

also \begin{tabular} 

\begin{center} . . . . . . . . . . . . . . . . . . . . . (4.2.1) – 67 

Start of the environment center. Each line of text terminated by \\ 

appears centered. See also \centering.


\begin{command name} . . . . . . . . . . . . . . . . . (2.2) – 20 

Most declaration commands, such as the font styles and sizes, can be 

used as environment names. For example, \begin{small} switches 

to font size \small until the countercommand \end{small} is given. 

\begin{alltt} . . . . . . . . . . . . . . . . . . . . (4.9.1) – 111 

When the alltt package has been loaded, this environment prints 

out original text in typewriter typeface, maintaining line breaks, special 

symbols, and so on, except for \ { } which function as usual. 

This allows commands to be executed within the typewriter text. 

\begin{bmatrix} [m][a] . . . . . . . . . . . . . . . (12.2.4) – 266 

Like the matrix environment, but enclosed in square brackets []. 

\begin{Bmatrix} [m][a] . . . . . . . . . . . . . . . (12.2.4) – 266 

Like the matrix environment, but enclosed in curly braces { }. 

\begin{cases} [m][a] . . . . . . . . . . . . . . . . (12.2.6) – 277 

With the amsmath package, writes math expressions on several lines, 

terminated by \\, in left-justified columns, separated by &, with a 

curly brace enclosing all lines at the left, the whole being centered 

vertically. 

\begin{description} . . . . . . . . . . . . . . . . . . (4.3.3) – 70 

Start of the environment description to produce an indented list 

with labels. The label text is the argument label in the command 

\item[label]. 

\begin{displaymath} . . . . . . . . . . . . . . . . . . (5.1) – 120 

Switches from text to displayed math mode for producing a formula 

on a line by itself. Functions the same as \[. 

\begin{document} . . . . . . . . . . . . . . . . . . . . (1.5.2) – 12 

Start of the outermost environment of a text document. This command 

terminates the preamble. It is obligatory for every L AT E X document, 

as is its counterpart \end{document} for ending the document. 

\begin{enumerate} . . . . . . . . . . . . . . . . . . . (4.3.2) – 70 

Start of the environment enumerate to produce a numbered, indented 

listing. The style of numbering depends on the depth of 

nesting; at the first level, it consists of a running Arabic number that 

is incremented with each call to \item. 

\begin{eqnarray} [m] . . . . . . . . . . . . . . . . . (5.4.7) – 138 

Switches from text to displayed math mode to produce a set of equations 

or a multiline formula in the form of a three-column table {rcl}. 

The individual lines of the formula are ended with the command \\; 

the fields within a line are separated by & characters. Each line is 

given a sequential equation number unless the command \nonumber 

appears within it.


\begin{eqnarray*} [m] . . . . . . . . . . . . . . . . (5.4.7) – 138 

Is the same as the eqnarray environment except that no equation 

numbers are printed. 

\begin{equation} [m] . . . . . . . . . . . . . . . . . . (5.1) – 120 

Switches from text to displayed math mode to produce a formula on 

a line by itself, including an automatic sequential equation number. 

\begin{falign} [a] . . . . . . . . . . . . . . . . . (12.2.6) – 275 

With the amsmath package, is the same as the align environment 

except that spacing is inserted between the column pairs to fill up 

the entire line. 

\begin{figure}[loc] . . . . . . . . . . . . . . . . . . (7.1) – 169 

Float environment for entering text for a figure. The optional argument 

loc can be any combination of the letters h, t, b, and p to 

determine the various positioning possibilities. Default is tbp. The 

character ! may additionally be given to ignore all float spacing and 

number restrictions set by the float style parameters. 

\begin{figure*}[loc] . . . . . . . . . . . . . . . . . . (7.1) – 169 

The same as the figure environment except that the figure is inserted 

over the width of two columns when the option twocolumn or the 

command \twocolumn has been selected. The standard form figure 

will only fill the width of one column. 

\begin{filecontents}{file name} [p] . . . . . . . . . (D.2.9) – 448 

An environment that may be given only before \documentclass, it 

writes its lines literally to a file of the specified name, if that file does 

not already exist. It also adds comment lines stating its source. In 

this way, non-standard files may be included in the main document 

file for shipment to other installations. If a file with the stated name 

already exists, it is not overwritten, but a warning message is issued. 

\begin{filecontents*}{file name} [p] . . . . . . . . (D.2.9) – 448 

Is the same as the filecontents environment, except that no comment 

lines are written to the file. The file will contain exactly the 

contents of the environment, and nothing more. 

\begin{flushleft} . . . . . . . . . . . . . . . . . . . (4.2.2) – 67 

Start of the flushleft environment in which each line of text is 

left justified, that is, it begins flush with the left margin but is not 

expanded to match the right edge. The equivalent declaration is 

\raggedright. 

\begin{flushright} . . . . . . . . . . . . . . . . . . (4.2.2) – 67 

Start of the flushright environment in which each line of text is 

right justified, that is, the right-hand side is flush with the right 

margin, but the line is not expanded to start exactly at the left edge. 

The equivalent declaration is \raggedleft.


\begin{gather} [a] . . . . . . . . . . . . . . . . . (12.2.6) – 273 

With the amsmath package, switches to displayed math mode to 

produce several lines of equations, all centered with no alignment. 

Lines are terminated by \\ commands. Each line receives an equation 

number, unless the *-form of the environment has been selected. 

\begin{gathered}[pos] [m][a] . . . . . . . . . . . . (12.2.6) – 276 

With the amsmath package, is like the gather environment but is 

used as an element within math mode. The optional argument pos 

determines the vertical positioning relative to neighboring elements: 

t or b for top or bottom, no argument for centering. 

\begin{itemize} . . . . . . . . . . . . . . . . . . . . (4.3.1) – 69 

Start of the itemize environment for producing labeled, indented 

listings. The type of label depends on the depth of nesting; at the 

first level it is a • generated by each \item command. 

\begin{longtable} . . . . . . . . . . . . . . . . . . (4.8.4) – 108 

With the longtable package, begins a table with the same syntax as 

tabular but which will continue to other pages as necessary. The 

head and footlines on the continued pages can be specified, terminated 

with \endhead, \endfirsthead, \endfoot, \endlastfoot. 

\begin{letter}{recipient} . . . . . . . . . . . . . . (16.1) – 352 

Start of a letter with the document class letter. Name and address 

of the recipient are given within the second pair of brackets; lines of 

text within this argument are ended with the command \\. 

\begin{list}{standard label}{list decl} . . . . . . . . . (4.4) – 74 

Start of a generalized list environment. The label is defined by 

standard label, which is generated by each \item command. The 

desired list declarations are contained in list decl (see page 75). 

\begin{lrbox}{\boxname} . . . . . . . . . . . . . . . (4.7.1) – 87 

Functions in a way similar to the command \sbox, except that it is the 

text of the environment that is stored in the LR box named \boxname, 

which has previously been created with \newsavebox{\boxname}. 

The contents of the box may be printed as often as desired with 

\usebox{\boxname}. 

\begin{math} . . . . . . . . . . . . . . . . . . . . . . (5.1) – 119 

Switches from text to in-line math mode to produce formulas within 

a line of text. This environment has the same effect as \( or $ in text 

mode. 

\begin{matrix} [m][a] . . . . . . . . . . . . . . . . (12.2.4) – 266 

With the amsmath package, is the same as the array environment 

except that the column specifier argument may be omitted, without


which up to 10 centered columns may be entered. This maximum 

may be changed with the counter MaxMatrixCols. The environments 

pmatrix, bmatrix, Bmatrix, vmatrix, and Vmatrix function the 

same as matrix but are enclosed in braces (), [], {}, | | and , 

respectively. 

\begin{minipage}[pos][height][inner pos]{width} 

. . . . . . . . . . . . . . . . . . . . . . . . (4.7.3), (4.7.5) – 88, 90 

Environment to format text within a ‘minipage’ of width width. Its vertical 

positioning with respect to the surrounding text is determined 

by the optional argument pos: t for alignment with its top line, b 

with its bottom line, and centered with no argument. The other two 

optional arguments are: height to give the total height, and inner pos 

to specify how the text is to be positioned inside it. Possible values 

are t for top, b for bottom, c for centered, and s to be stretched 

out to fill the whole vertical space. The default is the value of the 

external positioning pos option. The height argument may contain 

the parameters \height, \depth, \width, and \totalheight. 

\begin{multicols}{num cols}[header][pre space] 

. . . . . . . . . . . . . . . . . . . . . . . . (3.2.8), (B.5.4) – 51, 395 

This environment is provided by the multicol package in the tools 

collection (Section B.5.4). It switches to printing the text in num cols 

columns, with header printed in one column across the top. A new 

page is inserted only if the remaining space on the current page 

is less than \premulticols or the optional argument pre space. 

A new page is inserted at the end if the remaining space is less 

than \postmulticols. The columns on the last page are balanced. 

Column separation and rule are set by the lengths \columnsep and 

\columnseprule. 

With the starred version multicols*, the columns of text are not 

balanced on the last page of the environment. 

\begin{multline} [a] . . . . . . . . . . . . . . . . (12.2.6) – 271 

With the amsmath package, switches to displayed math mode to 

produce a single equation over several lines. Line breaks are forced 

by \\ commands. The first line is to the far left, the last to the right, all 

others centered. With \shoveleft{form} and \shoveright{form}, 

single lines consisting of form may be pushed to the left or right. 

The single equation number appears at the right of the last line, or at 

the left of the first line, depending on class options reqno (default) 

and leqno, respectively. With the *-form of the environment, the 

equation number is suppressed. 

\begin{note} . . . . . . . . . . . . . . . . . . . . (15.1.2) – 326 

In slides class, the environment for producing a note for the current 

slide. Notes are numbered with the current slide number followed 

by a hyphen and running number, for example 8-1, 8-2, etc.


\begin{overlay} . . . . . . . . . . . . . . . . . . (15.1.2) – 325 

In slides class, the environment for producing an overlay for the 

current slide. Overlays are numbered with the current slide number 

followed by a lower case letter, for example 3-a, 3-b, etc. See also 

\begin{slide}. 

\begin{picture}(x dimen,y dimen) . . . . . . . . . (13.1.2) – 288 

Environment to generate a picture with the width x dimen and height 

y dimen, where the unit of length has previously been specified by 

the declaration \unitlength. 

\begin{picture}(x dimen,y dimen)(x offset,y offset) (13.1.6) – 301 

Most general form of the call to the picture environment. The 

picture is displaced to the left by x offset and downwards by y offset. 

\begin{pmatrix} [m][a] . . . . . . . . . . . . . . . (12.2.4) – 266 

Like the matrix environment, but enclosed in round parentheses (). 

\begin{quotation} . . . . . . . . . . . . . . . . . . . (4.2.3) – 67 

Start of the quotation environment in which text is indented on 

both sides relative to the normal page margins. Paragraphs within 

the environment are marked with an additional indentation of the 

first line. 

\begin{quote} . . . . . . . . . . . . . . . . . . . . . (4.2.3) – 67 

The same as the quotation environment except that the first line 

of a paragraph is not indented but instead additional line spacing 

comes between paragraphs. 

\begin{slide} . . . . . . . . . . . . . (15.1.2), (15.2.1) – 325, 331 

In slides and seminar classes, the main environment for producing 

a slide. 

\begin{sloppypar} . . . . . . . . . . . . . . . . . . . (2.8.3) – 36 

Inside this environment word spacings are allowed to stretch more 

generously than usual so that paragraphs are broken up into lines 

with fewer word divisions. See also \sloppy. The countercommand 

is \fussy. 

\begin{split} [m][a] . . . . . . . . . . . . . . . . (12.2.6) – 272 

With the amsmath package, is used within a math environment such 

as equation to write a formula over several lines. Line breaks are 

forced with \\ commands and the lines are horizontally aligned 

on the & alignment marker. Any equation number is generated by 

the outer environment. It is either centered with the class option 

centertags (default) or with tbtags it appears at the right of the 

last line, or at the left of the first line, depending on class options 

reqno (default) and leqno, respectively.


\begin{subarray}{pos}{1st line\\..\\last line} [m][a] (12.2.2) – 261 

With the amsmath package, sets multiline text for superscripts and 

subscripts, like \substack, but the parameter pos may take on values 

c or l for centered or left-justified lines. 

\begin{subequations} [a] . . . . . . . . . . . . . . (12.2.7) – 277 

With the amsmath package, numbers equations within it with a fixed 

main number and sequence of lower case letters attached, as 7a, 7b, 

7c, . . . . 

\begin{tabbing} . . . . . . . . . . . . . . . . . . . . (4.6.1) – 81 

Start of the tabbing environment in which special tabbing commands 

become operational: \= sets a tab stop, \> jumps to the next stop, 

\< goes back a stop, \\ terminates and starts a new line, \+ sets the 

left margin one tab stop further, \- moves the left margin back one 

stop. 

\begin{table}[loc] . . . . . . . . . . . . . (4.8.5), (7.1) – 109, 169 

Float environment for entering text for a table. The optional argument 

loc can be any combination of the letters h, t, b, and p to determine 

the various positioning possibilities. Default is tbp. The character 

! may additionally be given to ignore all float spacing and number 

restrictions set by the float style parameters. 

\begin{table*}[loc] . . . . . . . . . . . . . . . . . . (7.1) – 169 

The same as the table environment except that the table is inserted 

over the width of two columns when the option twocolumn or the 

command \twocolumn has been selected. The standard form table 

will only fill the width of one column. 

\begin{tabular}[pos]{cols} . . . . . . . . . . . . . . (4.8.1) – 95 

Start of the tabular environment for producing tables. The argument 

cols contains a formatting character for each column in the 

table: c for centered text, l for left, r for right justification, or p{wd} 

for a column of width wd in which the text may extend over several 

lines. 

When the entry @{text} appears between any two of the above column 

formatting characters, text is inserted in every row between those two 

columns. Where the character | appears, a vertical line is drawn in 

every row. 

The optional argument pos specifies how the table is to be vertically 

aligned with the surrounding text: with no argument, it is centered, 

otherwise with t the top line, with b the bottom line is aligned with 

the external baseline. 

The text entries of the individual columns are separated by & and the 

rows are terminated by \\.


\begin{tabular*}{width}[pos]{cols} . . . . . . . . . . (4.8.1) – 95 

The same as \begin{tabular} except that the total width of the 

table is given by the argument width. This may only be achieved 

successfully if there is rubber spacing between the columns. This 

may be added with @{\extracolsep\fill} somewhere within the 

cols format definition. 

\begin{thebibliography}{sample label} . . . . . . . (9.3.1) – 217 

Environment to generate a list of literature references. The sample 

label is the longest reference marker that will appear. Each entry 

in the bibliography starts with the command \bibitem which prints 

the marker for that entry; lines after the first are indented by an 

amount equal to the width of sample label. 

\begin{theindex} . . . . . . . . . . . . . . . . . . . (9.4.1) – 225 

Environment to produce a keyword index in two-column format. 

Entries are made with the \item, \subitem, \subsubitem, or 

\indexspace commands. 

\begin{theorem type}[extra title] . . . . . . . . . . . . (4.5) – 80 

Environment to invoke a theorem-like structure that has previously 

been defined by the user with the \newtheorem command. The name 

of the environment, theorem type, something like theorem or axiom, 

is the first argument of the \newtheorem command. The extra title 

is text that added after the name and number of the structure in () 

parentheses. 

\begin{titlepage} . . . . . . . . . . . . . . . . . . . (3.3.1) – 52 

Environment to produce a title page without a page number. The 

user has total control over the composition of this page. 

\begin{trivlist} . . . . . . . . . . . . . . . . . . . . (4.4.5) – 79 

Environment to generate a trivial list without a sample label and 

list declarations. The parameters \leftmargin, \labelwidth, and 

\itemsep are all set to 0 pt while \listparindent = \parindent 

and \parsep = \parskip. 

\begin{verbatim} . . . . . . . . . . . . . . . . . . . . (4.9) – 110 

Environment to print out source text, that is, as from a typewriter. 

Blank lines, line breaking, and commands are all output literally 

without any interpretation or formatting. 

\begin{verbatim*} . . . . . . . . . . . . . . . . . . . (4.9) – 110 

The same as the verbatim environment except that blanks are 

printed as to make them visible. 

\begin{verse} . . . . . . . . . . . . . . . . . . . . . (4.2.4) – 68 

Environment for setting rhymes, poems, verses, etc. Stanzas are 

separated by blank lines, individual lines by the \\ command.


\begin{vmatrix} [m][a] . . . . . . . . . . . . . . . (12.2.4) – 266 

Like the matrix environment, but enclosed in vertical lines | |. 

\begin{Vmatrix} [m][a] . . . . . . . . . . . . . . . (12.2.4) – 266 

Like the matrix environment, but enclosed in double vertical lines 

. 

\belowdisplayskip [m] . . . . . . . . . . . . . . . . (5.5.4) – 149 

Vertical spacing between a long displayed formula and the following 

text. A new value may be assigned with the \setlength command: 

\setlength{\belowdisplayskip}{\abovedisplayskip} 

sets \belowdisplayskip to the same value as \abovedisplayskip. 

See further examples under \abovedisplayskip. 

\belowdisplayshortskip [m] . . . . . . . . . . . . . (5.5.4) – 149 

Vertical spacing between a short displayed formula and the following 

text. Value is set with the \setlength command as in the above 

example. 

\beta [m] produces β . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\bfdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 

This command defines the series attribute that is selected with the 

\bfseries command. It may be redefined with \renewcommand: 

\renewcommand{\bfdefault}{b} 

\bfseries . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 

This declaration switches to a font in the current family and shape, 

but with the bold series attribute. 

\bibitem[label]{key} entry text . . . . . . (9.3.1), (9.3.4) – 217, 221 

Command to enter the text for a literature reference in the 

thebibliography environment. The reference word key is used 

in the main body of the text with citation commands to refer to this 

entry. In standard L AT E X, the bibliography list will be sequentially 

numbered, except for those entries with an optional label, in which 

case label replaces the number. In author–year bibliographies, the 

label must have a special form to transfer author and year texts to 

the citation commands. 

\bibliography{file list} . . . . . . . . . . (9.3.2), (14.1) – 219, 309 

For producing a bibliography with the aid of the BIBT E X program; 

file list is the root name of one or more files containing the literature 

databases to be searched. 

\bibliographystyle{style} . . . . . . . . (9.3.2), (14.1) – 219, 310 

In conjunction with the BIBT E X program, this command selects the 

style in which the bibliography entries are to be written. Choices for 

style are plain, unsrt, alpha, and abbrv, or plainnat, unsrtnat, 

abbrvnat with natbib. Many other contributed styles also exist.


\bibname . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the heading for the bibliography in book and 

report document classes. In English, this is ‘Bibliography’ but may 

be altered for adaptation to other languages. 

\bigbr symbol [m] . . . . . . . . . . . . . . . . . . . (5.5.3) – 148 

A bracket symbol larger than normal, but smaller than \Big. Example: 

\big(. 

\Bigbr symbol [m] . . . . . . . . . . . . . . . . . . . (5.5.3) – 148 

A bracket symbol larger than \big, but smaller than \bigg. Example: 

\Big[. 

\bigcap [m] produces . . . . . . . . . . . . . . . . (5.3.7) – 128 

\bigcirc [m] produces ○ . . . . . . . . . . . . . . . (5.3.3) – 125 

\bigcup [m] produces . . . . . . . . . . . . . . . . (5.3.7) – 128 

\biggbr symbol [m] . . . . . . . . . . . . . . . . . . (5.5.3) – 148 

A bracket symbol larger than \Big, but smaller than \Bigg. Example: 

\bigg|. 

\Biggbr symbol [m] . . . . . . . . . . . . . . . . . . (5.5.3) – 148 

The largest bracket symbol. Example: \Bigg\langle. 

\bigodot [m] produces . . . . . . . . . . . . . . . (5.3.7) – 128 

\bigoplus [m] produces . . . . . . . . . . . . . . . (5.3.7) – 128 

\bigotimes [m] produces . . . . . . . . . . . . . . (5.3.7) – 128 

\bigtriangledown [m] produces ▽ . . . . . . . . . . . (5.3.3) – 125 

\bigtriangleup [m] produces △ . . . . . . . . . . . . (5.3.3) – 125 

\bigskip . . . . . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

Inserts large vertical spacing of the amount \bigskipamount. See 

also \medskip and \smallskip. 

\bigskipamount 

Standard value for the amount of vertical spacing that is inserted 

with the command \bigskip. May be changed with the \setlength 


\setlength{\bigskipamount}{5ex plus1.5ex minus2ex} 

\bigsqcup [m] produces . . . . . . . . . . . . . . . (5.3.7) – 128 

\biguplus [m] produces . . . . . . . . . . . . . . . (5.3.7) – 128 

\bigvee [m] produces . . . . . . . . . . . . . . . . (5.3.7) – 128 

\bigwedge [m] produces . . . . . . . . . . . . . . . (5.3.7) – 128 

\binom{over}{under} [m][a] . . . . . . . . . . . . . (12.2.3) – 264 

With the amsmath package, prints a 

binomial expression: 

n 

\[\binom{n}{k}\] yields . 

k


\bmod [m] . . . . . . . . . . . . . . . . (5.3.8), (12.2.5) – 129, 268 

Command to produce the function name ‘mod’ in the form 

a\bmod b = a mod b 

\boldmath . . . . . . . . . . . . . . . . . . . . . . . (5.4.9) – 143 

Switches to bold face for math mode. This command must be given 

in text mode, however, before going into math mode. To set only part 

of a formula in bold face, use \mbox{\boldmath$...$} to return 

temporarily to text mode. 

\boldsymbol{symbol} [m][a] . . . . . . . . . . . . . (12.2.1) – 258 

When one of the packages amsmath or amsbsy has been loaded, this 

command prints symbol in bold face. Unlike \mathbf, it also affects 

math symbols and lower case Greek letters. 

\bot [m] produces ⊥ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\botfigrule . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 173 

A command that is executed before a float at the bottom of a page. 

It is normally defined to do nothing, but may be redefined to add a 

rule between the float and the main text. It must not add any net 

vertical spacing. 

\renewcommand{\botfigrule}{\vspace*{-.4pt} 

\rule{\columnwidth}{.4pt}} 

\bottomfraction . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

Maximum fraction of a page that may be taken up by floats at the 

bottom. May be set to a new value with: 

\renewcommand{\bottomfraction}{decimal frac}. 

bottomnumber . . . . . . . . . . . . . . . . . . . . . . (7.3) – 171 

Maximum number of floats that may appear at the bottom of a page. 

Set to a new number with \setcounter{bottomnumber}{num}. 

\bowtie [m] produces ⋈ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\Box [m] produces . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\boxed{formula} [m][a] . . . . . . . . . . . . . . . (12.2.5) – 270 

With the amsmath package, sets the mathematical formula in a box. 

\breve{x} [m] . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Breve accent over the math variable x: \breve{a} = ă. 

\Breve{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \breve, but with multiple 


\bullet [m] produces • . . . . . . . . . . . . . . . . (5.3.3) – 125 

\c{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces a cedilla under x: \c{C} = Ç.


\cap [m] produces ∩ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\caption[short form]{caption text} . . . . . . . . . . . (7.4) – 173 

Produces a numbered title or caption with the text caption text within 

the float environments figure or table. The short form is the 

abbreviated text appearing in the list of figures or tables, which is 

the same as the caption text if it is omitted. 

\captionslanguage . . . . . . . . . . . . (D.4.1), (11.1) – 459, 255 

A command used in several language adaptations to redefine the 

headings of special sections such as ‘Chapter’ and ‘Contents’. It 

occurs in packages like german as well as in the babel system. This 

command is normally part of the definition of the \selectlanguage 


\cc{list} . . . . . . . . . . . . . . . . . . . . . . . . (16.1) – 353 

Command within document class letter to generate ‘cc:’, copies, 

followed by a list of names list at the end of the letter. 

\ccname . . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command in the letter document class containing the word to be 

printed by the \cc command. In English, this is ‘cc’ but may be 

altered for adaptation to other languages. 

\cdot [m] produces · . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\cdots [m] produces · · · . . . . . . . . . . . . . . . . (5.2.6) – 123 

\centering . . . . . . . . . . . . . . . . . . . . . . . (4.2.1) – 67 

Declaration to switch to centered lines of text, each input line being 

terminated by \\. See also \begin{center}. 

\centerline{text} . . . . . . . . . . . . . . . . . . . . (4.2.1) – 67 

An additional T E X command that sets text centered on a horizontal 

line by itself. 

\cfrac[pos]{over}{under} [m][a] . . . . . . . . . . (12.2.3) – 265 

With the amsmath package, produces a continued fraction. The optional 

argument pos may be l or r to have the numerator left or right 

justified on the horizontal rule, otherwise it is centered. 

\chapter[short title]{title} . . . . . . . . . . . . . . . . (3.3.3) – 55 

Starts a new chapter on a new page, with an automatic sequential 

chapter number and title as header. If the optional short title is given, 

it appears in place of title in the table of contents and in the running 

head at the top of the pages. 

\chapter*{title} . . . . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

Starts a new chapter on a new page, with title as header, but without a 

chapter number. The entry does not appear in the table of contents.


\chaptername . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the chapter heading. In English, this is ‘Chapter’ 

but may be altered for adaptation to other languages. 

\check{x} [m] . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Háček accent over the math variable x: \check{a} = ǎ. 

\Check{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \check, but with multiple 


\CheckCommand{\com name}[narg][opt]{def } . . . . . (D.2.5) – 445 

Tests that the current definition of \com name is as expected. If not, 

an error message is issued. This is used to ensure that important 

commands have not been altered by other packages. 

\chi [m] produces χ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\circ [m] produces ◦ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\circle{diameter} . . . . . . . . . . . . . . . . . (13.1.4) – 295 

Picture element command to produce a circle with diameter diameter 

in the picture environment. To be used as an argument in a \put 

or \multiput command. 

\circle*{diameter} . . . . . . . . . . . . . . . . . (13.1.4) – 295 

Like \circle but produces a solid circle, filled in black. 

\cite[note]{key} . . . . . . . . . . . . . (9.3.3), (14.1) – 219, 310 

Literature citation using the identifier key to produce a reference 

label in the text. The optional note text is included with the label. 

\citep[pre-note][post-note]{key} . . . . . . . . . . . (9.3.4) – 221 

With the natbib package, inserts a parenthetical literature citation 

as ‘[Jones et al., 1999]’, using the identifier key. The optional note 

texts are included within the parentheses, before and after; if only 

one is present, it is a post-note. In numerical citation mode, the 

citation number is printed as ‘[21]’, just as with the standard \cite 

command. Multiple keys may be given. 

\citet{key} . . . . . . . . . . . . . . . . . . . . . . (9.3.4) – 221 

With the natbib package, inserts an in-text literature citation as 

‘Jones et al. [1999]’, using the identifier key. In numerical citation 

mode, it prints the author’s name before the citation number, as 

‘Jones et al. [21]’. 

\ClassError{class name}{error text}{help} [p] . . . . . (D.2.7) – 446 

Writes an error message error text to the monitor and transcript file, 

labeled with the class name, and halts processing, waiting for a user 

response as for a L AT E X error. If H〈return〉 is typed, the help text is 

printed. Both error text and help may contain \MessageBreak for a 

new line, \space for a forced space, and \protect before commands 

that are to have their names printed literally and not interpreted.


\ClassInfo{class name}{info text} [p] . . . . . . . . . (D.2.7) – 447 

Is like \ClassWarningNoLine except that the text info text is only 

written to the transcript file, and not to the monitor. 

\ClassWarning{class name}{warning text} [p] . . . . . (D.2.7) – 446 

Writes warning text to the monitor and transcript file, labeled with 

the class name and the current line number of the input file. Processing 

continues. The warning text is formatted in the same way as 

that for \ClassError. 

\ClassWarningNoLine{class name}{warning text} [p] . . (D.2.7) – 446 

Is like \ClassWarning except that the current line number of the 

input file is not printed. 

\cleardoublepage . . . . . . . . . . . . . . . . . . . . (2.7.4) – 34 

Ends the current page and outputs all unprocessed floats on to one 

or more float pages. The next page will be a right-hand one, with an 

odd page number. 

\clearpage . . . . . . . . . . . . . . . . . . . . . . . (2.7.4) – 33 

Ends the current page and outputs all unprocessed floats on to one 

or more float pages. 

\cline{n-m} . . . . . . . . . . . . . . . . . . . . . . (4.8.1) – 97 

In tabular environment, produces a horizontal rule from the beginning 

of column n to the end of column m. Example: \cline{2-5}. 

\closing{regards} . . . . . . . . . . . . . . . . . . . (16.1) – 353 

End of the text in the letter environment; regards stands for the 

desired terminating text. 

\clubsuit [m] produces ♣ . . . . . . . . . . . . . . . (5.3.6) – 127 

\color col spec . . . . . . . . . . . . . . . . . . . . . . (6.2) – 167 

A command made available with the color package. It is a declaration 

that switches the color in which the text is printed to that specified. 

It remains in effect until the end of the current environment or until 

countermanded by another \color command. The col spec is either 

the name of a color defined (or predefined) by \definecolor or 

of the form [model]{specs}, where the arguments have the same 

meaning as they do for \definecolor. Examples: 

\color[rgb]{0.5,0.5,0} \color{magenta} 

\colorbox col spec{text} . . . . . . . . . . . . . . . . . (6.2) – 167 

A command made available with the color package. The text is set 

in an LR box with the specified color as the background color. The 

col spec is the same as for \color.


\columnsep . . . . . . . . . . . . . . . . . . . . . . . (3.1.1) – 40 

Declaration for the amount of intercolumn spacing in two-column 

page formatting. May be changed with the \setlength command: 

\setlength{\columnsep}{1pt} 

\columnseprule . . . . . . . . . . . . . . . . . . . . . (3.1.1) – 40 

Declaration for the thickness of the vertical rule separating the 

columns in two-column page formatting. Value is set with the 

\setlength command: 

\setlength{\columnseprule}{1pt} 

\cong [m] produces ≅ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\contentsline{sec type}{\numberline{sec num}title text}{page} 

Command that appears in the .toc file for every entry in the table 

of contents, which is read when the \tableofcontents command 

is given. Such commands may be altered or added to the.toc file by 

means of the text editor. The entry sec type stands for the sectioning 

level, such as section, while sec num is its number (for example, 

2.3) and page is the page number where the entry appears. 

\contentsname . . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the heading for the table of contents. In English, 

this is ‘Contents’ but may be altered for adaptation to other 

languages. 

\coprod [m] produces . . . . . . . . . . . . . . . . (5.3.7) – 128 

\copyright produces © . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\cos [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘cos’ in formulas. 

\cosh [m] . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘cosh’ in formulas. 

\cot [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘cot’ in formulas. 

\coth [m] . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘coth’ in formulas. 

\csc [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘csc’ in formulas. 

\cup [m] produces ∪ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\CurrentOption [p] . . . . . . . . . . . . . . . . . . (D.2.3) – 443 

A command that may only be used in the definition of options, 

especially for default options. It contains the name of the option 

being processed.


\d{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces a ‘dot under’ accent: \d{o} = o . . 

\dag produces † . . . . . . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\dagger [m] produces † . . . . . . . . . . . . . . . . (5.3.3) – 125 

\dashbox{dash}(x dimen,y dimen)[pos]{text} . . . (13.1.4) – 291 

Picture element command to produce a dashed frame with width 

x dimen and height y dimen, using a dash length of dash in the 

picture environment. Without the optional pos, the contents text 

are centered within the frame, otherwise they are positioned at the 

left (l), right (r), top (t), or bottom (b), or a combination thereof, such 

as lt. This command is used as an argument in a \put or \multiput 


\dashv [m] produces ⊣ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\date{date text} . . . . . . . . . . . . . . . (3.3.1), (16.1) – 52, 353 

1. The command \maketitle normally prints the current date on 

the title page. The declaration \date will replace the date with 

whatever text is given in date text. 

2. Prints the text date text instead of the automatic current date 

in a letter. 

\datelanguage . . . . . . . . . . . . . . . . . . . . . (11.1) – 255 

A command used in several language adaptations to redefine the 

\today command according to the requirements of language. It occurs 

in german package as well as in the babel system. It may also be 

used for ‘dialects’, such a \dateUSenglish and \dateenglish. This 

command is normally part of the definition of the \selectlanguage 


\dblfigrule . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 173 

A command that is executed after a two-column float at the top of a 

page. It is normally defined to do nothing, but may be redefined to 

add a rule between the float and the main text. It must not add any 

net vertical spacing. 

\renewcommand{\dblfigrule}{\vspace*{-.4pt} 

\rule{\textwidth}{.4pt}} 

\dblfloatpagefraction . . . . . . . . . . . . . . . . . (7.3) – 172 

For two-column page formatting, the fraction of a float page that 

must be filled with floats before a new page is called. A new value is 

assigned with 

\renewcommand{\dblfloatpagefraction}{decimal frac} 

\dbinom{over}{under} [m][a] . . . . . . . . . . . . (12.2.3) – 264 

With the amsmath package, produces a binomial as \binom does, but 

in \displaystyle size.


\dblfloatsep . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

For two-column page formatting, the vertical spacing between floats 

that extend over both columns. A new value is set with the 


\setlength{\dblfloatsep}{12pt plus 2pt minus 4pt} 

\dbltextfloatsep . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

For two-column page formatting, the vertical spacing between floats 

extending over both columns at the top of the page and the following 

text. A new value is set with the \setlength command. 

\dbltopfraction . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

For two-column page formatting, the maximum fraction of a page that 

may be occupied at the top by floats extending over both columns. 

A new value is assigned with 

\renewcommand{\dbltopfraction}{decimal frac}. 

dbltopnumber . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

For two-column page formatting, the maximum number of floats that 

may appear at the top of a page extending over both columns. A new 

value is assigned with 

\setcounter{dbltopnumber}{num}. 

\ddag produces ‡ . . . . . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\ddagger [m] produces ‡ . . . . . . . . . . . . . . . . (5.3.3) – 125 

\dddot{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, a triple dot accent in math formulas: 

\dddot{a} = ... 

a. 

\ddddot{x} [m][a] . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, a four-dot accent in math formulas: 

\ddddot{a} = .... 

a . 

\ddot{x} [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

A double dot accent in mathematical formulas: \ddot{a} = ä. 

\Ddot{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \ddot, but with multiple 


\ddots [m] produces . . . . . . . . . . . . . . . . . . . (5.2.6) – 123 

\DeclareErrorFont{code}{fam}{ser}{shp}{sz} [p] . . (A.3.6) – 376 

If no valid font can be found, even after substituting the attributes 

as given by \DeclareFontSubstitution, the font declared with this 

command is selected as a last resort.


\DeclareFixedFont{\cmd}{code}{fam}{ser}{shp}{sz} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.2) – 373 

Defines \cmd to be a font declaration that selects a font with the 

fixed attributes given in the definition. 

\DeclareFontEncoding{code}{text set}{math set} [p] . (A.3.6) – 376 

Declares code to be the name of a new encoding scheme; text set is a 

set of commands that is to be executed when switching to text mode, 

math set a set for math mode. 

\DeclareFontEncodingDefaults{text set}{math set} [p] (A.3.6) – 376 

Declares the sets of commands to be executed by all encodings when 

switching to text and math modes; the additional commands specific 

to each encoding are executed afterwards. 

\DeclareFontFamily{code}{fam}{opt} [p] . . . . . . (A.3.6) – 376 

Declares fam to be a new font family with the encoding code; the 

encoding must previously have been declared. The commands in opt 

are executed every time this family is selected. 

\DeclareFontShape{code}{fam}{ser}{shp}{def }{opt} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.6) – 377 

Associates external font names with the given font attribute values; 

the actual definition def relates font sizes to font names, as explained 

on page 377. The commands in opt are executed every time this shape 

is selected. 

\DeclareFontSubstitution{code}{fam}{ser}{shp} [p] (A.3.6) – 376 

Declares the font attributes that are to be substituted in case there 

is no valid font corresponding to the set of attributes selected. The 

order of substitution is shape, series, and family; the encoding is 

never substituted. 

\DeclareGraphicsExtensions{ext list} [p] . . . . . . . (6.1.7) – 165 

Establishes the list of default extensions for graphics files that can be 

imported with the \includegraphics command and the graphics 

or graphicx packages; ext list is a comma-separated list of file extensions, 

such as .eps,.ps. 

\DeclareGraphicsRule{ext}{type}{bb}{cmd} [p] . . . (6.1.7) – 165 

Associates a graphics file extension ext with a graphics file type and a 

file extension (bb) where the bounding box information is to be read, 

and an operating command (cmd) that is to be executed on the file 

to make it available for importation. For example 

\DeclareGraphicsRule{.eps.gz}{eps} 

{.eps.bb}{‘gunzip -c #1} 

Here the command must be prefixed with ‘ and #1 represents the 

name of the file to be processed.


\DeclareMathAccent{\cmd}{type}{sym fnt}{pos} [p] . (A.3.4) – 375 

Declares \cmd to be a math accent command, printed with the character 

in position pos of the symbol font with the internal name 

sym fnt. The type is either \mathord or \mathalpha; in the latter 

case the symbol changes with math alphabet. 

\DeclareMathAlphabet{\cmd}{code}{fam}{ser}{shp} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.3) – 373 

Defines \cmd to be a math alphabet for setting letters in math mode; 

the font selected in all math versions is that with the specified font attributes. 

If a different font is to be invoked for certain math versions, 

they are defined individually with \SetMathAlphabet afterwards. 

However, if the shape attribute is left empty, the alphabet command 

is indeed created, but remains undefined in all versions, requiring an 

explicit \SetMathAlphabet declaration for each one. 

\DeclareMathDelimiter{\cmd}{type}{sym fnt1}{pos1} 

{sym fnt2}{pos2} [p] . . . . . . . . . . . . . . (A.3.4) – 375 

Declares \cmd to be a math delimiter in two sizes: the smaller 

variant is the character in position pos1 of the symbol font with the 

internal name sym fnt1, while the larger is from position pos2 of font 

sym fnt2. 

\DeclareMathOperator{\cmd}{name} [p][a] . . . . . (12.2.5) – 268 

With the amsopn or amsmath packages, defines \cmd to be a math 

mode command to print the text name as a function name in an 

upright font with appropriate spacing. With the *-form, the raising 

and lowering operators ˆ and _ produce limits, above or below the 

name. 

\DeclareMathRadical{\cmd}{sym fnt1}{pos1} 

{sym fnt2}{pos2} [p] . . . . . . . . . . . . . . (A.3.4) – 375 

Declares \cmd to be a math radical symbol: the smaller variant is the 

character in position pos1 of the symbol font with the internal name 

sym fnt1, while the larger is from position pos2 of font sym fnt2. 

\DeclareMathSizes{text}{math text}{script}{sscript} [p] (A.3.4) – 375 

Sets the point sizes for the three math styles \textstyle, 

\scriptstyle, and \scriptscriptstyle when the text is being 

printed in point size text. The unit pt is not included. 

\DeclareMathSymbol{\symbol}{type}{sym fnt}{pos} [p] (A.3.4) – 375 

Declares \symbol to be a command that prints the character in position 

pos with the symbol font with the internal name sym fnt. This 

same symbol is printed in all math alphabets, but may be different 

for other math versions if the symbol font has been \Set... to have 

different attributes in that math version.


\DeclareMathVersion{ver} [p] . . . . . . . . . . . . (A.3.3) – 374 

Declares ver to be a new math version for math alphabets and 

symbol fonts. Initially, the version will use those fonts defined 

by \Declare... commands, which may be redefined for this math 

version with appropriate \Set... commands. 

\DeclareOldFontCommand{\cmd}{text specs}{math specs} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.2) – 373 

Defines \cmd to be a font declaration that invokes text specs in text 

mode, and math specs in math mode. The new command behaves in 

math mode as a 2.09 font declaration, although math specs must be 

a math font command without its argument. It is meant to define 

commands to be compatible with L AT E X 2.09 and should generally be 

avoided. 

\DeclareOption{option}{code} [p] . . . . . . . . . . . (D.2.3) – 442 

In a class or package file, this command defines the set of commands 

(code) that is to be associated with the given option. These commands 

are executed when \ExecuteOptions or \ProcessOptions is called. 

After the latter, all definitions are erased, to save memory. The code 

is internally stored in a command named \ds@option. 

\DeclareOption*{code} [p] . . . . . . . . . . . . . . . (D.2.3) – 442 

In a class or package file, this command defines the default set of 

commands that is associated with every undefined option. Special 

commands that may be used within code are \CurrentOption (the 

name of the option) and \OptionNotUsed. Example: 

\DeclareOption*{\InputIfFileExists 

{\CurrentOption.sty}{}{\OptionNotUsed}} 

\DeclareRobustCommand{\cmd}[narg][opt]{def } . . . (D.2.5) – 445 

Defines or redefines the command \cmd in the same way as 

\newcommand except that the result is robust: it may be used in 

the argument of another command without a \protect command 

before it. 

\DeclareRobustCommand*{\cmd}[narg][opt]{def } . . (D.2.6) – 445 

The same as \DeclareRobustCommand except that the arguments to 

\cmd must be ‘short’, not containing any new paragraphs. 

\DeclareSymbolFont{sym fnt}{code}{fam}{ser}{shp} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.4) – 374 

Declares the math font with the given attributes to be a symbol font 

that may be addressed by other commands with the name sym fnt. 

The symbol font applies to all math versions unless redefined with a 

\SetSymbolFont command.


\DeclareSymbolFontAlphabet{\cmd}{sym fnt} [p] . . (A.3.4) – 375 

Defines \cmd to be a math alphabet that uses the font declared with 

a \DeclareSymbolFont command to have the internal name sym fnt. 

This is preferred over \DeclareMathAlphabet if there is a defined 

symbol font with the necessary attributes for the math alphabet. 

\DeclareTextAccent{\cmd}{code}{pos} [p] . . . . . . (A.3.7) – 379 

Defines \cmd to be an accent command when encoding code is active; 

it uses the symbol in font position pos as the accent. 

\DeclareTextAccentDefault{\cmd}{code} [p] . . . . (A.3.7) – 380 

Declares the default encoding that is to be taken if the accent command 

\cmd is called in an encoding for which it is not explicitly 

defined. 

\DeclareTextCommand{\cmd}{code}[narg][opt]{def } [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.7) – 379 

Defines \cmd in the same way as \newcommand except the definition 

is only valid when encoding code is active. 

\DeclareTextCommandDefault{\cmd}[narg][opt]{def } [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.7) – 380 

Creates a default definition for the command \cmd for all encodings 

for which it is not explicitly defined. 

\DeclareTextComposite{\cmd}{code}{let}{pos} [p] . (A.3.7) – 379 

Defines \cmd followed by the letter let to be the single character in 

font position pos when encoding code is active. It thus defines the 

action of accent commands in T1 encoding when the accented letter 

exists as a separate symbol. For example: 

\DeclareTextComposite{\’}{T1}{e}{233} 

The command must already have been defined for the encoding 

with either \DeclareTextAccent or \DeclareTextCommand (with 

one argument). 

\DeclareTextCompositeCommand{\cmd}{code}{let}{def } [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.7) – 379 

Is the same as \DeclareTextComposite except that any definition 

may be assigned to the command/letter combination, and not just a 

single symbol. 

\DeclareTextFontCommand{\cmd}{font specs} [p] . . . (A.3.2) – 373 

Defines \cmd to be a text font command that sets its argument with 

the font spec specifications. Example: 

\DeclareTextFontCommand{\textbf}{\bfseries} 

\DeclareTextSymbol{\cmd}{code}{pos} [p] . . . . . . (A.3.7) – 379 

Defines \cmd to print the symbol in font position pos when encoding 

code is active.


\DeclareTextSymbolDefault{\cmd}{code} [p] . . . . (A.3.7) – 380 

Declares the default encoding that is to be taken if the symbol command 

\cmd is called in an encoding for which it is not explicitly 

defined. 

\definecolor{name}{model}{specs} . . . . . . . . . . (6.2) – 167 

A command made available with the color package. It associates 

the name of a color (name) with the specifications specs according 

to a certain model. Possible values for model are rgb (red, green, 

blue), cmyk (cyan, magenta, yellow, black), gray, and named. In each 

case, specs is a comma-separated list of numbers between 0 and 1 

specifying the strength of the relevant component. In the case of the 

named model, specs is an internal name for the color that is known 

by the driver program. Examples: 

\definecolor{litegrn}{cmyk}{0.25,0,0.75.0} 

\definecolor{brown}{named}{RawSienna} 

A number of colors are predefined in all color drivers: red, green, 

blue, yellow, cyan, magenta, black, and white. 

\deg [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘deg’ in formulas. 

\DeleteShortVerb{\c} . . . . . . . . . . . . (4.9.1, B.5.3) – 111, 393 

When the standard package shortvrb has been loaded, this command 

counteracts the effects of a previous \MakeShortVerb{\c}, 

allowing the character c to have its original meaning once more. 

\Delta [m] produces ∆ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\delta [m] produces δ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\depth . . . . . . . . . . . . . . . . . . . . . . . . . . (4.7.1) – 86 

A length parameter equal to the depth of a box (baseline to bottom); it 

may only be used in the width specification of \makebox, \framebox, 

or \savebox, or in the height specification of a \parbox or minipage 

environment. 

\framebox[20\depth]{text} 

\det [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘det’ in formulas. Can be 

combined with a lower limit by means of the subscript command. 

\dfrac{numerator}{denominator} [m][a] . . . . . . . (12.2.3) – 264 

With the amsmath package, produces a fraction as \frac does, but 

in \displaystyle size. 

\DH . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character Ð.


\dh . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character ð. 

\Diamond [m] produces ♦ . . . . . . . . . . . . . . . (5.3.3) – 125 

\diamond [m] produces ⋄ . . . . . . . . . . . . . . . . (5.3.3) – 125 

\diamondsuit [m] produces ♦ . . . . . . . . . . . . . (5.3.6) – 127 

\dim [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘dim’ in formulas. 

\discretionary{before}{after}{without} . . . . . . . . (2.8.1) – 35 

Hyphenation suggestion within a word. The word may be divided 

such that before is at the end of one line, and after at the start of the 

next line. If no division occurs, without is printed. 

\displaybreak[num] [m][a] . . . . . . . . . . . . . (12.2.7) – 278 

With the amsmath package, allows a manual page break in a multiline 

math formula when given just before \\; if num is present, it takes 

on values of 0–4 with increasing encouragement for a break, where 4 

is the same as no value, a forced page break. Automatic page breaks 

are impossible in multiline formulas unless \allowdisplaybreaks 

has been issued in the preamble. 

\displaystyle [m] . . . . . . . . . . . . . . . . . . (5.5.2) – 146 

Switches to font size \displaystyle as the active font within a math 

formula. 

\div [m] produces ÷ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\DJ . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character Ð. 

\dj . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character đ. 

\documentclass[options]{class}[version] [p] . . . . . . (3.1) – 37 

Normally the first command in a L AT E X document, determining the 

overall characteristics. Standard values for class are: 

article, report, book, letter, and slides 

of which only one may be selected. In addition, various options may 

be chosen, their names separated by commas. Possibilities are: 

10pt, 11pt, 12pt, 

letterpaper, legalpaper, executivepaper, 

a4paper, a5paper, b5paper, landscape, 

onecolumn, twocolumn, 

oneside, twoside, 

notitlepage, titlepage, 

leqno, fleqn, openbib, 

draft, final


These and any additional options are all global, meaning that they 

also apply to any packages specified with a following \usepackage 


The optional version is a date, given in the form yyyy/mm/dd, as for 

example 1994/08/01. If the date of the class file read in is earlier 

than this, a warning message is printed. 

\doublebox{text} . . . . . . . . . . . . . . . . . . . . (4.7.9) – 94 

With the fancybox package, is a variant of \fbox, drawing a doubled 

box around text; the thicknesses and separation of the lines depend 

on the length \fboxrule. 

\dot{x} [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

A dot accent in mathematical formulas: \dot{a} = ˙a. 

\Dot{x} [m][a] . . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \dot, but with multiple 


\doteq [m] produces . = . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\dotfill . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 30 

Fills up the space in a line with a dotted leader: . . . . . . . = \dotfill. 

\dots produces . . . . . . . . . . . . . . . . . . . . . . (5.2.6) – 123 

\dots [m][a] . . . . . . . . . . . . . . . . . . . . . (12.2.2) – 264 

With the amsmath package, places continuation dots in math mode 

automatically at a height determined by the following symbol. 

\dotsb [m][a] produces dots for binary operator: · · · . (12.2.2) – 264 

\dotsc [m][a] produces dots for commas: . . . . . . . . (12.2.2) – 264 

\dotsi [m][a] produces dots for integral signs: · · · . . (12.2.2) – 264 

\dotsm [m][a] produces dots for multiplication: · · · . . (12.2.2) – 264 

\doublerulesep . . . . . . . . . . . . . . . . . . . . . (4.8.2) – 98 

The distance between double rules inside the tabular and array 

environments. New value assigned with \setlength outside of the 

environment. 

\Downarrow [m] produces ⇓ . . . . . . . . . . . . . . (5.3.5) – 127 

\downarrow [m] produces ↓ . . . . . . . . . . . . . . . (5.3.5) – 127 

\ell [m] produces ℓ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\em . . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.1) – 62 

This declaration switches to an emphatic font, one that has the 

current family and series, but with a different shape attribute. It 

normally toggles between an upright and an italic shape.


\emph{text} . . . . . . . . . . . . . . . . . . . . . . . (4.1.1) – 62 

This command sets its argument in an emphatic font, one that has 

the current family and series, but with a different shape attribute. It 

normally toggles between an upright and an italic shape. 

\emptyset [m] produces ∅ . . . . . . . . . . . . . . . (5.3.6) – 127 

\encl{enclosures} . . . . . . . . . . . . . . . . . . . (16.1) – 353 

Command in the document class letter to add the word ‘encl:’ with 

the list enclosures at the end of a letter. For International L AT E X, the 

actual word printed is contained in the command \enclname. 

\enclname . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command in the letter document class containing the word to be 

printed by the \encl command. In English, this is ‘encl’ but may be 


\end{environment} . . . . . . . . . . . . . . . . . . . (2.2) – 19 

Command to terminate an environment started with a command 

\begin{environment}. 

\endfirsthead . . . . . . . . . . . . . . . . . . . . (4.8.4) – 108 

Within the longtable environment of the longtable package, this 

command ends those specifications that are to be added to the top 

of the table (the head) on the first page only. 

\endfoot . . . . . . . . . . . . . . . . . . . . . . . (4.8.4) – 108 


command ends those specifications that are to be added at the bottom 

of the table (the foot) before it is continued to other pages. 

\endhead . . . . . . . . . . . . . . . . . . . . . . . (4.8.4) – 108 


command ends those specifications that are to be added to the top 

of the table (the head) when it is continued to other pages. 

\endlastfoot . . . . . . . . . . . . . . . . . . . . . (4.8.4) – 108 


command ends those specifications that are to be added at the bottom 

of the table (the foot) on the last page. 

\enlargethispage{size} . . . . . . . . . . . . . . . . . (2.7.4) – 34 

The \textheight parameter is temporarily increased by the length 

size, in order to improve a bad page break. On the following pages, 

\textheight will have its normal value once more. 

\enlargethispage* . . . . . . . . . . . . . . . . . . . (2.7.4) – 34 

Is the same as \enlargethispage except that any additional spacing 

between the lines is removed as necessary to maximize the amount 

of text on the page.


\ensuremath{math cmds} . . . . . . . . . . . . . . . (8.3.1) – 186 

Sets math cmds in math mode; may be called in both text and math 

modes. Its main use is to define new commands that require math 

mode but can be called from either mode. 

\epsilon [m] produces ɛ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\eqref{marker} [a] . . . . . . . . . . . . . . . . . (12.2.7) – 278 

With the amsmath package, is a variation on the \ref command, 

and prints the equation number defined with \label{marker} in 

parentheses, as (5.6). 

\equiv [m] produces ≡ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\eta [m] produces η . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\euro . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.8) – 25 

With the eurosym package, prints the euro symbol e from the eurosym 

METAFONT fonts. With the eurosans package, it prints the 

symbol from the Adobe (PostScript) euro fonts. In both cases, the 

symbol is sans serif but changes to bold face or slanted to match the 

current font. 

\EUR . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.8) – 25 

With the europs package, prints the euro symbol € from the Adobe 

(PostScript) euro fonts, such that it matches the font family and other 

attributes. \EURofc prints the invariable symbol €. 

\evensidemargin . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

Sets the left margin for the even-numbered pages. It is effective in 

the document class book and, when the option twoside has been 

selected, in the other classes. A new value is assigned with the 


\setlength{\evensidemargin}{2.5cm} 

\ExecuteOptions{option list} [p] . . . . . . . . . . . . (D.2.3) – 443 

In a class or package file, this command executes all the option 

definitions in option list. This is normally invoked just prior to 

\ProcessOptions to establish certain options as default. 

\exists [m] produces ∃ . . . . . . . . . . . . . . . . (5.3.6) – 127 

\exp [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘exp’ in formulas. 

\extracolsep{extra width} . . . . . . . . . . . . . . . (4.8.1) – 96 

Tabular command for setting extra spacing between all the following 

columns in a table. This command is inserted as an @-expression 

into the column definition field of the tabular environment: 

\begin{tabular}{lr@{\extracolsep{2.5mm}}lcr}


\fbox{text} produces a frame around text . . . . . . . . (4.7.1) – 86 

\fancypage{cmds1}{cmds2} . . . . . . . . . . . . . . . (4.7.9) – 95 

With the fancybox package, places a framed box around the contents 

of all subsequent pages; cmds1 exclude the head and footlines, cmds2 

includes them. The arguments set box parameters like \fboxrule 

but must end with a box command like \shadowbox. Usually one set 

of cmds is left blank. 

\fboxrule . . . . . . . . . . . . . . . . . . . . . . . . (4.7.8) – 93 

The line thickness for the frames drawn by \fbox and \framebox 

commands. A new value is assigned with the \setlength command: 

\setlength{\fboxrule}{1pt} 

\fboxsep . . . . . . . . . . . . . . . . . . . . . . . . (4.7.8) – 93 

The distance between the frame and text in the \fbox and \framebox 

commands. A new value is assigned with the \setlength command: 

\setlength{\fboxsep}{1mm} 

\fcolorbox col spec1 col spec2{text} . . . . . . . . . . . (6.2) – 167 

A command made available with the color package. Like \colorbox, 

the text is set in an LR box with the col spec2 as the background color, 

but with a frame of color col spec1 around it. The col specs are either 

both defined names or employ the same model. Examples: 

\fcolorbox[rgb]{1,0,0}{0,1,0}{Text} 

\fcolorbox{red}{green}{Text} 

\figurename . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the name for a figure caption. In English, this 

is ‘Figure’ but may be altered for adaptation to other languages. 

\fill . . . . . . . . . . . . . . . . . . . . . . . . . . (2.4.2) – 22 

A rubber length with a natural size of zero that can stretch to any 

size necessary to fill up the horizontal or vertical space available. 

\flat [m] produces ♭ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\floatpagefraction . . . . . . . . . . . . . . . . . . (7.3) – 172 

The fraction of a float page that must be filled with floats before a 

new page is called. A new value is assigned with 

\renewcommand{\floatpagefraction}{decimal frac} 

\floatsep . . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

The vertical spacing between floats that appear at the top or bottom 

of a page. A new value is set with the \setlength command: 

\setlength{\floatsep}{12pt plus 2pt minus 4pt} 

\flushbottom . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

A declaration that puts vertical spacing between paragraphs so that 

the last line on every page is at the same position. Standard for the 

book document class and for the twoside option.


\fnsymbol{counter} . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of the given counter as a ‘footnote symbol’: 

* † ‡ § ** †† ‡‡ 

\fontencoding{enc} . . . . . . . . . . . . . . . . . . (A.1) – 368 

This command selects the font encoding scheme. Possible values of 

enc are OT1 for the standard and T1 for the Cork encodings. Other 

values are also possible. 

\fontfamily{fam} . . . . . . . . . . . . . . . . . . . (A.1) – 368 

This command selects the ‘family’ of fonts. Possible values of fam 

for standard L AT E X with the Computer Modern fonts are cmr, cmss, 

cmtt, and cmfi. 

\fontseries{ser} . . . . . . . . . . . . . . . . . . . . (A.1) – 368 

This command selects the ‘series’ of fonts within a ‘family’. Possible 

values of ser for standard L AT E X are m (medium) and bx (bold extended). 

\fontshape{form} . . . . . . . . . . . . . . . . . . . . (A.1) – 368 

This command selects the ‘shape’ of fonts. Possible values of form are 

n (normal), it (italic), sl (slanted), sc (small caps), and u (‘unslanted’ 

italic). 

\fontsize{sz}{line sp} . . . . . . . . . . . . . . . . . (A.1) – 368 

This command selects the font size. The argument sz specifies the 

size of the characters in points (without the dimension pt) and line sp 

determines the value of the interline spacing (\baselineskip), with 

an explicit dimension. Example: \fontsize{12}{14pt}. 

\footnote[num]{footnote text} . . . . . (4.10.1), (4.10.2) – 112, 114 

Produces a footnote containing the text footnote tex. The optional 

argument num will be used as the footnote number in place of the 

next number in the automatic sequence. 

\footnotemark[num] . . . . . . . . . . . . . . . . (4.10.3) – 114 

Produces a footnote marker in the current text. The optional argument 

num will be used as the footnote number in place of the next 

number in the automatic sequence. May be used within structures 

where \footnote is not normally permitted, such as LR boxes, tables, 

math formulas. 

\footnoterule . . . . . . . . . . . . . . . . . . . (4.10.6) – 118 

This is an internal command to produce the horizontal rule between 

the regular text on a page and the footnote text at the bottom. May 

be changed with, e.g., 

\renewcommand{\footnoterule} 

{\rule{wdth}{hght}\vspace{-hght}}


\footnotesep . . . . . . . . . . . . . . . . . . . . (4.10.6) – 117 

The vertical spacing between two footnotes. A new value is assigned 

with the \setlength command: 

\setlength{\footnotesep}{6.5pt} 

\footnotesize . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \footnotesize, which is smaller than 

\small but larger than \scriptsize. 

\footnotetext[num]{footnote text} . . . . . . . . . (4.10.3) – 114 

Produces a footnote with the text footnote text but without generating 

a marker in the current text. The marker that is used for the 

footnote itself at the bottom of the page derives from the current 

value of the counter footnote, which remains unchanged, or from 

the value of the optional argument num. This command may be used 

together with the \footnotemark command to insert footnotes into 

structures where they are otherwise not allowed, such as LR boxes, 

tables, and math formulas. The \footnotetext command must be 

given outside of that structure. 

\footskip . . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The distance from the bottom edge of the text body to the lower edge 

of the footline. A value is assigned with the \setlength command: 

\setlength{\footskip}{25pt} 

\forall [m] produces ∀ . . . . . . . . . . . . . . . . (5.3.6) – 127 

\foreignlanguage{language}{text} . . . . . . . . . . (11.1) – 254 

In the babel system, sets a short text in the selected language. 

\frac{numerator}{denominator} [m] . . . . . . . . . (5.2.3) – 122 

Math command for generating fractions. 

\frame{text} . . . . . . . . . . . . . . . . . . . . . (13.1.4) – 298 

Produces a frame without any intervening spacing around text. 

Mainly used as a picture element in a \put or \multiput command 

within the picture environment. 

\framebox[width][pos]{text}} . . . . . . . . . . . . . (4.7.1) – 86 

Produces a frame of width width around text. By default, the text is 

centered within the frame, but may be left or right justified by giving 

the optional argument pos as l or r. It may also have the value s, to 

stretch the text to the given width. 

\framebox(x dimen,y dimen)[pos]{text} . . . . . . (13.1.4) – 291 

Picture element command to produce a frame of width x dimen and 

height y dimen within the picture environment. Without the optional 

argument pos, the text is centered vertically and horizontally.


The text may be left or right justified, and/or aligned at the top or 

bottom, by setting pos to a combination of the letters l, r, t, and b, 

such as tr for top, right; pos may also contain s to stretch the text 

to the full width. The command is to be used as the argument of a 

\put or \multiput command. 

\frenchspacing . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

After this command has been given, no additional horizontal spacing 

is inserted at the end of a sentence. The countercommand is 

\nonfrenchspacing. 

\frontmatter . . . . . . . . . . . . . . . . . . . . . . (3.3.5) – 57 

In the book class, introduces the material that comes at the beginning 

(preface, table of contents) by turning off the chapter numbering of 

the \chapter command and switching to Roman numbers for the 

pagination. 

\frown [m] produces ⌢ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\fussy . . . . . . . . . . . . . . . . . . . . . . . . . . (2.8.3) – 36 

Countercommand of \sloppy that allows larger interword spacings 

than normal. After \fussy has been given, the normal spacings apply 

once more. 

\Gamma [m] produces Γ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\gamma [m] produces γ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\gcd [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘gcd’ in formulas. A lower 

limit may be given as a subscript. 

\ge [m] produces ≥ . . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\genfrac{left}{right}{thkns}{mathsz}{over}{under} [m][a] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . (12.2.3) – 265 

With the amsmath package, produces a generalized fraction with 

delimiters left and right, line thickness thkns, math font size mathsz 

0–3, and with over on top of under. If mathsz is empty, the sizing is 

automatic. 

\geq [m] produces ≥ . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\gets [m] produces ← . . . . . . . . . . . . . . . . . (5.3.5) – 127 

\gg [m] produces ≫ . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\glossary{glossary entry} . . . . . . . . . . . . . . . (9.4.4) – 230 

Write a \glossaryentry command to the.glo file if \makeglossary 

has been issued in the preamble; else it does nothing. 

\glossaryentry{glossary entry}{page number} . . . . (9.4.4) – 230 

The form in which the entry is written to the .glo file by the 

\glossary command.


\graphpaper[num](x,y)(lx,ly) . . . . . . . . . . . (13.1.5) – 301 

A command added with the graphpap package for use in the picture 

environment. It plots a labeled grid system with the lower left corner 

at (x,y), lx wide and ly high. Grid lines are drawn every num units, 

with the fifth ones thicker. If num is not specified, it is assumed to 

be 10. All arguments must be integers, not decimal fractions. 

\grave{x} [m] . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

A grave accent over the math variable x: \grave{a} = à. 

\Grave{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \grave, but with multiple 


\guillemotleft . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol «. 

\guillemotright . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol ». 

\guilsinglleft . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol ‹. 

\guilsinglright . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol ›. 

\H{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Hungarian double acute accent: \H{o} = ő. 

\hat{x} [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Circumflex over the math variable x: \hat{a} = â. 

\Hat{x} [m][a] . . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \hat, but with multiple 


\hbar [m] produces ℏ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\headheight . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The height of the head at the top of each page. A new value is 

assigned with the \setlength command: 

\setlength{\headheight}{25pt} 

\headsep . . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

Vertical spacing between the lower edge of the page head and the 

top of the main text. A new value is assigned with the \setlength 


\setlength{\headsep}{0.25in}


\headtoname . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command in the letter document class containing the text that 

precedes the recipient’s name in the headline after the first page. 

In English, this is ‘To’ but may be altered for adaptation to other 

languages. 

\heartsuit [m] produces ♥ . . . . . . . . . . . . . . (5.3.6) – 127 

\height . . . . . . . . . . . . . . . . . . . . . . . . . (4.7.1) – 86 

A length parameter equal to the natural height of a box (distance from 

the baseline to the top); it may only be used in the width specification 

of \makebox, \framebox, or \savebox, or in the height specification 

of a \parbox or a minipage environment. 

\framebox[6\height]{text} 

\hfill . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 30 

A horizontal rubber spacing with a natural length of zero that can 

be stretched to any value. Used to fill up a horizontal line with blank 

spacing. This command is an abbreviation for \hspace{\fill}. 

\hline . . . . . . . . . . . . . . . . . . . . . . . . . . (4.8.1) – 97 

Produces a horizontal line in the array and tabular environments 

over the width of the entire table. 

\hoffset . . . . . . . . . . . . . . . . . . . . . . . . . 602, 603 

Horizontal offset of the output page from the printer border set by 

the printer driver. This printer border is normally 1 inch from the 

left edge of the paper. The standard value of \hoffset is 0 pt so 

that the left reference margin of the page is identical with the printer 

margin. A new value is assigned with the \setlength command: 

\setlength{\hoffset}{-1in} 

\hom [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘hom’ in formulas. 

\hookleftarrow [m] produces ←↪ . . . . . . . . . . . . (5.3.5) – 127 

\hookrightarrow [m] produces ↩→ . . . . . . . . . . . (5.3.5) – 127 

\hrulefill . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 30 

Fills up the space in a line with a rule: = \hrulefill. 

\hspace{width} . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

Produces horizontal spacing of length width. It is ignored if it occurs 

at the beginning or end of a line. 

\hspace*{width} . . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

Produces horizontal spacing of length width even at the beginning or 

end of a line.


\huge . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \huge, which is smaller than \Huge but 

larger than \LARGE. 

\Huge . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the largest font size available \Huge, which is larger than 

\huge. 

\hyperlink{name}{link} . . . . . . . . . . . . . . (10.2.4) – 247 

With the hyperref package, makes the contents of link to be a link to 

the target with the name name, established with the \hypertarget 

command. The link may be text, a symbol, or a graphics loaded with 

\includegraphics. The link is framed with a colored box, unless 

the option colorlinks has been set to true, in which case the link 

is set in a text color determined by the option linkcolor. 

\hypertarget{name}{text} . . . . . . . . . . . . . (10.2.4) – 247 

With the hyperref package, prints the text argument as normal, but 

makes it a target for internal links, named name. The text may be 

empty. 

\hypersetup{key = value, . . . } . . . . . . . . . . . . (10.2.4) – 241 

With the hyperref package, allows parameters (key) to be assigned 

values (value); the parameters may also be set as options in the 

\usepackage loading command, by with \hypersetup, these values 

may be changed within the document. The list of possible keys and 

values is given on pages 242–246. 

\hyphenation{hyphenation list} [p] . . . . . . . . . . . (2.8.2) – 35 

Sets up a list of hyphenation exceptions. The hyphenation list consists 

of a collection of words containing hyphens at the places where 

word division may occur: hy-phen-a-tion per-mit-ted. 

\i produces ı . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

\idotsint [m][a] produces ··· 

. . . . . . . . . . . (12.2.2) – 260 

\iff [m] produces ⇐⇒ . . . . . . . . . . . . . . . . . (5.3.5) – 127 

\IfFileExists{file name}{true}{false} . . . . . . . . (D.2.8) – 447 

Tests if the file file name can be found in the places where L AT E X looks 

for files; if so, the code true is executed; otherwise, false. This is like 

\InputIfFileExists except the file is not input. 

\iflanguage{language}{yes text}{no text} . . . . . . (11.1) – 254 

In the multilingual babel system, tests if language is the currently 

selected language and, if so, executes yes text, otherwise no text.


\[ifthen]ifthenelse{test}{then text}{else text} . . . . . (8.3.5) – 193 

A conditional command available when the standard package ifthen 

has been loaded. If the logical statement test evaluates to 〈true〉 

then then text is inserted, otherwise else text. The logical statement 

may be relational (two numbers with one of < = > between 

them), an even–odd test (\isodd{number}), a comparison 

of two texts (\equal{text1}{text2}), a comparison of two lengths 

(\lengthtest{length1 op length2}, op is one of < = >), or a test 

of a boolean switch (\boolean{switch}). Switches are created with 

\newboolean{switch} and set with \setboolean{switch}{value}, 

where value is true or false. Logical statements may be combined 

with logical operators \and, \or, and \not, and grouped with $ and 

$. 

\Im [m] produces ℑ . . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\imath [m] produces ı . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\in [m] produces ∈ . . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\include{file} . . . . . . . . . . . . . . . . . . . . . (9.1.2) – 209 

Inserts the contents of the file with the root name file and extension 

.tex into the current text at the point where the command appears. 

A new page is always started! Together with \includeonly, this 

command allows portions of the document to be processed as though 

the rest of the text were present. 

\includegraphics[llx,lly][urx,ury]{file name} . . . . . (6.1.2) – 155 

A command made available with the graphics package that imports 

external graphics stored in the file file name. The coordinates of the 

bounding box are given by llx, lly (lower left corner) and by urx, ury 

(upper right corner). It is the contents of this bounding box that 

are used for further manipulation: scaling, rotating. It is also the 

(manipulated) bounding box that is used to reserve space in the text; 

any graphics that extend beyond the limits of this box will also be 

printed, but overlapping any other material that may be beside it. 

The bounding box coordinates may have units attached to them; the 

default units are big points bp (72 per inch). 

If the bounding box coordinates are omitted, the information is 

obtained in some other manner, depending on the type of graphics 

file. For an encapsulated PostScript file, this information is taken 

from the graphics file itself. 

If llx and lly are not specified, they are assumed to be 0. That is, if 

only one set of optional coordinates are given, they refer to the upper 

right corner. 

\includegraphics*[llx,lly][urx,ury]{file name} . . . . (6.1.2) – 155 

The same as \includegraphics except that any graphics extending 

beyond the bounding box are not included: the figure is clipped.


\includegraphics[key=value,. . . ]{file name} . . . . . (6.1.3) – 157 

With the graphicx package, this command has a different syntax in 

which the scaling, rotating, clipping are effected through key=value 

pairs, like width=7cm, angle=90, scale=.5. 

\includeonly{file list} [p] . . . . . . . . . . . . . . . (9.1.2) – 209 

Only those files whose names are in file list, separated by commas, 

will be read in by the \include commands. The \include commands 

for other file names are ignored. Nevertheless, all the auxiliary files 

are read in so that the page and section numbers will be correct, as 

are the cross-reference markers. 

\indent . . . . . . . . . . . . . . . . . . . . . . . . . (3.2.4) – 46 

The first line of the next paragraph is to be indented. 

\index{index entry} . . . . . . . . . . . . (9.4.2), (9.4.2) – 225, 226 

Writes a \indexentry command to the .idx file if the \makeindex 

command has been issued in the preamble; otherwise it does nothing. 

The MakeIndex program (Section 9.4.3) can process this file if the 

entries are in the forms 

\index{main entry} 

\index{main entry!sub entry} 

\index{main entry!sub entry!sub sub entry} 

making up a theindex environment with the entries alphabetically 

ordered and organized with \item, \subitem, and \subsubitem 


\indexentry{index entry}{page number} . . . . . . . (9.4.2) – 226 

The form in which the entry is written to the.idx file by the \index 


\indexname . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the heading for the index. In English, this is 

‘Index’ but may be altered for adaptation to other languages. 

\indexspace . . . . . . . . . . . . . . . . . . . . . . (9.4.1) – 225 

Command within theindex environment to produce a blank line. 

\inf [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘inf’ in formulas. A lower 

limit may be set as a subscript. 

\infty [m] produces ∞ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\intertext{insert text} [m][a] . . . . . . . . . . . . (12.2.1) – 259 

When the package amsmath is loaded, this command inserts text as 

a left-justified line between lines of an equation, without affecting 

their horizontal alignment.


\input{file} . . . . . . . . . . . . . . . . . . . . . . (9.1.1) – 207 

Inserts the contents of the file with the root name file and extension 

.tex into the current text at the point where the command appears. 

The file that is read in may also contain further \input commands. 

\InputIfFileExists{file name}{true}{false} . . . . . (D.2.8) – 447 

Tests if the file file name can be found in the places where LAT EX looks for files; if so, the code true is executed and the file is input; 

otherwise, false is executed. 

\int [m] produces 

. . . . . . . . . . . . . . . . . . (5.2.5) – 123 

\iint [m][a] produces 

. . . . . . . . . . . . . . . (12.2.2) – 260 

\iiint [m][a] produces 

. . . . . . . . . . . . . . (12.2.2) – 260 

\iiiint [m][a] produces 

. . . . . . . . . . . . . (12.2.2) – 260 

\intextsep . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

The vertical spacing between floats in the middle of a page and the 

surrounding text. A new value is assigned with the \setlength 


\setlength{\intextsep}{10pt plus2pt minus3pt} 

\invisible . . . . . . . . . . . . . . . . . . . . . (15.1.2) – 326 

In slides class, a declaration that makes the following text be printed 

in ‘invisible ink’, that is, it takes up as much space as though it were 

there. It remains in effect until the end of the environment, or end of 

the curly braces, in which it was issued, or until \visible is given. 

It is used for making overlays. 

\iota [m] produces ι . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\itdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 

This command defines the shape attribute that is selected with the 

\itshape command. It may be redefined with \renewcommand: 

\renewcommand{\itdefault}{it} 

\item[label] . . . . . . . . . . . . . . . . . . (4.3), (4.4.1) – 69, 74 

Produces a label and the start of an item text in a list environment. 

Without the optional argument, the label is generated according to 

the type of environment, for example numbers for the enumerate 

environment. The optional argument inserts label in place of this 

standard item label. 

\item . . . . . . . . . . . . . . . . . . . . . . . . . (9.4.1) – 225 

Produces a main entry in theindex environment. 

\itemindent . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 77 

The amount by which the label and the text of the first line after each 

\item is indented in a list environment. The standard value is 0 pt, 

but a new value may be assigned with the \setlength command: 

\setlength{\itemindent}{1em}


\itemsep . . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 75 

The amount of vertical spacing in addition to \parsep that is inserted 

between the \item texts in a list environment. A new value may be 


\setlength{\itemsep}{2pt plus1pt minus1pt} 

\itshape . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 

This declaration switches to a font in the current family and series, 

but with the italic shape attribute. 

\j produces j . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

\jmath [m] produces j . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\Join [m] produces ⋈ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\jot . . . . . . . . . . . . . . . . . . . . . . . . . . (5.5.4) – 149 

The amount of vertical spacing between the formula lines of an 

eqnarray or eqnarray* environment. Standard value is 3 pt. A new 

value may be assigned with the \setlength command: 

\setlength{\jot}{4.5pt} 

\k{x} . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the ogonek accent \k{A} = Ą. 

\kappa [m] produces κ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\ker [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘ker’ in formulas. 

\kill . . . . . . . . . . . . . . . . . . . . . . . . . . (4.6.2) – 82 

Removes the preceding sample line in a tabbing environment that 

was given only to set the tabs and not to be printed at this point. 

\L produces Ł . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\l produces ł . . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\label{marker} . . . . . . . . . . . . . . . . . . . . (9.2.1) – 213 

Sets a marker in the text at this position with the name marker. 

It may be referred to either earlier or later in the document with 

the command \ref{marker} to output the counter that was then 

current, such as the section, equation, or figure number, or with the 

command \pageref{marker} to print the page number where the 

marker was set. 

\labelenum n . . . . . . . . . . . . . . . . . . . . . . (4.3.5) – 73 

A set of commands to produce the standard labels for the nesting 

levels of the enumerate environments, where n is one of i, ii, iii, 

or iv. For example, 

\renewcommand{\labelenumii}{\arabic{enumii}.)} 

changes the standard labels of the second-level enumerate environment 

to be 1.), 2.), etc.


\labelitem n . . . . . . . . . . . . . . . . . . . . . . (4.3.5) – 73 

A set of commands to produce the standard labels for the nesting 

levels of the itemize environments, where n is one of i, ii, iii, or 

iv. For example, 

\renewcommand{\labelitemi}{$\Rightarrow$} 

changes the standard labels of the outermost itemize environment 

to ⇒. 

\labelsep . . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 77 

In a list environment, the distance between the label box and the 

list text. A new value is assigned with the \setlength command: 

\setlength{\labelsep}{5pt} 

\labelwidth . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 77 

In a list environment, the width of the box reserved for the label. A 

new value is assigned with the \setlength command: 

\setlength{labelwidth}{2.2cm} 

\Lambda [m] produces Λ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\lambda [m] produces λ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\langle [m] produces 〈 . . . . . . . . . . . . . . . . . (5.4.1) – 132 

\language{num} . . . . . . . . . . . . . . . . . . . (11.1) – 255 

In T E X versions 3.0 and later, the set of hyphenation patterns number 

num is made active. The patterns must be previously loaded into 

the format file by an initex run in which \language{num} was given 

before those patterns were read in. 

\large . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \large, which is smaller than \Large but 

larger than \normalsize. 

\Large . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \Large, which is smaller than \LARGE but 

larger than \large. 

\LARGE . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \LARGE, which is smaller than \huge but 

larger than \Large. 

\LaTeX produces L AT E X . . . . . . . . . . . . . . . . . . . (2.1) – 18 

\LaTeXe produces L AT E X 2ε . . . . . . . . . . . . . . . . . (2.1) – 19 

\lceil [m] produces ⌈ . . . . . . . . . . . . . . . . . (5.4.1) – 132 

\ldots produces . . . . . . . . . . . . . . . . . . . . . (5.2.6) – 123 

\le [m] produces ≤ . . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\leadsto [m] produces ⇝ . . . . . . . . . . . . . . . (5.3.5) – 127


\leftlbrack [m] . . . . . . . . . . . . . . . . . . . . (5.4.1) – 131 

Adjusts the size of the bracket symbol lbrack to fit the height of 

the formula between the \left ... \right pair. For example, 

\left[. If there is to be no matching bracket, the \left and \right 

commands must still be given to specify the part of the formula to 

be sized, but the missing bracket is given as a period (for example, 

\right.). 

\Leftarrow [m] produces ⇐ . . . . . . . . . . . . . . (5.3.5) – 127 

\leftarrow [m] produces ← . . . . . . . . . . . . . . (5.3.5) – 127 

\leftharpoondown [m] produces ↽ . . . . . . . . . . . (5.3.5) – 127 

\leftharpoonup [m] produces ↼ . . . . . . . . . . . . (5.3.5) – 127 

\lefteqn [m] . . . . . . . . . . . . . . . . . . . . . (5.4.7) – 140 

A command inside the eqnarray environment that outputs its argument 

as though it had zero width, thus having no effect on the 

column widths. It is used mainly for the first row of a multi-row 

formula. 

\leftmargin . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 76 

In a list environment, the amount by which the left edge of the text 

is indented relative to the surrounding text. A new value is assigned 

with the \setlength command. For nested list environments, 

different values for the indentation can be specified by adding i 

. . . vi to the declaration name, such as 

\setlength{\leftmarginiii}{0.5cm} 

See also . . . . . . . . . . . . . . . . . . . . . . (4.3.4) – 71. 

\Leftrightarrow [m] produces ⇔ . . . . . . . . . . . (5.3.5) – 127 

\leftrightarrow [m] produces ↔ . . . . . . . . . . . (5.3.5) – 127 

\leftroot{shift} [m][a] . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, used in the index to a \sqrt command 

to shift it slightly to the left. The shift is a number specifying how 

many units to move it. Example: 

\sqrt[\leftroot{-1}\uproot{3}\beta]{k} 

\leq [m] produces ≤ . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\lfloor [m] produces ⌊ . . . . . . . . . . . . . . . . (5.4.1) – 132 

\lg [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘lg’ in formulas. 

\lhd [m] produces ⊳ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\lim [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘lim’ in formulas. A lower 


\liminf [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘lim inf’ in formulas. A 

lower limit may be set as a subscript.


\limits [m] . . . . . . . . . . . . . . . . (5.2.5), (5.3.7) – 123, 128 

Places the upper and lower limits above and below the appropriate 

symbols where these would normally go just after them. 

\limsup [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘lim sup’ in formulas. A 

lower limit may be set as a subscript. 

\line(∆x,∆y){length} . . . . . . . . . . . . . . . (13.1.4) – 293 

A picture element command within a picture environment for drawing 

horizontal and vertical lines of any length as well as slanted lines 

at a limited number of angles. For horizontal and vertical lines, the 

length argument is the actual length in units of \unitlength. For 

slanted lines, length is the length of the projection on to the x-axis 

(horizontal displacement). The slope is determined by the (∆x, ∆y) 

arguments, which take on integral values such that −6 ≤ ∆x ≤ 6 

and −6 ≤ ∆y ≤ 6. This command is the argument of a \put or 

\multiput command. 

\linebreak[n] . . . . . . . . . . . . . . . . . . . . . (2.7.2) – 31 

A recommendation to break the line of text at this point such that 

it fills the horizontal space available (left and right justified). The 

urgency of the recommendation is given by the integral number 

n between 0 and 4, with the higher numbers meaning a stronger 

recommendation. A value of 4 is the same as the command without 

the optional argument and means an obligatory line break. 

\linethickness{thickness} . . . . . . . . . . . . . (13.1.4) – 300 

Sets the thickness of the horizontal and vertical lines in the picture 

environment. The argument thickness is a length specification with 

units, for example, 1.2mm. 

\linewidth . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 47 

A length that is set to the current text line width, whether in one 

column of a two-column page, in a minipage or parbox. This must 

never be changed, but is used when that width is needed, say to set 

the width of an included graphics with 

\includegraphics[width=0.8\linewidth]{..} 

\listfigurename . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the heading for the list of figures. In English, 

this is ‘List of Figures’ but may be altered for adaptation to other 

languages. 

\listfiles [p] . . . . . . . . . . . . . . . (9.1.1), (D.2.9) – 208, 447 

When given in the preamble, causes a list of all files read in during 

the processing to be printed to the monitor and to the transcript file 

at the end of the run. The list includes version number, date, and 

any additional information entered with one of the \Provides... 



\listoffigures . . . . . . . . . . . . . . . . . . . . . (3.4.4) – 59 

Produces a list of figures with the entries from all the \caption 

commands in figure environments. 

\listoftables . . . . . . . . . . . . . . . . . . . . . (3.4.4) – 59 

Produces a list of tables with the entries from all the \caption 

commands in table environments. 

\listparindent . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 77 

Depth of indentation for the first line of a paragraph inside a list 

environment. A new value may be assigned with the \setlength 


\setlength{\listparindent}{1em} 

\listtablename . . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the heading for the list of tables. In English, 

this is ‘List of Tables’ but may be altered for adaptation to other 

languages. 

\ll [m] produces ≪ . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\ln [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘ln’ in formulas. 

\LoadClass[options]{class}[version] [p] . . . . . . . . (D.2.2) – 442 

This command may only be invoked within a class file to load another 

class file. It may only be called once within any class file. The file 

loaded must have the extension .cls. Any options specified in the 

\documentclass command are not passed over as global options. 


example 1994/08/01. If the date of the class file is earlier than this, 

a warning message is printed. 

\LoadClassWithOptions{class}[version] [p] . . . . . . (D.2.2) – 442 

Like \LoadClass except all the currently specified options are automatically 

passed to class. 

\location{number} . . . . . . . . . . . . . . . . . . (16.1) – 352 

In the letter document class, enters the sender’s room number. In 

the standard L AT E X letter class, number is only output if \address 

has not been called. It is intended to be used in company letterheads. 

\log [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘log’ in formulas.


\Longleftarrow [m] produces ⇐= . . . . . . . . . . . (5.3.5) – 127 

\longleftarrow [m] produces ←− . . . . . . . . . . . (5.3.5) – 127 

\Longleftrightarrow [m] produces ⇐⇒ . . . . . . . . (5.3.5) – 127 

\longleftrightarrow [m] produces ←→ . . . . . . . . (5.3.5) – 127 

\longmapsto [m] produces ↦−→ . . . . . . . . . . . . . (5.3.5) – 127 

\Longrightarrow [m] produces =⇒ . . . . . . . . . . . (5.3.5) – 127 

\longrightarrow [m] produces −→ . . . . . . . . . . . (5.3.5) – 127 

\lq produces ‘, identical to the ‘ symbol. 

\lvert [m][a] produces | (left delimiter) . . . . . . . . (12.2.5) – 270 

\lVert [m][a] produces (left delimiter) . . . . . . . . (12.2.5) – 270 

\mainmatter . . . . . . . . . . . . . . . . . . . . . . . (3.3.5) – 57 

In the book class, introduces the main body of text after the front 

matter, by resetting the page numbering to 1 with Arabic numbers, 

and by reactivating the chapter numbering with the \chapter command. 

It undoes the effects of \frontmatter. 

\makebox[width][pos]{text} . . . . . . . . . . . . . . . (4.7.1) – 86 

Produces a box of width width containing text centered horizontally, 

unless pos is given to specify that it is to be left (l) or right (r) 

justified. It may also have the value s, to stretch the text to width. 

\makebox(x dimen,y dimen)[pos]{text} . . . . . . . (13.1.4) – 291 

Picture element command to produce a box of width x dimen and 

height y dimen within the picture environment. Without the optional 

argument pos, the text is centered vertically and horizontally. 

The text may be left or right justified, and/or aligned at the top or 

bottom, by setting pos to a combination of the letters l, r, t, and b, 

such as tr for top, right; pos may also contain s to stretch the text 

to the full width. The command is used as the argument of a \put 

or \multiput command. 

\makeglossary [p] . . . . . . . . . . . . . . . . . . . (9.4.4) – 230 

Command to activate the \glossary commands in the text. 

\makeindex [p] . . . . . . . . . . . . . . . . . . . . . (9.4.2) – 226 

Command to activate the \index commands in the text. 

\makelabel{text} . . . . . . . . . . . . . . . . . . . . (4.4.1) – 75 

An internal command that is called by the \item command to produce 

the actual label text within a list environment. 

\makelabels . . . . . . . . . . . . . . . . . . . . . . (16.1) – 355 

Produces address labels in the letter document class using the 

entries from the \begin{letter} environment. 

\MakeLowercase{text cmd} . . . . . . . . . . . . . . (D.3.2) – 455 

Converts text cmd (text and commands) to lower case.


\MakeShortVerb{\c} . . . . . . . . . . . . . (4.9.1, B.5.3) – 111, 393 

When the standard package shortvrb has been loaded, this command 

makes the character c a shorthand form for \verbc: everything 

that appears between two occurrences of c is printed literally, 

in typewriter type. With \DeleteShortVerb{\c}, the character is 

restored to its normal meaning. Example: \MakeShortVerb{\|} 

\maketitle . . . . . . . . . . . . . . . . . . . . . . . (3.3.1) – 54 

Produces a title page using entries in the \author and \title commands, 

and optionally those in the \date and \thanks commands. 

\MakeUppercase{text cmd} . . . . . . . . . . . . . . (D.3.2) – 455 

Converts text cmd (text and commands) to upper case. 

\mapsto [m] produces ↦→ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\marginpar[left text]{right text} . . . . . . . . . . . (4.10.5) – 117 

Produces a marginal note at the right of the text containing right text. 

With two-sided formatting, the marginal note goes into the left margin 

on the even pages, in which case the optional left text will be written 

instead. For two-column text, the marginal notes always go into the 

‘outer’ margin, and again left text will be used for the left margin. 

\marginparpush . . . . . . . . . . . . . . . . . . . (4.10.6) – 118 

The minimum vertical separation between two marginal notes. A 

new value may be assigned with the \setlength command. 

\marginparsep . . . . . . . . . . . . . . . . . . . (4.10.6) – 118 

The spacing between the edge of the text and a marginal note. A new 

value may be assigned with the \setlength command. 

\marginparwidth . . . . . . . . . . . . . . . . . . (4.10.6) – 118 

The width of the box reserved for marginal notes. A new value may 

be assigned with the \setlength command. 

\markboth{left head}{right head} . . . . . . . . . . . . (3.2.1) – 43 

Sets the text entries for the left and right page headlines in twosided 

formatting when the page style myheadings has been selected, 

or when the automatic entries of page style headings are to be 

changed. 

\markright{head} . . . . . . . . . . . . . . . . . . . . (3.2.1) – 43 

Sets the text entry for the page headline when the page style 

myheadings has been selected, or when the automatic entry of page 

style headings is to be manually changed. In two-sided formatting, 

only the right headline is set with this command. 

\mathbf{text} [m] . . . . . . . . . . . . . (5.4.2), (A.3.3) – 133, 373 

This command sets text in a bold font (\bfseries) within math 

mode. Spaces are ignored as usual.


\mathcal{text} [m] . . . . . . . (5.3.2), (5.4.2), (A.3.3) – 125, 133, 373 

This command sets text in calligraphic letters within math mode: 

$\mathcal{ABC}$ = ABC. 

\mathindent . . . . . . . . . . . . . . . . . . . . . . . (3.1.1) – 40 

The indentation of displayed formulas from the left margin when the 

option fleqn has been selected. A new value may be assigned with 

the \setlength command: 

\setlength{\mathindent}{25pt} 

\mathit{text} [m] . . . . . . . . . . . . . (5.4.2), (A.3.3) – 133, 373 

This command sets text in a text italic font (\itshape) within math 

mode. It differs from \mathnormal in that the spacing between the 

letters is as in regular text. Compare mathit and mathnor mal. 

\mathnormal{text} [m] . . . . . (5.3.1), (5.4.2), (A.3.3) – 125, 133, 373 

This command sets text in the normal (italic) math font within math 

mode. In this font, capital Greek letters are also set in italics: 

$\Gamma\mathnormal{\Gamma}$ = Γ Γ . 

\mathring{x} [m] . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

A ring accent in mathematical formulas: \mathring{a} = ˚a. 

\mathrm{text} [m] . . . . . . . . . . . . . (5.4.2), (A.3.3) – 133, 373 

This command sets text in a Roman font (\rmfamily) within math 

mode. Spaces are ignored as usual. 

\mathsf{text} [m] . . . . . . . . . . . . . (5.4.2), (A.3.3) – 133, 373 

This command sets text in a sans serif font (\sffamily) within math 


\mathtt{text} [m] . . . . . . . . . . . . . (5.4.2), (A.3.3) – 133, 373 

This command sets text in a typewriter font (\ttfamily) within math 


\mathversion{ver} . . . . . . . . . . . . . . . . . . (A.3.3) – 373 

Selects the current math version. Possible values are bold and 

normal; new values may be created with the \DeclareMathVersion 


\max [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘max’ in formulas. A lower 


\mbox{text} produces an LR box around text . . . . . . . (4.7.1) – 86 

\mddefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 

This command defines the series attribute that is selected with the 

\mdseries command. It may be redefined with \renewcommand: 

\renewcommand{\mddefault}{m}


\mdseries . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 

This declaration switches to a font in the current family and shape, 

but with the medium series attribute. 

\medskip . . . . . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

Inserts large vertical spacing of the amount \medskipamount. See 

also \bigskip and \smallskip. 

\medskipamount 

Standard value for the amount of vertical spacing that is inserted 

with the command \medskip. May be changed with the \setlength 


\setlength{\medskipamount}{3ex plus1ex minus1ex} 

\medspace [m][a] . . . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this is an alias for \:, a medium space in 

a math formula. 

\MessageBreak . . . . . . . . . . . . . . . . . . . . (D.2.7) – 446 

Forces a new line in the texts of error, warning, and information messages. 

These are the only places where it may be invoked, otherwise 

it does nothing. 

\mho [m] produces . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\mid [m] produces | . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\min [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘min’ in formulas. A lower 


\mod{arg} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.5) – 268 

With the amsopn or amsmath packages, command to produce the 

function name ‘mod’ in formulas in the form: 

y\mod{a+b} = y mod a + b 

\models [m] produces ⊨ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\mp [m] produces ∓ . . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\mspace{mu} [m][a] . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, inserts spacing in math formulas; mu is 

a math space in units of mu (=1/18 em): \mspace{-4mu}. 

\mu [m] produces µ . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\multicolumn{n}{col}{text} . . . . . . . . . . . . . . . (4.8.1) – 97 

Merges the next n columns in the array and tabular environments, 

formatting the text entry text according to the single column definition 

col, which may be l, c, r as well as |.


\multiput(x,y)(∆x,∆y){n}{pic elem} . . . . . . . (13.1.3) – 289 

Multiple positioning command in the picture environment. The 

object pic elem is placed n times, at (x, y), (x +∆x, y +∆y), . . . (x + 

(n − 1)∆x, y + (n − 1)∆y). 

\multlinegap [m][a] . . . . . . . . . . . . . . . . . (12.2.6) – 271 

A length that determines the left and right margins of formulas 

produced with the A M S-L AT E X multline environment; initial value is 

10 pt but may be reset by the user. 

\nabla [m] produces ∇ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\name{sender} . . . . . . . . . . . . . . . . . . . . . (16.2) – 356 

In the letter document class, enters the sender’s name. 

\natural [m] produces ♮ . . . . . . . . . . . . . . . . (5.3.6) – 127 

\nearrow [m] produces ↗ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\NeedsTeXFormat{format}[version] [p] . . . . . . . . (D.2.1) – 440 

Declares the T E X format that is necessary for processing the file. This 

should be the first statement in the file. At the moment, the only 

legitimate value for format is LaTeX2e. The version, if included, must 

be given as a date in the form yyyy/mm/dd, specifying the earliest 

possible release date of the format that is consistent with all the 

features employed in the file. Example: 


\neg [m] produces ¬ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\negmedspace [m][a] . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this inserts a negative medium space in a 

math formula. 

\negthickspace [m][a] . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this inserts a negative thick space in a 

math formula. 

\negthinspace [m][a] . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this is an alias for \!, a negative thin 

space in a math formula. 

\neq [m] produces ≠ . . . . . . . . . . . . . . . . . . (5.3.4) – 127 

\newboolean{switch} . . . . . . . . . . . . . . . . . (8.3.5) – 194 

Requires the standard L AT E X package ifthen. Creates a 

new boolean switch. The value of the switch is set with 

\setboolean{switch}{value}, where value is true or false. Its 

value is tested with \boolean{switch}, which may be used as a logical 

statement in the test part of \ifthenelse and \whiledo.


\newcommand{com name}[narg][opt]{def } . . . . . . . (8.3) – 185 

Defines a user command with the name \com name to be def. The 

first optional argument narg ≤ 9 specifies how many variable arguments 

the command is to have, which appear in the def as the 

replacement characters #1 to #narg. If the second optional argument 

is present, the first argument of the new command is optional, 

and takes on the value opt if it is not explicitly given. 

\newcommand*{com name}[narg][opt]{def } . . . . . . (D.2.6) – 445 

The same as \newcommand except that the arguments to \com name 

must be ‘short’, not containing any new paragraphs. 

\newcounter{counter name}[in counter] . . . . . . . (8.1.2) – 182 

Establishes a new counter with the name counter name. The optional 

argument in counter is the name of an existing counter which, when 

incremented, resets the new counter to zero; that is, the new counter 

is a sub-counter of in counter. 

\newenvironment{env name}[narg][opt]{beg def }{end def } 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8.4) – 195 

Defines a user environment with the name env name which has the 

\begin definition beg def and the \end definition end def. The 

optional argument narg ≤ 9 specifies how many variable arguments 

the environment is to have, which appear in the beg def as the 

replacement characters #1 to #narg. If the second optional argument 

is present, the first argument of the \begin command is optional, 

and takes on the value opt if it is not explicitly given. 

\newenvironment*{env name}[narg][opt]{beg def }{end def } 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (D.2.6) – 445 

The same as \newenvironment except that the arguments to 

\begin{env name} must be ‘short’, not containing any new paragraphs. 

\newfont{\font cmd}{\font name scaled size} . . . . . (4.1.5) – 66 

Establishes the relation between the font file name file name magnified 

by the scaling factor size and a font selection command 

\font cmd. After \font cmd has been called, \baselineskip, the 

interline spacing, still has its previous value. 

\newlength{\length cmd} . . . . . . . . . . . . . . . . (8.2) – 185 

Creates a new length command with the name \length cmd and 

initializes it to 0 pt. New values may be assigned as for all length 

commands with the \setlength command: 

\setlength{\length cmd}{length} 

The quantity length must have units (cm, pt, etc.) and may be a 

rubber length.


\newline . . . . . . . . . . . . . . . . . . . . . . . . (2.7.2) – 31 

Terminates and starts a line of text without right justifying it. 

\newpage . . . . . . . . . . . . . . . . . . . . . . . . (2.7.4) – 33 

Terminates and starts a new page, leaving the rest of the page blank. 

\newsavebox{\boxname} . . . . . . . . . . . . . . . . (4.7.1) – 87 

Creates a storage box with the name \boxname in which LR boxes 

may be saved with the \savebox command. 

\newtheorem{type}[num like]{title} 

\newtheorem{type}{title}[in ctr] . . . . . . . . . . . . (4.5) – 80 

Defines a new theorem-like environment named type which when 

called prints a theorem declaration with the name title in bold face, 

followed by an automatic sequential number, and the actual text of 

the environment in italic. The optional argument num like is the 

name of another theorem structure which is to share the same numbering 

counter. The other optional argument in ctr is the name of a 

sectioning counter, such as chapter, which is to reset the theorem 

counter every time it is incremented. That is, the theorem counter is 

a sub-counter of in ctr. Only one of the optional arguments may be 

given. 

\NG . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character Ŋ. 

\ng . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character ŋ. 

\ni [m] produces ∋ . . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\nocite{key} . . . . . . . . . . . . . . . . . . . . . (14.1) – 310 

The entry in the literature database with the keyword key will be 

included in the bibliography without any citation (reference) in the 

text. With \nocite{*}, all entries in all databases will be included. 

\nofiles [p] . . . . . . . . . . . . . . . . . . . . . . . (B.6) – 396 

Issued in the preamble, this command suppresses the output of the 

auxiliary files.aux,.glo,.idx,.lof,.lot, and.toc. 

\noindent . . . . . . . . . . . . . . . . . . . . . . . . (3.2.4) – 46 

The first line of the next paragraph will not be indented. 

\nolimits [m] . . . . . . . . . . . . . . . . . . . . . (5.3.7) – 128 

Places the upper and lower limits after the appropriate symbols where 

these would normally go just above or below them.


\nolinebreak[n] . . . . . . . . . . . . . . . . . . . . (2.7.2) – 32 

A recommendation not to break the line of text at this point. The 

urgency of the recommendation is given by the integral number 

n between 0 and 4, with the higher numbers meaning a stronger 

recommendation. A value of 4 is the same as the command without 

the optional argument and means absolutely no line break here. 

\nonfrenchspacing . . . . . . . . . . . . . . . . . . . (2.7.1) – 29 

Countermands \frenchspacing, switching back to the standard formatting 

in which extra word spacing is inserted at the end of a 

sentence. 

\nonumber [m] . . . . . . . . . . . . . . . . . . . . . (5.4.7) – 139 

The formula line in an eqnarray environment in which this command 

appears will not contain an equation number. 

\nopagebreak[n] . . . . . . . . . . . . . . . . . . . . (2.7.4) – 33 

A recommendation not to break the page at this point. The urgency of 

the recommendation is given by the integral number n between 0 and 

4, with the higher numbers meaning a stronger recommendation. A 

value of 4 is the same as the command without the optional argument 

and means absolutely no page break here. 

\normalcolor . . . . . . . . . . . . . . . . . . . . . . (6.2) – 167 

A command that normally does nothing. However, if the color 

package has been loaded, it resets the color for text to be the color 

that was in effect at the end of the preamble, normally black. A 

\color command in the preamble can alter this ‘standard’ color. 

This command is called by many internal L AT E X macros, to reset 

the text color when printing headlines and headings. Other packages 

should also use it so that they are consistent with the color package. 

\normalfont . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 65, 371 

This declaration switches to the font with the default family, shape, 

and series attributes. 

\normalmarginpar . . . . . . . . . . . . . . . . . . (4.10.5) – 117 

Countermands \reversemarginpar, switching back to the standard 

placement of marginal notes in the ‘outer’ margin. 

\normalsize . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \normalsize, the size selected by the 

option in the \documentclass or \documentstyle commands. It is 

smaller than \large but larger than \small. 

\not [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.4) – 127 

Changes the following comparison symbol into its negative counterpart: 

\not\cong = ≅


\notag{mark} [m][a] . . . . . . . . . . . . . . . . . (12.2.6) – 271 

Within one of the A M S-L AT E X alignment environments, suppresses 

the automatic equation number. 

\notin [m] produces ∉ . . . . . . . . . . . . . . . . . (5.3.4) – 127 

\nu [m] produces ν . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\numberwithin{ctr}{in ctr} [a] . . . . . . . . . . . (12.2.7) – 277 

With the amsmath package, redefines the counter ctr to be a subcounter 

of in ctr, meaning it is reset every time in ctr is incremented. 

The value of in ctr is printed with that of ctr. This is normally used 

to make equations in an article to be numbered within sections: 

\numberwithin{equation}{section} 

\nwarrow [m] produces ↖ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\O produces Ø . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\o produces ø . . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\oddsidemargin . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

Sets the left margin for the odd-numbered pages in document class 

book or when the option twoside has been selected for other classes. 

In all other cases, it sets the left margin for all pages. A new value is 


\setlength{\evensidemargin}{1.5cm} 

\odot [m] produces ⊙ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\OE produces Œ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\oe produces œ . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\oint [m] produces . . . . . . . . . . . . . . . . . . (5.3.7) – 128 

\Omega [m] produces Ω . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\omega [m] produces ω . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\ominus [m] produces ⊖ . . . . . . . . . . . . . . . . (5.3.3) – 125 

\onecolumn . . . . . . . . . . . . . . . . . . . . . . . (3.2.7) – 51 

Starts a new page and switches from two-column to one-column page 

formatting. 

\onlynotes{note nums} . . . . . . . . . . . . . . . (15.1.3) – 327 

In slides class, a command to be issued in the preamble to generate 

only those notes whose numbers appear in note nums. The command 

behaves the same as \onlyslides for slides. 

\onlyslides{slide nums} . . . . . . . . . . . . . . (15.1.3) – 327 

In slides class, a command to be issued in the preamble to generate 

only those slides whose numbers appear in slide nums. The numbers 

are separated by commas, and may include a range with a hyphen: 

\onlyslides{4,10-13,23}.


\opening{dear} . . . . . . . . . . . . . . . . . . . . (16.1) – 353 

In the letter environment of the letter class, this sets the 

form of the salutation at the start of the letter text; for example, 

\opening{Dear George,}. 

\oplus [m] produces ⊕ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\OptionNotUsed [p] . . . . . . . . . . . . . . . . . . (D.2.3) – 443 

A command that may only be used in the definition of options, 

especially default options. It declares the \CurrentOption to be 

unprocessed. This is used if the processing of a default option 

should fail, say because some file is missing. L AT E X is then informed 

that this requested option is still outstanding. 

\oslash [m] produces ⊘ . . . . . . . . . . . . . . . . (5.3.3) – 125 

\otimes [m] produces ⊗ . . . . . . . . . . . . . . . . (5.3.3) – 125 

\oval(x dimen,y dimen)[part] . . . . . . . . . . . (13.1.4) – 296 

Picture element command to produce an oval with width x dimen 

and height y dimen in the picture environment. The optional part 

argument may take on values of t, b, l, and r to draw only the top, 

bottom, left, or right halves of the oval. A combination of these 

values may be given to draw only a quarter of the oval, such as tl 

or lt for the top left part. To be used as an argument in a \put or 

\multiput command. 

\ovalbox{text} . . . . . . . . . . . . . . . . . . . . . (4.7.9) – 94 

With the fancybox package, is a variant of \fbox, drawing a framed 

box with round corners around text; the thickness of the lines is 

given by \thinlines. 

\Ovalbox{text} . . . . . . . . . . . . . . . . . . . . . (4.7.9) – 94 

With the fancybox package, is the same as \ovalbox but the thickness 

of the lines is given by \thicklines. 

\overbrace{sub form} [m] . . . . . . . . . . . . . . . (5.4.4) – 136 

Produces a horizontal curly brace over the math formula sub form. 

Any following superscript will be placed centered above the horizon- 

tal brace. 

\overbrace{a+b} = 

 

a + b 

\overbrace{x+y+z}ˆ{\xi\eta\zeta} = 

ξηζ 

 

x + y + z 

\overleftarrow{expr} [m][a] . . . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long leftwards pointing arrow 

over the mathematical expression expr. 

\overleftrightarrow{expr} [m][a] . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long double arrow over the 

mathematical expression expr.


\overline{sub form} [m] . . . . . . . . . . . . . . . (5.4.4) – 136 

Produces a horizontal bar over the math formula sub form: 

\overline{a-b} = a − b 

\overrightarrow{expr} [m][a] . . . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long rightwards pointing arrow 

over the mathematical expression expr. 

\overset{char}{\symbol} [m][a] . . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places char over the math symbol \symbol 

in superscript size. 

\P produces . . . . . . . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\PackageError{pkg name}{error text}{help} [p] . . . . (D.2.7) – 446 

Writes an error message error text to the monitor and transcript file, 

labeled with the package name, and halts processing, waiting for a 

user response as for a L AT E X error. If H〈return〉 is typed, the help text is 

printed. Both error text and help may contain \MessageBreak for a 

new line, \space for a forced space, and \protect before commands 

that are to have their names printed literally and not interpreted. 

\PackageInfo{pkg name}{info text} [p] . . . . . . . . (D.2.7) – 447 

Is like \PackageWarningNoLine except that the text info text is only 

written to the transcript file, and not to the monitor. 

\PackageWarning{pkg name}{warn text} [p] . . . . . . (D.2.7) – 446 

Writes warn text to the monitor and transcript file, labeled with 

the package name and the current line number of the input file. 

Processing continues. The warn text is formatted in the same way as 

that for \PackageError. 

\PackageWarningNoLine{pkg name}{warn text} [p] . . (D.2.7) – 446 

Is like \PackageWarning except that the current line number of the 

input file is not printed. 

\pagebreak[n] . . . . . . . . . . . . . . . . . . . . . (2.7.4) – 33 

A recommendation to break the page at this point. The urgency of 

the recommendation is given by the integral number n between 0 and 

4, with the higher numbers meaning a stronger recommendation. A 

value of 4 is the same as the command without the optional argument 

and means an obligatory page break. 

\pagecolor col spec . . . . . . . . . . . . . . . . . . . (6.2) – 167 

A command made available with the color package. Sets the background 

color starting with the current page. All following pages have 

the same background color until \pagecolor is called once more. 

The col spec is the same as for \color.


\pagename . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command in the letter document class containing the text for page 

numbers after the first page. In English, this is ‘Page’ but may be 


\pagenumbering{style} . . . . . . . . . . . . . . . . . (3.2.3) – 45 

Determines the style of the page numbering and resets the page 

counter to 1. Possible values for style are: arabic, roman, Roman, 

alph, and Alph. 

\pageref{marker} . . . . . . . . . . . . . . . . . . . (9.2.1) – 213 

Prints the number of the page where marker has been set by a 

\label{marker} command. 

\pagestyle{style} [p] . . . . . . . . . . . . . . . . . . (3.2) – 42 

Determines the page style, that is, the contents of the head and 

footlines on every page. Possible values for style are: plain, empty, 

headings, and myheadings. 

\path{directory} . . . . . . . . . . . . . . . . . . . . (4.9.2) – 112 

With the url package, prints directory literally, in typewriter font, 

with line breaks after non-letters, without hyphens. It functions 

much the same as the \url command, but is logically distinct since 

it is encoding something different. 

\paperheight . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The total height of the page as specified by the page size option in 

the \documentclass command line. With the default lettersize 

option, this is 11 in; with a4paper, it is 29.7 cm. The additional 

option landscape interchanges the values of \paperwidth and 

\paperheight. 

\paperwidth . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The total width of the page as specified by the page size option in 

the \documentclass command line. With the default lettersize 

option, this is 8.5 in; with a4paper, it is 21 cm. The additional 

option landscape interchanges the values of \paperwidth and 

\paperheight. 

\par . . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.1) – 23 

Ends the current paragraph and begins a new one. This command is 

equivalent to a blank line. 

\paragraph[short title]{title} . . . . . . . . . . . . . . (3.3.3) – 55 

The second last command in the sectioning hierarchy, coming between 

\subsubsection and \subparagraph. It formats title with 

the current sub-subsection number and an automatic sequential paragraph 

number. If the optional short title is given, it appears in place 

of title in the table of contents.


\paragraph*{title} . . . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \paragraph but without a number or an entry in the 

table of contents. 

\parallel [m] produces . . . . . . . . . . . . . . . (5.3.4) – 126 

\parbox[pos][height][inner pos]{width}{text} (4.7.3), (4.7.5) – 88, 90 

Produces a vertical box of width width in which text is set in lines 

that are left and right justified to this width. The vertical positioning 

with respect to the surrounding text is determined by the optional 

argument pos: t for alignment with its top line, b with its bottom 

line, and centered with no argument. The two additional optional 

arguments: height to give the total height, and inner pos to specify 

how the text is to be positioned inside it. Possible values are t for top, 

b for bottom, c for centered, and s to be stretched out to fill the whole 

vertical space. The default is the value of the external positioning pos 

option. The height argument may contain the parameters \height, 

\depth, \width, and \totalheight. 

\parindent . . . . . . . . . . . . . . . . . . . . . . . (3.2.4) – 46 

The amount of indentation for the first line of a paragraph. A new 

value may be assigned with the \setlength command: 

\setlength{\parindent}{1.5em} 

\parsep . . . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 76 

The vertical spacing between paragraphs within a list environment. 

A new value may be assigned with the \setlength command: 

\setlength{\parsep}{2pt plus1pt minus1pt} 

\parskip . . . . . . . . . . . . . . . . . . . . . . . . (3.2.4) – 46 

The vertical spacing between paragraphs. A new value may be assigned 


\setlength{\parskip}{3pt plus1pt minus2pt} 

\part[short title]{title} . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The highest command in the sectioning hierarchy. It begins a new 

‘Part’ with an automatic sequential part number and the heading 

title. The following sectioning numbers are not influenced by the 

part number. If the optional short title is given, it appears in place of 

title in the table of contents. 

\part*{title} . . . . . . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \part but without a number or an entry in the table of 

contents. 

\partial [m] produces ∂ . . . . . . . . . . . . . . . . (5.3.6) – 127 

\partname . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the part heading. In English, this is ‘Part’ but 

may be altered for adaptation to other languages.


\partopsep . . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 75 

The additional vertical spacing at the beginning and/or end of a listing 

when a blank line precedes or follows the environment commands. 

A new value may be assigned with the \setlength command: 

\setlength{\partopsep}{2pt plus1pt minus1pt} 

\PassOptionsToClass{options}{class} [p] . . . . . . . (D.2.3) – 443 

Assigns the options in the list options to the specified class file, which 

is later loaded with \LoadClass. This command must be called from 

a class file, or from another file input by a class file. It may be used in 

the definition of options, or in a configuration file to activate options. 

\PassOptionsToPackage{options}{package} [p] . . . . (D.2.3) – 443 

Assigns the options in the list options to the specified package file, 

which is later loaded with \RequirePackage. This command may be 

called from a class or package file. It may be used in the definition 

of options, or in a configuration file to activate certain options. 

\perp [m] produces ⊥ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\Phi [m] produces Φ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\phi [m] produces φ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\Pi [m] produces Π . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\pi [m] produces π . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\pm [m] produces ± . . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\pmb{symbol} [m][a] . . . . . . . . . . . . . . . . . (12.2.1) – 258 

When one of the packages amsmath or amsbsy has been loaded, this 

command prints symbol in simulated bold face. This is done by 

printing it several times slightly displaced. 

\pmod{arg} [m] . . . . . . . . . . . . . (5.3.8), (12.2.5) – 129, 268 

Command to produce the function name ‘mod’ in formulas in the 

form: 

y\pmod{a+b} = y (mod a + b) 

\pod{arg} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.5) – 268 

With the amsopn or amsmath packages, command to produce the 

function name ‘mod’ in formulas in the form: 

y\pod{a+b} = y (a + b) 

\poptabs . . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Restores the last set of tabular stops in the tabbing environment 

that has been saved with \pushtabs. 

\pounds produces £ . . . . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\Pr [m] . . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘Pr’ in formulas. A lower 

limit may be set as a subscript.


\prec [m] produces ≺ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\preceq [m] produces . . . . . . . . . . . . . . . . (5.3.4) – 126 

\prime [m] produces ′ (identical to the ’ symbol) . . . . (5.3.6) – 127 

\printindex . . . . . . . . . . . . . . . . . . . . . . (9.4.3) – 228 

A command defined in the makeidx.sty file that generates theindex 

environment after the program MakeIndex has processed the.idx file. 

\ProcessOptions [p] . . . . . . . . . . . . . . . . . . (D.2.3) – 443 

In a class or package file, this command processes the requested 

options by executing the \ds@ commands for each one in the order 

in which they were defined. The \ds@ commands are then erased. 

\ProcessOptions* [p] . . . . . . . . . . . . . . . . . (D.2.3) – 443 

This is the same as \ProcessOptions except that the \ds@ commands 

are executed in the order in which the options were requested. 

\prod [m] produces . . . . . . . . . . . . . . . . . (5.3.7) – 128 

\propto [m] produces ∝ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\protect . . . . . . . . . . . . . . . . . . . . . . . (D.2.5) – 445 

Fragile commands may be used in moving arguments when they are 

preceded by the \protect command. Example: 

\section{The \protect\pounds{} Sign}. 

\providecommand{\com name}[narg][opt]{def } . . . (8.3.1) – 187 

The same as \newcommand except that if a command with the name 

\com name already exists, the new definition is ignored. 

\providecommand*{\com name}[narg][opt]{def } . . . (D.2.6) – 445 

The same as \providecommand except that the arguments to 

\com name must be ‘short’, not containing any new paragraphs. 

\ProvidesClass{class}[version] [p] . . . . . . . . . . (D.2.1) – 441 

At the beginning of a class file, this statement declares the name of 

the class and its version, to be checked against the name and version 

in the \documentclass or \LoadClass command that input it. The 

version specification, if present, consists of three parts: date, version 

number, and additional information. Example: 

\ProvidesClass{thesis}[1995/01/25 v3.8 U of Saigon] 

\ProvidesFile{file name}[version] . . . . . . . . . . (D.2.1) – 442 

At the beginning of a general file, this statement declares its name 

and version. No checking is done when the file is read in with 

\input, but the information is printed out if \listfiles has been 

activated. This command is not limited to the preamble as the other 

\Provides.. commands are.


\ProvidesPackage{class}[version] [p] . . . . . . . . . (D.2.1) – 441 

At the beginning of a package file, this statement declares the name 

of the package and its version, to be checked against the name and 

version in the \usepackage or \RequirePackage command that 

input it. The version specification, if present, consists of three parts: 

date, version number, and additional information. Example: 

\ProvidesPackage{notes}[1995/02/13 1.2 G. Smith] 

\ProvideTextCommand{\cmd}{code}[narg][opt]{def } [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.7) – 379 

Defines \cmd in the same way as \providecommand except the definition 

is only valid when encoding code is active. If the command 

\cmd is already defined for that encoding, it is not redefined. 

\ProvideTextCommandDefault{\cmd}{code}[narg][opt]{def } [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.7) – 380 

Is the same as \DeclareTextCommandDefault except that if a default 

encoding definition already exists for \cmd, then no redefinition 

occurs and the previous definition is retained. 

\ps text . . . . . . . . . . . . . . . . . . . . . . . . (16.1) – 353 

Adds a postscript to a letter in the letter document class. 

\Psi [m] produces Ψ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\psi [m] produces ψ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\pushtabs . . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Saves the current set of tabulator stops in the tabbing environment. 

It may be recalled with the \poptabs command. 

\put(x,y){pic elem} . . . . . . . . . . . . . . . . . (13.1.3) – 289 

The positioning command within a picture environment. The picture 

element pic elem is placed with its reference point at the location 

(x, y). 

\qbezier[num](x1,y1)(x2,y2)(x3,y3) . . . . . . (13.1.4) – 299 

This command can be given within the picture environment to 

draw a quadratic Bézier curve from point (x1,y1) to (x3,y3), using 

(x2,y2) as the extra Bézier point. The curve is drawn as num + 

1 dots if the optional argument num is given, otherwise num is 

calculated automatically to produce a solid line. It is the same as 

\bezier except that num is optional. 

\quad . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 30 

Inserts horizontal spacing of size 1 em. 

\qquad . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.1) – 30 

Inserts horizontal spacing of size 2 em.


\quotedblbase . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol „. 

\quotesinglbase . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol ‚. 

\r{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces a circle accent: \r{o} = ˚o. 

\raggedbottom . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The standard page formatting for article, report, and letter 

document classes when the twoside option has not been selected. 

The spacing between paragraphs is fixed so that the last line will vary 

from page to page. The opposite command is \flushbottom. 

\raggedleft . . . . . . . . . . . . . . . . . . . . . . . (4.2.2) – 67 

After this declaration, the lines of text will only be right justified and 

the left margin will be uneven. The individual lines are terminated 

by \\. See also \begin{flushright}. 

\raggedright . . . . . . . . . . . . . . . . . . . . . . (4.2.2) – 67 

After this declaration, the lines of text will only be left justified and 

the right margin will be uneven. The individual lines are terminated 

by \\. See also \begin{flushleft}. 

\raisebox{lift}[height][depth]{text} . . . . . . . . . . (4.7.2) – 87 

An LR box containing text is raised an amount lift above the current 

baseline. If lift is negative, the box is lowered. The optional arguments 

state that it is to be treated as though it extended by height 

above and by depth below the baseline regardless of its true extents. 

\raisetag{len} [m][a] . . . . . . . . . . . . . . . . (12.2.6) – 271 

Within one of the A M S-L AT E X alignment environments, raises the 

equation number or marker by len above its normal position. 

\rangle [m] produces 〉 . . . . . . . . . . . . . . . . . (5.4.1) – 132 

\rceil [m] produces ⌉ . . . . . . . . . . . . . . . . . (5.4.1) – 132 

\Re [m] produces ℜ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\ref{marker} . . . . . . . . . . . . . . . . . . . . . (9.2.1) – 213 

Prints the number of the section, equation, figure, or table where 

marker has been set by a \label{marker} command. 

\reflectbox{text} . . . . . . . . . . . . . . . . . . . (6.1.2) – 156 

A command made available with the graphics package that reflects 

the contents text as an LR box such that left and right are reversed.


\refname . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

Command containing the heading for the bibliography in article 

document class. In English, this is ‘References’ but may be altered 

for adaptation to other languages. 

\refstepcounter{counter} . . . . . . . . . . . . . . (8.1.3) – 182 

Increases the value of the number stored in counter by one, the same 

as \stepcounter, but also makes the specified counter the relevant 

one for the \label–\ref cross-referencing commands. 

\renewcommand{com name}[narg][opt]{def } . . . . . . (8.3) – 185 

The same as \newcommand except that the command \com name 

must already exist, otherwise an error message is printed. 

\renewcommand*{com name}[narg][opt]{def } . . . . (D.2.6) – 445 

The same as \renewcommand except that the arguments to 

\com name must be ‘short’, not containing any new paragraphs. 

\renewenvironment{env}[narg][opt]{beg}{end} . . . . (8.4) – 195 

The same as \newenvironment except that the environment env 

must already exist, otherwise an error message is printed. 

\renewenvironment*{env}[narg][opt]{beg}{end} . . (D.2.6) – 445 

The same as \renewenvironment except that the arguments to 

\begin{env} must be ‘short’, not containing any new paragraphs. 

\RequirePackage[options]{packages}[version] [p] . . . (D.2.2) – 442 

This command is the equivalent of \usepackage within a class or 

package file. It loads one or more package files with the extension 

.sty. More than one package may be specified in packages, the names 

being separated by commas. Any options listed will be applied to all 

packages. Furthermore, any options listed in the \documentclass 

command will also be applied to the package files. 


example 1994/08/01. If the date of the package file is earlier than 

this, a warning message is printed. 

\RequirePackageWithOptions{package}[version] [p] . (D.2.2) – 442 

Like \RequirePackage except all the currently specified options are 

automatically passed to package. 

\resizebox{h length}{v length}{text} . . . . . . . . . (6.1.2) – 156 

A command made available with the graphics package that scales 

the contents text as an LR box such that the horizontal size becomes 

h length and the vertical size v length. If either size is given as !, the 

one scale factor is applied to both dimensions.


\resizebox*{h length}{v length}{text} . . . . . . . . . (6.1.2) – 156 

The same as \resizebox except that the vertical size v length refers 

to the total height plus depth of the LR box. 

\reversemarginpar . . . . . . . . . . . . . . . . . (4.10.5) – 117 

Changes the placement of marginal notes from the standard (right 

or ‘outer’ margin) to the opposite side. Can be countermanded with 

\normalmarginpar. 

\rfloor [m] produces ⌋ . . . . . . . . . . . . . . . . . (5.4.1) – 132 

\rhd [m] produces ⊲ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\rho [m] produces ρ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\rightrbrack [m] . . . . . . . . . . . . . . . . . . . . (5.4.1) – 131 

Adjusts the size of the bracket symbol rbrack to fit the height of 

the formula between the \left ... \right pair. For example, 

\right]. If there is to be no matching bracket, the \left and 

\right commands must still be given to specify the part of the 

formula to be sized, but the missing bracket is given as a period (for 

example, \left.). 

\Rightarrow [m] produces ⇒ . . . . . . . . . . . . . . (5.3.5) – 127 

\rightarrow [m] produces → . . . . . . . . . . . . . . (5.3.5) – 127 

\rightharpoondown [m] produces ⇁ . . . . . . . . . . (5.3.5) – 127 

\rightharpoonup [m] produces ⇀ . . . . . . . . . . . (5.3.5) – 127 

\rightleftharpoons [m] produces ⇌ . . . . . . . . . (5.3.5) – 127 

\rightmargin . . . . . . . . . . . . . . . . . . . . . . (4.4.2) – 76 

In a list environment, the amount by which the right edge of the 

text is indented relative to the right side of the surrounding text. 

Standard value is 0 pt. A new value is assigned with the \setlength 


\setlength{\rightmargin}{0.5cm} 

\rmdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 

This command defines the family attribute that is selected with the 

\rmfamily command. It may be redefined with \renewcommand: 

\renewcommand{\rmdefault}{ptm} 

\rmfamily . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 

This declaration switches to a font in the current series and shape, 

but with the Roman family attribute. 

\Roman{counter} . . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of the counter as an upper case Roman 

numeral. 

\roman{counter} . . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Prints the current value of the counter as a lower case Roman numeral.


\rotatebox{angle}{text} . . . . . . . . . . . . . . . . (6.1.2) – 156 

A command made available with the graphics package that rotates 

the contents text as an LR box through the angle expressed in degrees. 

The rotation is counterclockwise about the left-hand end of 

the baseline of the box. 

\rq produces ’, identical to the ’ symbol. 

\rule[lift]{width}{height} . . . . . . . . . . . . . . . (4.7.6) – 91 

Produces a black rectangle of width width and height height, raised 

above the baseline by an amount lift, if this optional argument is given. 

A value of ‘0 pt’ for either the width or height creates an invisible 

horizontal or vertical strut that may be used to make spacing. 

\rvert [m][a] produces | (right delimiter) . . . . . . . (12.2.5) – 270 

\rVert [m][a] produces (right delimiter) . . . . . . . (12.2.5) – 270 

\S produces § . . . . . . . . . . . . . . . . . . . . . . . (2.5.5) – 23 

\SS produces SS, the upper case version of \ss, ß . . . . . (2.5.6) – 24 

\savebox{\boxname}[width][pos]{text} . . . . . . . . (4.7.1) – 87 

Functions the same as the \makebox command except that the box 

contents are not output but saved under the name \boxname, which 

has been previously defined with \newsavebox. The box may be 

set any place in the text as often as desired with the command 


\savebox{\boxname}(x dim,y dim)[pos]{sub pic} . (13.1.4) – 300 

In the picture environment, a sub-picture sub pic may be stored as 

a box of width x dim and height y dim under the name \boxname, 

which has been previously defined with \newsavebox. The pos argument 

functions as it does for the picture \makebox. The box 

may be set any place in the picture environment with the command 


\sb produces a subscript, identical to the _ symbol. 

\sbox{\boxname}{text} . . . . . . . . . . . . . . . . . (4.7.1) – 87 

Stores text in an LR box named \boxname that has previously been 

created with \newsavebox{\boxname}. The contents of the box may 

be printed as often as desired with \usebox{\boxname}. 

\scalebox{h scale}[v scale]{text} . . . . . . . . . . . (6.1.2) – 156 

A command made available with the graphics package that scales 

the contents text as an LR box with the horizontal factor h scale 

and optionally with the (different) vertical factor v scale. If v scale is 

missing, it is the same as h scale. 

\scdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 


\scshape command. It may be redefined with \renewcommand: 

\renewcommand{\scdefault}{sc}


\scriptscriptstyle [m] . . . . . . . . . . . . . . . (5.5.2) – 146 

Switches to font size \scriptscriptstyle as the active font inside 

a math formula. 

\scriptsize . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \scriptsize, which is smaller than 

\footnotesize but larger than \tiny. 

\scriptstyle [m] . . . . . . . . . . . . . . . . . . . (5.5.2) – 146 

Switches to font size \scriptstyle as the active font inside a math 

formula. 

\scshape . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 


but with the Caps and Small Caps shape attribute. 

\searrow [m] produces ↘ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\sec [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘sec’ in formulas. 

\section[short title]{title} . . . . . . . . . . . . . . . . (3.3.3) – 55 

Begins a new section, formatting title with the current chapter number 

(book and report classes only) and an automatic sequential section 

number. If the optional short title is given, it appears in place of title 

in the table of contents and the running head at the top of the pages. 

\section*{title} . . . . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \section but without a number or an entry in the table 

of contents. 

\see{reference} . . . . . . . . . . . . . . . . . . . . (9.4.2) – 227 

A command defined in the makeidx package for use with the MakeIndex 

program. It is called within an \index command to refer to 

another entry in the keyword index as ‘see reference’. Given in the 

form: 

\index{entry|see{reference}} 

Note: the above text is correct with | in place of \ for |see. 

\seealso{reference} . . . . . . . . . . . . . . . . . . (9.4.2) – 227 

Like \see, but will print the text ‘see also reference’ in the index. 

More precisely, it prints the text stored in the command \alsoname. 

Requires the makeidx package. 

\seename . . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 460 

A command defined in the package makeidx containing the text for 

the command \see. In English, this is ‘see’ but may be altered for 

adaptation to other languages.


\selectfont . . . . . . . . . . . . . . . . . . . . . . . (A.1) – 369 

Activates the font with the current set of attributes, making it the 

current font in which text is set. It should normally follow an attribute 

selection command. Example: 

\fontshape{sl}\selectfont 

\selectlanguage{language} . . . . . . . . . . . . . (11.1) – 255 

Command in multi-language packages such as german and in the 

babel system for changing the language. The names of titles, the 

form of the date command \today, special language-specific commands, 

and the hyphenation patterns are all changed. Example: 

\selectlanguage{english} 

\setboolean{switch}{value} . . . . . . . . . . . . . . (8.3.5) – 194 

Requires the standard L AT E X package ifthen. Sets the value of 

a boolean switch to 〈true〉 or 〈false〉, depending on value, which 

must be true or false. The switch must have been created with 

\newboolean{switch}. Its value is tested with \boolean{switch}, 

which may be used as a logical statement in the test part of 

\ifthenelse and \whiledo. 

\setcounter{counter}{value} . . . . . . . . . . . . . (8.1.3) – 182 

Assigns the integral number value to the counter counter. 

\setlength{\length cmd}{length spec} . . . . . . . . . (8.2) – 184 

The length command with the name \length cmd is assigned the 

length value length spec, which may be a fixed or rubber length. 

See also . . . . . . . . . . . . . . . . . (2.4.1), (2.4.2) – 21, 21 

\SetMathAlphabet{\cmd}{ver}{code}{fam}{ser}{shp} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.3) – 373 

Defines the math alphabet that has been declared with 

\DeclareMathAlphabet to take the specified font attributes in the 

math version ver. 

\setminus [m] produces \ . . . . . . . . . . . . . . . (5.3.3) – 125 

\SetSymbolFont{sym fnt}{ver}{code}{fam}{ser}{shp} [p] 

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A.3.4) – 374 

Defines the symbol font that has been declared with 

\DeclareSymbolFont to take the specified font attributes in the 

math version ver. 

\settime{secs} . . . . . . . . . . . . . . . . . . . (15.1.3) – 327 

In the slides class, if the option clock has been selected, a time 

marker, in minutes, appears at the bottom of the notes. This command 

sets the internal timer to the specified number of seconds. See 

also \addtime.


\settodepth{\length cmd}{text} . . . . . . . . . . . . (8.2) – 184 

The length command with the name \length cmd is assigned a value 

equal to the depth of text below the baseline. 

\settoheight{\length cmd}{text} . . . . . . . . . . . . (8.2) – 184 


equal to the height of text above the baseline. 

\settowidth{\length cmd}{text} . . . . . . . . . . . . (8.2) – 184 


equal to the length of text as it would be set in an LR box. 

\sfdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 


\sffamily command. It may be redefined with \renewcommand: 

\renewcommand{\sfdefault}{phv} 

\sffamily . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 


but with the sans serif family attribute. 

\shadowbox{text} . . . . . . . . . . . . . . . . . . . . (4.7.9) – 94 

With the fancybox package, is a variant of \fbox, drawing a shadowed 

box around text; the thickness of the shadow is given by the 

length \shadowsize. 

\sharp [m] produces ♯ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\shortstack[pos]{text} . . . . . . . . . . . . . . . (13.1.4) – 297 

Formats the text into a single column, where the individual 

rows are terminated by \\. The optional positioning argument 

pos takes on values of l or r to set the text left or right justified, 

otherwise it is centered. Example: 

\shortstack{aa\\bbb\\cc\\x\\yy\\zzz} 

aa 

bbb 

cc 

x 

yy 

zzz 

\sideset{pre}{post}\symbol [m][a] . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places superscripts and subscripts snugly 

before (pre) and after (post) the math symbol \symbol. Example: 

$\sideset{_\dagˆ*}{_\dagˆ*}\prod$ yields ∗∗ 

\Sigma [m] produces Σ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\sigma [m] produces σ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\signature{name} . . . . . . . . . . . . . . . . . . (16.1) – 351 

In the letter document class, supplies the name of the writer that 

should go below the signature if this is different from the entry in 

\name. 

† 

†


\sim [m] produces ∼ . . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\simeq [m] produces . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\sin [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘sin’ in formulas. 

\sinh [m] . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘sinh’ in formulas. 

\sldefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 


\slshape command. It may be redefined with \renewcommand: 

\renewcommand{\sldefault}{sl} 

\sloppy . . . . . . . . . . . . . . . . . . . . . . . . . (2.8.3) – 36 

After this command has been given, word spacings are allowed to 

stretch more generously than usual so that paragraphs are broken up 

into lines with fewer word divisions. It is countermanded by \fussy. 

See also \begin{sloppypar}. 

\slshape . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 


but with the slanted shape attribute. 

\small . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the font size \small, which is smaller than \normalsize 

but larger than \footnotesize. 

\smallskip . . . . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

Inserts large vertical spacing of amount \smallskipamount. See also 

\medskip and \bigskip. 

\smallskipamount 

Standard value for the amount of vertical spacing that is inserted with 

the command \smallskip. May be changed with the \setlength 


\setlength{\smallskipamount}{1ex plus0.5ex minus0.3ex} 

\smash[pos]{text} [m][a] . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this T E X command acquires an optional 

argument pos that may be b or t, to effectively zero the depth or 

height of the text. With no pos, both height and depth are zeroed. 

\smile [m] produces ⌣ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\sp produces a superscript, identical to the ˆ symbol. 

\spadesuit [m] produces ♠ . . . . . . . . . . . . . . (5.3.6) – 127 

\sqcap [m] produces ⊓ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\sqcup [m] produces ⊔ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\sqrt[n]{arg} [m] . . . . . . . . . . . . . . . . . . . (5.2.4) – 122 

Basic math command to produce a root sign. The height and length 

of the sign are made to fit the contents arg. The optional argument 

n is the degree of the root: \sqrt[3]{2} = 3√ 2, \sqrt{2} = √ 2.


\sqsubset [m] produces ⊏ . . . . . . . . . . . . . . . (5.3.4) – 126 

\sqsubseteq [m] produces ⊑ . . . . . . . . . . . . . . (5.3.4) – 126 

\sqsupset [m] produces ⊐ . . . . . . . . . . . . . . . (5.3.4) – 126 

\sqsupseteq [m] produces ⊒ . . . . . . . . . . . . . . (5.3.4) – 126 

\ss produces ß . . . . . . . . . . . . . . . . . . . . . . (2.5.6) – 24 

\stackrel{upper}{lower} [m] . . . . . . . . . . . . . (5.4.5) – 137 

Places one mathematical symbol upper on top of another lower, such 

that the upper one appears in a smaller typeface: 

\stackrel{\alpha}{\longrightarrow} = α 

−→ 

\star [m] produces ⋆ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\stepcounter{counter} . . . . . . . . . . . . . . . . (8.1.3) – 182 

Increases the value of the number stored in counter by one. 

\stretch{decimal num} . . . . . . . . . . . . . . . . . (8.2) – 184 

A rubber length with a natural value of 0 pt but with a stretchability 

that is decimal num times that of \fill. 

\subitem{sub entry} . . . . . . . . . . . . . . . . . . (9.4.1) – 225 

In theindex environment, a command to produce a second-level 

entry after an \item command. 

\subparagraph[short title]{title} . . . . . . . . . . . . . (3.3.3) – 55 

The last command in the sectioning hierarchy, coming after 

\paragraph. It formats title with the current paragraph number 

and an automatic sequential sub-paragraph number. If the optional 

short title is given, it appears in place of title in the table of contents. 

\subparagraph*{title} . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \subparagraph but without a number or an entry in 

the table of contents. 

\subsection[short title]{title} . . . . . . . . . . . . . . (3.3.3) – 55 

The command in the sectioning hierarchy that comes between the 

\section and \subsubsection. It formats title with the current 

section number and an automatic sequential subsection number. If 

the optional short title is given, it appears in place of title in the table 

of contents. 

\subsection*{title} . . . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \subsection but without a number or an entry in the 


\substack{1st line\\..\\last line} [m][a] . . . . . . . (12.2.2) – 261 

With the amsmath package, produces centered multiline indices or 

limits; it must immediately follow ˆ or _ and be enclosed in { }.


\subsubitem{sub sub entry} . . . . . . . . . . . . . . (9.4.1) – 225 

In theindex environment, a command to produce a third-level entry 

under a \subitem command. 

\subsubsection[short form]{title} . . . . . . . . . . . (3.3.3) – 55 

The command in the sectioning hierarchy coming between 

\subsection and \paragraph. It formats title with the current subsection 

number and an automatic sequential sub-subsection number. 

If the optional short title is given, it appears in place of title in the 


\subsubsection*{title} . . . . . . . . . . . . . . . . . (3.3.3) – 55 

The same as \subsubsection but without a number or an entry in 

the table of contents. 

\subset [m] produces ⊂ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\subseteq [m] produces ⊆ . . . . . . . . . . . . . . . (5.3.4) – 126 

\succ [m] produces ≻ . . . . . . . . . . . . . . . . . (5.3.4) – 126 

\succeq [m] produces . . . . . . . . . . . . . . . . (5.3.4) – 126 

\sum [m] produces . . . . . . . . . . . . . . . . . . (5.2.5) – 123 

\sup [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘sup’ in formulas. A lower 


\suppressfloats[loc] . . . . . . . . . . . . . . . . . . (7.2) – 171 

Any floats given between this command and the end of the current 

page will be suspended at least until the next page. If the optional loc 

is given as one of t or b (not both), only floats with that placement 

parameter are suspended. 

\supset [m] produces ⊃ . . . . . . . . . . . . . . . . (5.3.4) – 126 

\supseteq [m] produces ⊇ . . . . . . . . . . . . . . . (5.3.4) – 126 

\surd [m] produces √ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\swarrow [m] produces ↙ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\symbol{n} . . . . . . . . . . . . . . . . . . . . . . . (4.1.6) – 66 

Produces the symbol in the current character font that is stored 

under the internal number n. 

\t{xy} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces a ‘tie-after’ accent over two letters: \t{oo} = oo. 

\tabbingsep . . . . . . . . . . . . . . . . . . . . . . . (4.6.4) – 83 

Determines the spacing between the text ltext and the current tabular 

stop when ltext\’ is given in a tabbing environment. A new value 

may be assigned with the \setlength command.


\tabcolsep . . . . . . . . . . . . . . . . . . . . . . . (4.8.2) – 98 

Determines the half-column spacing in the tabular environment. A 

new value may be assigned with the \setlength command: 

\setlength{\tabcolsep}{3mm} 

\tableofcontents . . . . . . . . . . . . . . . . . . . . (3.4.2) – 58 

Prints the table of contents from information in the sectioning commands 

and additional entries. 

\tablename . . . . . . . . . . . . . . . . . . . . . . (D.4.1) – 459 

Command containing the name for a table caption. In English, this 

is ‘Table’ but may be altered for adaptation to other languages. 

\tabularnewline[len] . . . . . . . . . . . . . . . . . . (4.8.1) – 98 

Terminates a row in the tabular or array environments, adding 

vertical spacing len if it is specified. This is equivalent to \\[len] 

except that there is no ambiguity as to whether it is terminating a 

row in the table or a line of text within a column entry. If something 

like \raggedright is given in the last column, then this command 

must be used in place of \\. 

\tag{mark} [m][a] . . . . . . . . . . . . . . . . . . (12.2.6) – 271 

Within one of the A M S-L AT E X alignment environments, prints mark 

in place of the equation number, in parentheses. The *-form prints 

it without parentheses. 

\tan [m] . . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘tan’ in formulas. 

\tanh [m] . . . . . . . . . . . . . . . . . . . . . . . (5.3.8) – 128 

Command to produce the function name ‘tanh’ in formulas. 

\tau [m] produces τ . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\tbinom{over}{under} [m][a] . . . . . . . . . . . . (12.2.3) – 264 

With the amsmath package, produces a binomial as \binom does, but 

in \textstyle size. 

\telephone{number} . . . . . . . . . . . . (16.1), (16.2) – 352, 356 

In the letter document class, enters the sender’s telephone number. 

In the standard L AT E X letter.sty, number is only output if \address 

has not been called. It is intended to be used in company letter styles 

such as mpletter. 

\TeX produces T E X . . . . . . . . . . . . . . . . . . . . (2.1) – 19 

\text{short text} [m][a] . . . . . . . . . . . . . . . (12.2.1) – 259 

When one of the packages amsmath or amstext has been loaded, this 

command prints short text as normal text within a math formula. If 

used in subscripts or superscripts, automatic sizing takes place.


\textsym name . . . . . . . . . . . . . . . . . . . . . (D.6) – 462 

An alternative means to produce certain special symbols that 

otherwise are only available in math mode or through ligature 

combinations: 

\textbullet (•); \textemdash (—); \textendash (–); 

\textexclamdown (¡); \textperiodcentered (·); 

\textquestiondown (¿); \textquotedblleft (“); 

\textquotedblright (”); \textquoteleft (‘); \textquoteright 

(’); \textvisiblespace ( ) 

\textasciicircum (ˆ); \textasciitilde (˜); \textbackslash (\); 

\textbar (|); \textgreater (>); \textless (


\textit{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

This command sets its argument in a font in the current family 

and series, but with the italic shape attribute. It is equivalent to 

{\itshape text}. 

\textmd{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

This command sets its argument in a font in the current family and 

shape, but with the medium series attribute. It is equivalent to 

{\mdseries text}. 

\textnormal{text} . . . . . . . . . . . . . . . . . . . . (4.1.3) – 65 

This command sets its argument in the font with the default family, 

series, and shape attributes. It is equivalent to {\normalfont text}. 

\textquotedbl . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the symbol ". 

\textregistered produces ® . . . . . . . . . . . . . (D.6) – 462 

\textrm{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current series and shape, but with the Roman 

family attribute. It is equivalent to {\rmfamily text}. 

\textsc{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current family and series, but with the Caps 

and Small Caps shape attribute. It is equivalent to {\scshape text}. 

\textsf{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current series and shape, but with the sans 

serif family attribute. It is equivalent to {\sffamily text}. 

\textsl{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current family and series, but with the slanted 

shape attribute. It is equivalent to {\slshape text}. 

\textstyle [m] . . . . . . . . . . . . . . . . . . . . (5.5.2) – 146 

Switches to font size \textstyle inside a math formula. 

\textsuperscript{char} . . . . . . . . . . . . . . . . (D.6) – 462 

Produces a superscript in the current text, rather than math, font: 

\textsuperscript{12} = 12 . 

\texttrademark produces . . . . . . . . . . . . . . (D.6) – 462 

\texttt{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current series and shape, but with the 

typewriter family attribute. It is equivalent to {\ttfamily text}. 

\textup{text} . . . . . . . . . . . . . . . . . . . . . . (4.1.4) – 65 

Sets text in a font in the current family and series, but with the 

upright shape attribute. It is equivalent to {\upshape text}.


\textwidth . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The total width reserved for the text on a page. For two-column 

formatting, this is the width of both columns plus the gap between 

them. A new value may be assigned with the \setlength command. 

\tfrac{numerator}{denominator} [m][a] . . . . . . . (12.2.3) – 264 

With the amsmath package, produces a fraction as \frac does, but 

in \textstyle size. 

\TH . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character Þ. 

\th . . . . . . . . . . . . . . . . . . . . . . . . . . (G.4.5) – 503 

When T1 encoding is active, prints the character þ. 

\thanks{footnote text} . . . . . . . . . . . . . . . . . . (3.3.1) – 54 

Produces a footnote to an author’s name on the title page when 

\maketitle is called. 

\thecounter . . . . . . . . . . . . . . . . . . . . . . (8.1.4) – 183 

Internal commands for formatting and printing counter values, making 

possible use of other counters. For example, \thesubsection 

might be defined to be \thesection.\roman{subsection}. A new 

definition may be made with \renewcommand{\thecounter}{def }. 

\Theta [m] produces Θ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\theta [m] produces θ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\thicklines . . . . . . . . . . . . . . . . . . . . . (13.1.4) – 300 

In the picture environment, this command sets all the sloping lines 

and arrows, circles, and ovals to be drawn with thicker than normal 

lines. 

\thickspace [m][a] . . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this is an alias for \;, a thick space in a 

math formula. 

\thinlines . . . . . . . . . . . . . . . . . . . . . (13.1.4) – 300 

In the picture environment, resets the line thickness for sloping 

lines and arrows, circles, and ovals back to the standard value after 

\thicklines has been given. 

\thinspace [m][a] . . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, this is an alias for \,, a thin space in a 

math formula. 

\thisfancypage{cmds1}{cmds2} . . . . . . . . . . . . (4.7.9) – 95 

With the fancybox package, places a framed box around the contents 

of the current page only; cmds1 exclude the head and footlines, cmds2 

includes them. The arguments set box parameters like \fboxrule 

but must end with a box command like \shadowbox. Usually one set 

of cmds is left blank.


\thispagestyle{style} . . . . . . . . . . . . . . . . . (3.2) – 42 

Changes the page style for the current page only. Possible values for 

style are: plain, empty, headings, and myheadings. 

\tilde{x} [m] . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Produces a tilde (squiggle) over the math variable x: \tilde{a} = 

ã. 

\Tilde{x} [m][a] . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \tilde, but with multiple 


\tiny . . . . . . . . . . . . . . . . . . . . . . . . . . (4.1.2) – 62 

Switches to the smallest font size available \tiny, smaller than 

\scriptsize. 

\times [m] produces × . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\title{text} . . . . . . . . . . . . . . . . . . . . . . . (3.3.1) – 52 

Enters the text for the title page that is produced by \maketitle. 

\to [m] produces → . . . . . . . . . . . . . . . . . . (5.3.5) – 127 

\today . . . . . . . . . . . . . . . . . . . (2.5.11), (D.4.2) – 27, 461 

Prints the current date in the American fashion. This form may be 

changed to British or to that of other languages by redefining the 

command with the help of the internal T E X commands \day, \month, 

and \year. 

\top [m] produces ⊤ . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\topfigrule . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 173 

A command that is executed after a float at the top of a page. It is 

normally defined to do nothing, but may be redefined to add a rule 

between the float and the main text. It must not add any net vertical 

spacing. 

\renewcommand{\topfigrule}{\vspace*{-.4pt} 

\rule{\columnwidth}{.4pt}} 

\topfraction . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

The maximum fraction of a page that may be occupied at the top by 

floats at the top of the page. A new value is assigned with 

\renewcommand{\topfraction}{decimal frac} 

\topmargin . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The size of the margin from the top of the page to the page head. A 

new value may be assigned by the \setlength command: 

\setlength{topmargin}{0.5in}


topnumber . . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 171 

The maximum number of floats that may appear at the top of a page. 

A new value is assigned with: 

\setcounter{topnumber}{num} 

\topsep . . . . . . . . . . . . . . . . . . . (4.4.2), (5.5.4) – 75, 149 

The extra vertical spacing, in addition to \parskip, inserted at the 

beginning and end of a listing environment. When document class 

option fleqn has been chosen, it is also inserted at the beginning 

and end of displayed math formulas. A new value may be assigned 


\setlength{\topsep}{4pt plus2pt minus2pt} 

\topskip . . . . . . . . . . . . . . . . . . . . . . . . (3.2.5) – 48 

The vertical distance from the top of the page body to the baseline 

of the first line of text. A new value may be assigned with the 


\setlength{\topskip}{12pt} 

\totalheight . . . . . . . . . . . . . . . . . . . . . . (4.7.1) – 86 

A length parameter equal to the total natural height of a box (height 

plus depth); it may only be used in the width specification of 

\makebox, \framebox, or \savebox, or in the height specification of 

a \parbox or a minipage environment. 

\framebox[6\totalheight]{text} 

totalnumber . . . . . . . . . . . . . . . . . . . . . . . (7.3) – 172 

The total number of floats that may appear on a page regardless of 

their positions. A new value is assigned with: 

\setcounter{totalnumber}{num} 

\triangle [m] produces △ . . . . . . . . . . . . . . . (5.3.6) – 127 

\triangleleft [m] produces ⊳ . . . . . . . . . . . . . (5.3.3) – 125 

\triangleright [m] produces ⊲ . . . . . . . . . . . . (5.3.3) – 125 

\ttdefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 


\ttfamily command. It may be redefined with \renewcommand: 

\renewcommand{\ttdefault}{pcr} 

\ttfamily . . . . . . . . . . . . . . . . . . . (4.1.3), A.2) – 64, 371 


but with the typewriter family attribute. 

\twocolumn[text] . . . . . . . . . . . . . . . . . . . . (3.2.7) – 51 

Begins a new page and switches to two-column page format. The 

optional text is set in one column extending over the two columns.


\typein[\cmd]{message} . . . . . . . . . . . . . . . (9.1.3) – 211 

Prints the message to the monitor and stops the program, waiting for 

a reply from the user. The text of the response is assigned to the L AT E X 

command named \@typein, or to \cmd if the optional argument has 

been given. After the return key is pressed, the processing continues. 

The typed-in text is inserted in place of \@typein if the optional 

argument was not given, otherwise it may be inserted as one pleases 

with the \cmd command. 

\typeout{message} . . . . . . . . . . . . . . . . . . (9.1.3) – 211 

Prints the message to the monitor and continues the processing. The 

message is also written to the.log file. 

\u{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces a breve accent: \u{o} = ŏ. 

\unboldmath . . . . . . . . . . . . . . . . . . . . . . (5.4.9) – 143 

Countermands the \boldmath command. It must be given outside 

of the math mode. Afterwards, formulas are set in standard ‘math 

italics’ once more. 

\underbrace{sub form} [m] . . . . . . . . . . . . . . (5.4.4) – 136 

Produces a horizontal curly brace beneath the math formula 

sub form. Any following subscript will be placed centered below 

the horizontal brace. 

\underbrace{a+b}: a + b 

 

\underbrace{x+y+z}_{\xi\eta\zeta}: x + y + z 

 

ξηζ 

\underleftarrow{expr} [m][a] . . . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long leftwards pointing arrow 

beneath the mathematical expression expr. 

\underleftrightarrow{expr} [m][a] . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long double arrow beneath the 

mathematical expression expr. 

\underline{text} . . . . . . . . . . . . . . . . . . . (5.4.4) – 136 

Underlines the text in both math and normal text modes: 

\underline{Text} = Text. 

\underrightarrow{expr} [m][a] . . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places a long rightwards pointing arrow 

beneath the mathematical expression expr.


\underset{char}{\symbol} [m][a] . . . . . . . . . . (12.2.2) – 262 

With the amsmath package, places char below the math symbol 

\symbol in subscript size. 

\unitlength . . . . . . . . . . . . . . . . . . . . . (13.1.1) – 288 

Defines the unit of length for the following picture environments. 

A value is assigned with the \setlength command: 

\setlength{\unitlength}{1.2cm} 

\unlhd [m] produces ⊴ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\unrhd [m] produces ⊵ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\Uparrow [m] produces ⇑ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\uparrow [m] produces ↑ . . . . . . . . . . . . . . . . (5.3.5) – 127 

\updefault . . . . . . . . . . . . . . . . . . . . . . (A.3.1) – 372 


\upshape command. It may be redefined with \renewcommand: 

\renewcommand{\updefault}{n} 

\Updownarrow [m] produces . . . . . . . . . . . . . (5.3.5) – 127 

\updownarrow [m] produces . . . . . . . . . . . . . (5.3.5) – 127 

\uplus [m] produces ⊎ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\upshape . . . . . . . . . . . . . . . . . . (4.1.3), (A.2) – 64, 371 


but with the upright shape attribute. 

\Upsilon [m] produces Υ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\upsilon [m] produces υ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\uproot{shift} [m][a] . . . . . . . . . . . . . . . . . (12.2.5) – 269 

With the amsmath package, used in the index to a \sqrt command to 

shift it slightly upwards. The shift is a number specifying how many 

units to move it. Example: 

\sqrt[\leftroot{-1}\uproot{3}\beta]{k} 

\url{address} . . . . . . . . . . . . . . . . . . . . . (4.9.2) – 112 

With the url package, prints address literally, normally in typewriter 

font, with line breaks after non-letters, without hyphens. Intended 

for Internet and email addresses. With the hyperref package, address 

becomes an active link in a PDF file. The address argument 

may be delimited either by curly braces as usual, or by some other 

character, as is done with \verb. Examples: 

\url{w_smith@xyz.com} or \url|w_smith@xyz.com| 

\urlstyle{style} . . . . . . . . . . . . . . . . . . . (4.9.2) – 112 

With the url package, sets the typeface for subsequent \url commands, 

where style is one of tt, rm, sf, or same, for typewriter, 

Roman, sans serif, or unchanged font, respectively.


\usebox{boxname} . . . . . . . . . . . . . . . . . . . (4.7.1) – 87 

Inserts into the text the contents of the box that was saved with the 

\sbox or \savebox command under the name \boxname, which has 

been previously created with the \newsavebox command. 

\usecounter{counter} . . . . . . . . . . . . . . . . . . (4.4.1) – 75 

Command in the list environment that specifies which counter is to 

be employed in the standard labels with the \item commands. This 

counter is incremented by one with each \item call. 

\usefont{code}{fam}{ser}{shp} . . . . . . . . . . . . . (A.1) – 369 

Activates the font with the given set of attributes in the current size. 

Is equivalent to selecting the given font attributes and then calling 

\selectfont. 

\usepackage[options]{packages}[version] [p] . . . . . . (3.1.2) – 41 

Loads one or more package files containing additional L AT E X or T E X 

definitions. The files have the extension.sty. More than one package 

may be specified in packages, the names being separated by commas. 

Any options listed will be applied to all packages. Furthermore, any 

options listed in the \documentclass command will also be applied 

to the package files. 


example 1994/08/01. If the date of the package file is earlier than 

this, a warning message is printed. 

Example: 

\usepackage{bezier,ifthen}[1994/06/01] 

\v{x} . . . . . . . . . . . . . . . . . . . . . . . . . . (2.5.7) – 24 

Produces háček accent: \v{o} = ǒ. 

\value{counter} . . . . . . . . . . . . . . . . . . . . (8.1.3) – 183 

The current value of the number stored in counter for use with commands 

that require a number. It does not output this number. For example, 

\setcounter{counter1}{\value{counter2}} sets counter1 

to the same value as that of counter2. 

\varepsilon [m] produces ε . . . . . . . . . . . . . . (5.3.1) – 125 

\varinjlim [m][a] produces ‘lim’ in formulas 

−→ 

\varliminf [m][a] produces ‘lim’ in formulas 

. . . . . 

. . . . . 

(12.2.5) – 268 

(12.2.5) – 268 

\varlimsup [m][a] produces ‘lim’ in formulas . . . . . (12.2.5) – 268 

\varphi [m] produces ϕ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\varpi [m] produces ϖ . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\varprojlim [m][a] produces ‘lim’ 

in formulas . . . . (12.2.5) – 268 

←− 

\varrho [m] produces ϱ . . . . . . . . . . . . . . . . (5.3.1) – 125 

\varsigma [m] produces ς . . . . . . . . . . . . . . . (5.3.1) – 125


\vartheta [m] produces ϑ . . . . . . . . . . . . . . . (5.3.1) – 125 

\vdash [m] produces ⊢ . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\vdots [m] produces . . . . . . . . . . . . . . . . . . . . (5.2.6) – 123 

\vec{x} [m] . . . . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

A vector symbol over the variable x: \vec{a} = a. 

\Vec{x} [m][a] . . . . . . . . . . . . . . . . . . . . (12.2.2) – 263 

With the amsmath package, can be used like \vec, but with multiple 


\vector(∆x,∆y){length} . . . . . . . . . . . . . . (13.1.4) – 295 

A picture element command within a picture environment for drawing 

horizontal and vertical arrows of any length as well as slanted 

arrows at a limited number of angles. For horizontal and vertical 

arrows, the length argument is the actual length in units of 

\unitlength. For slanted arrows, length is the length of the projection 

on to the x-axis (horizontal displacement). The slope is determined 

by the (∆x, ∆y) arguments, which take on integral values 

such that −4 ≤ ∆x ≤ 4 and −4 ≤ ∆y ≤ 4. This command is the 

argument of a \put or \multiput command. 

\vee [m] produces ∨ . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\verb|source text| . . . . . . . . . . . . . . . . . . . . (4.9) – 110 

Everything that comes between the |...| symbols is output in the 

typewriter font exactly as is with no interpretation of special symbols 

or commands. Any symbol other than * may be used as the switch 

character, illustrated here as |, as long as it does not appear in 

source text. 

\verb*|source text| . . . . . . . . . . . . . . . . . . . (4.9) – 110 

The same as \verb except that blanks are made visible with the 

symbol . 

\vfill . . . . . . . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

A vertical rubber spacing with a natural length of zero that can be 

stretched to any value. Used to fill up parts of a page with blank 

spacing. This command is an abbreviation for \vspace{\fill}. 

\visible . . . . . . . . . . . . . . . . . . . . . . (15.1.2) – 326 

In slides class, a declaration that countermands a previous 

\invisible command, making text printed again. It remains in 

effect until the end of the environment, or end of the curly braces, 

in which it was issued, or until \invisible is given. It is used for 

making overlays. 

\vline . . . . . . . . . . . . . . . . . . . . . . . . . . (4.8.1) – 97 

Prints a vertical rule within the column entry of a table in the tabular 

environment.


\voffset . . . . . . . . . . . . . . . . . . . . . . . . . 602, 603 

Vertical offset of the output page from the printer border set by the 

printer driver. This printer border is normally 1 inch from the top 

edge of the paper. The standard value of \voffset is 0 pt so that the 

top reference margin of the page is identical with the printer margin. 

A new value is assigned with the \setlength command: 

\setlength{\voffset}{-1in} 

\vpageref[current][non-current]{key} . . . . . . . . (9.2.4) – 215 

With the varioref package, this is equivalent to on page 

\pageref{key}, unless \label{key} is on the current or adjacent 

page, in which case appropriate text is automatically substituted, 

e.g., ‘on the next page’. The first optional argument is inserted for 

the current page, the second for other pages. 

\vref{key} . . . . . . . . . . . . . . . . . . . . . . (9.2.4) – 215 

With the varioref package, this is equivalent to \ref{key} on page 

\pageref{key}, unless \label{key} is on the current or adjacent 

page, in which case appropriate text is automatically substituted for 

the page specification, e.g., ‘on the next page’. 

\vspace{height} . . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

Produces vertical spacing of length height. It is ignored if it occurs 

at the beginning or end of a page. 

\vspace*{height} . . . . . . . . . . . . . . . . . . . . (2.7.3) – 32 

Produces vertical spacing of length height even at the beginning or 

end of a page. 

\wedge [m] produces ∧ . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\whiledo{test}{do text} . . . . . . . . . . . . . . . . (8.3.5) – 193 

A conditional command available when the standard package ifthen 

has been loaded. The do text is inserted repeatedly as long as 

the logical statement test evaluates to 〈true〉. The logical statement 

may be relational (two numbers with one of < = > between 

them), an even–odd test (\isodd{number}), a comparison 

of two texts (\equal{text1}{text2}), a comparison of two lengths 

(\lengthtest{length1 op length2}, op is one of < = >), or a test 

of a boolean switch (\boolean{switch}). Switches are created with 

\newboolean{switch} and set with \setboolean{switch}{value}, 

where value is true or false. Logical statements may be combined 

with logical operators \and, \or, and \not, and grouped with $ and 

$. 

\widehat{xyz} [m] . . . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Produces a wide \hat symbol over several characters: 

\widehat{xyz} = xyz.


\widetilde{xyz} [m] . . . . . . . . . . . . . . . . . (5.3.9) – 129 

Produces a wide \tilde symbol over several characters: 

\widetilde{xyz} = xyz. 

\width . . . . . . . . . . . . . . . . . . . . . . . . . . (4.7.1) – 86 

A length parameter equal to the natural width of a box; it may 

only be used in the width specification of \makebox, \framebox, or 

\savebox, or in the height specification of a \parbox or a minipage 

environment. 

\framebox[2\width]{text} 

\wp [m] produces ℘ . . . . . . . . . . . . . . . . . . . (5.3.6) – 127 

\wr [m] produces ≀ . . . . . . . . . . . . . . . . . . . (5.3.3) – 125 

\Xi [m] produces Ξ . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\xi [m] produces ξ . . . . . . . . . . . . . . . . . . . (5.3.1) – 125 

\xleftarrow[below]{above} [m][a] . . . . . . . . . (12.2.2) – 262 

With the amsmath package, draws leftward pointing arrow with above 

printed over it in superscript size, and optionally below beneath it in 

subscript size. 

\xrightarrow[below]{above} [m][a] . . . . . . . . . (12.2.2) – 262 

With the amsmath package, draws rightward pointing arrow with 

above printed over it in superscript size, and optionally below beneath 

it in subscript size. 

\zeta [m] produces ζ . . . . . . . . . . . . . . . . . (5.3.1) – 125

H.2 Summary tables and figures 

H.2. Summary tables and figures 595 

Table H.1 Font attribute commands (4.1.3) – p. 64. 

\rmfamily \textrm{text} Roman 

\sffamily \textsf{text} sans serif 

\ttfamily \texttt{text} typewriter 

\upshape \textup{text} upright 

\itshape \textit{text} italic 

\slshape \textsl{text} slanted 

\scshape \textsc{text} Small Caps 

\mdseries \textmd{text} medium 

\bfseries \textbf{text} bold face 

Table H.2 Math alphabet commands (5.4.2) – p. 133. 

\mathrm{text} Roman 

\mathsf{text} sansserif 

\mathnormal{text} nor mal 

\mathtt{text} typewriter 

\mathit{text} italic 

\mathbf{text} boldface 

\mathcal{text} CAL 

Table H.3 Font sizes (4.1.2) – p. 62. 

\tiny smallest 

\scriptsize very small 

\footnotesize smaller 

\small small 

\normalsize normal 

\large large 

\Large larger 

\LARGE even larger 

\huge still larger 

\Huge largest 

Table H.4 L AT E X 2.09 font declarations (F.2.1) – p. 485. 

\rm Roman \it Italic \sc Small Caps 

\bf Bold Face \sl Slanted \sf Sans Serif 

\tt Typewriter \mit Γ ΠΦ \cal CAL


Table H.5 Dimensions (2.4.1) – p. 21. 

mm millimeter bp big point (1 in = 72 bp) 

cm centimeter dd (1157 dd = 1238 pt) 

in inch (1 in = 2.54 cm) cc cicero (1 cc = 12 dd) 

pt point (1 in = 72.27 pt) sp (1 pt = 65536 sp) 

pc pica (1 pc = 12 pt) 

em The current width of a capital M 

ex The current height of the letter x 

Table H.6 Accents (2.5.7) – p. 24. 

ò =\‘{o} ó=\’{o} ô=\ˆ{o} ö=\"{o} õ=\˜{o} 

ō =\={o} ˙o=\.{o} ŏ=\u{o} ǒ=\v{o} ő=\H{o} 

oo=\t{oo} o¸=\c{o} o . =\d{o} ō=\b{o} ˚o=\r{o} 

Table H.7 Special letters from other languages (2.5.6) – p. 24. 

œ={\oe} Œ={\OE} æ={\ae} Æ={\AE} ˚a={\aa} ˚A ={\AA} ¡ =!‘ 

ø ={\o} Ø ={\O} ł ={\l} Ł ={\L} ß={\ss} SS={\SS} ¿=?‘ 

Table H.8 Special symbols (2.5.5) – 23. 

† \dag § \S © \copyright ‡ \ddag \P £ \pounds 

Table H.9 Command symbols (2.5.4) – 23. 

$ \$ % \% { \{ \_ & \& # \# } \} 

Table H.10 Greek letters (5.3.1) – p. 125. 

Lower case letters 

α \alpha θ \theta o o τ \tau 

β \beta ϑ \vartheta π \pi υ \upsilon 

γ \gamma ι \iota ϖ \varpi φ \phi 

δ \delta κ \kappa ρ \rho ϕ \varphi 

ɛ \epsilon λ \lambda ϱ \varrho χ \chi 

ε \varepsilon µ \mu σ \sigma ψ \psi 

ζ \zeta ν \nu ς \varsigma ω \omega 

η \eta ξ \xi 

Upper case letters 

Γ \Gamma Λ \Lambda Σ \Sigma Ψ \Psi 

∆ \Delta Ξ \Xi Υ \Upsilon Ω \Omega 

Θ \Theta Π \Pi Φ \Phi


Table H.11 Binary operation symbols (5.3.3) – p. 125. 

± \pm ∩ \cap ◦ \circ ○ \bigcirc 

∓ \mp ∪ \cup • \bullet \Box 

× \times ⊎ \uplus ⋄ \diamond ♦ \Diamond 

÷ \div ⊓ \sqcap ⊳ \lhd △ \bigtriangleup 

· \cdot ⊔ \sqcup ⊲ \rhd ▽ \bigtriangledown 

∗ \ast ∨ \vee ⊴ \unlhd ⊳ \triangleleft 

⋆ \star ∧ \wedge ⊵ \unrhd ⊲ \triangleright 

† \dagger ⊕ \oplus ⊘ \oslash \ \setminus 

‡ \ddagger ⊖ \ominus ⊙ \odot ≀ \wr 

∐ \amalg ⊗ \otimes 

Table H.12 Relational symbols (5.3.4) – p. 126. 

≤ \le 

≪ \ll 

\leq ≥ \ge 

≫ \gg 

\geq ≠ \neq 

. 

= \doteq 

∼ \sim 

\simeq 

⊂ \subset ⊃ \supset ≈ \approx ≍ \asymp 

⊆ \subseteq ⊇ \supseteq ≅ \cong ⌣ \smile 

⊏ \sqsubset ⊐ \sqsupset ≡ \equiv ⌢ \frown 

⊑ \sqsubseteq ⊒ \sqsupseteq ∝ \propto ⋈ \bowtie 

∈ \in ∋ \ni ≺ \prec ≻ \succ 

⊢ \vdash ⊣ \dashv \preceq \succeq 

⊨ \models ⊥ \perp \parallel \| | \mid | 

Table H.13 Negated relational symbols (5.3.4) – p. 127. 

< \not< > \not> = \not= 

≤ \not\le ≥ \not\ge ≡ \not\equiv 

≺ \not\prec ≻ \not\succ ∼ \not\sim 

\not\preceq \not\succeq \not\simeq 

⊂ \not\subset ⊃ \not\supset ≈ \not\approx 

⊆ \not\subseteq ⊇ \not\supseteq ≅ \not\cong 

⊑ \not\sqsubseteq ⊒ \not\sqsupseteq ≍ \not\asymp 

∈ \not\in ∉ \notin 

Table H.14 Brackets (5.4.1) – p. 132. 

( ( ) ) ⌊ \lfloor ⌋ \rfloor 

[ [ ] ] ⌈ \lceil ⌉ \rceil 

{ \{ } \} 〈 \langle 〉 \rangle 

| | \| ↑ \uparrow ⇑ \Uparrow 

/ / \ \backslash ↓ 

 

\downarrow 

\updownarrow 

⇓ 

 

\Downarrow 

\Updownarrow


Table H.15 Arrows (5.3.5) – p. 127. 

← \leftarrow \gets ←− \longleftarrow ↑ \uparrow 

⇐ \Leftarrow ⇐= \Longleftarrow ⇑ \Uparrow 

→ \rightarrow \to −→ \longrightarrow ↓ \downarrow 

⇒ \Rightarrow =⇒ \Longrightarrow ⇓ \Downarrow 

↔ \leftrightarrow ←→ \longleftrightarrow \updownarrow 

⇔ \Leftrightarrow ⇐⇒ \Longleftrightarrow \Updownarrow 

↦→ \mapsto ↦−→ \longmapsto ↗ \nearrow 

←↪ \hookleftarrow ↩→ \hookrightarrow ↘ \searrow 

↼ \leftharpoonup ⇀ \rightharpoonup ↙ \swarrow 

↽ \leftharpoondown ⇁ \rightharpoondown ↖ \nwarrow 

⇌ \rightleftharpoons ⇝ \leadsto 

Table H.16 Miscellaneous symbols (5.3.6) – p. 127. 

ℵ \aleph ′ \prime ∀ \forall \Box 

ℏ \hbar ∅ \emptyset ∃ \exists ♦ \Diamond 

ı \imath ∇ \nabla ¬ \neg △ \triangle 

j \jmath √ \surd ♭ \flat ♣ \clubsuit 

ℓ \ell ∂ \partial ♮ \natural ♦ \diamondsuit 

℘ \wp ⊤ \top ♯ \sharp ♥ \heartsuit 

ℜ \Re ⊥ \bot \| ♠ \spadesuit 

ℑ \Im ⊢ \vdash ∠ \angle ⋈ \Join 

\mho ⊣ \dashv \ \backslash ∞ \infty 

The underlined commands in Tables H.11, H.12, H.15, and H.16 can only be used 

in L AT E X 2ε if one of the packages latexsym or amsfonts has been loaded. 

Table H.17 Mathematical symbols in two sizes (5.3.7) – p. 128. 

 

\sum 

\bigcap 

\bigodot 

 

 

 

\int 

\bigcup 

\bigotimes 

 

 

 

\oint 

\bigsqcup \bigoplus 

 

\prod 

\bigvee 

\biguplus 

\coprod \bigwedge 

Table H.18 Function names (5.3.8) – p. 128. 

\arccos \cosh \det \inf \limsup \Pr \tan 

\arcsin \cot \dim \ker \ln \sec \tanh 

\arctan \coth \exp \lg \log \sin 

\arg \csc \gcd \lim \max \sinh 

\cos \deg \hom \liminf \min \sup


Table H.19 Math accents (5.3.9) – p. 129. 

â \hat{a} ă \breve{a} à \grave{a} ā \bar{a} 

ǎ \check{a} á \acute{a} ã \tilde{a} a \vec{a} 

˙a \dot{a} ä \ddot{a} ˚a \mathring{a} 

The following symbols are made available with the package amssymb. 

Table H.20 A M S arrows 

⇢ \dashrightarrow ⇠ \dashleftarrow 

⇔ \leftleftarrows ⇆ \leftrightarrows 

⇚ \Lleftarrow ↞ \twoheadleftarrow 

↢ \leftarrowtail ↫ \looparrowleft 

⇋ \leftrightharpoons ↶ \curvearrowleft 

↺ \circlearrowleft ↖ \Lsh 

⇈ \upuparrows ↿ \upharpoonleft 

⇃ \downharpoonleft ⊸ \multimap 

↭ \leftrightsquigarrow ⇒ \rightrightarrows 

⇄ \rightleftarrows ⇒ \rightrightarrows 

⇄ \rightleftarrows ↠ \twoheadrightarrow 

↣ \rightarrowtail ↬ \looparrowright 

⇌ \rightleftharpoons ↷ \curvearrowright 

↻ \circlearrowright ↗ \Rsh 

⇊ \downdownarrows ↾ \upharpoonright 

⇂ \downharpoonright ⇝ \rightsquigarrow 

Negated arrows 

↚ \nleftarrow ↛ \nrightarrow 

⇍ \nLeftarrow ⇏ \nRightarrow 

\nleftrightarrow \nLeftrightarrow 

Table H.21 A M S binary operation symbols 

∔ \dotplus \smallsetminus 

⋓ \Cap ⋒ \Cup 

⊼ \barwedge ⊻ \veebar 

⊼ \doublebarwedge ⊟ \boxminus 

⊠ \boxtimes ⊡ \boxdot 

⊞ \boxplus ⋇ \divideontimes 

⋉ \ltimes ⋊ \rtimes 

⋋ \leftthreetimes ⋌ \rightthreetimes 

⋏ \curlywedge ⋎ \curlyvee 

⊝ \circleddash ⊛ \circledast 

⊚ \circledcirc \centerdot 

⊺ \intercal


Table H.22 A M S Greek and Hebrew letters 

F \digamma ϰ \varkappa 

\beth \daleth ג \gimel 

Table H.23 A M S delimiters 

\ulcorner \urcorner \llcorner \lrcorner 

Table H.24 A M S relational symbols 

≦ \leqq \leqslant 

⋜ \eqslantless ≲ \lesssim 

\lessapprox ≊ \approxeq 

⋖ \lessdot ⋘ \lll 

≶ \lessgtr ⋚ \lesseqgtr 

\lesseqqgtr ≑ \doteqdot 

≓ \risingdotseq ≒ \fallingdotseq 

∽ \backsim ⋍ \backsimeq 

\subseteqq ⋐ \Subset 

⊏ \sqsubset ≼ \preccurlyeq 

⋞ \curlyeqprec ≾ \precsim 

\precapprox ⊳ \vartriangleleft 

⊴ \trianglelefteq ⊨ \vDash 

⊪ \Vvdash ⌣ \smallsmile 

⌢ \smallfrown ≏ \bumpeq 

≎ \Bumpeq ≧ \geqq 

\geqslant ⋝ \eqslantgtr 

≳ \gtrsim \gtrapprox 

⋗ \gtrdot ⋙ \ggg 

≷ \gtrless ⋛ \gtreqless 

\gtreqqless ≖ \eqcirc 

≗ \circeq ≜ \triangleq 

∼ \thicksim ≈ \thickapprox 

\supseteqq ⋑ \Supset 

⊐ \sqsupset ≽ \succcurlyeq 

⋟ \curlyeqsucc ≿ \succsim 

≿ \succapprox ⊲ \vartriangleright 

⊵ \trianglerighteq ⊩ \Vdash 

\shortmid \shortparallel 

≬ \between ⋔ \pitchfork 

∝ \varpropto ◭ \blacktriangleleft 

∴ \therefore ∋ \backepsilon 

◮ \blacktriangleright ∵ \because


Table H.25 A M S negated relational symbols 

≮ \nless \nleq 

≰ \nleqslant \nleqq 

\lneq ≨ \lneqq 

\lvertneqq ⋦ \lnsim 

\lnapprox ⊀ \nprec 

\npreceq ⋨ \precnsim 

\precnapprox ≁ \nsim 

\nshortmid ∤ \nmid 

⊬ \nvdash \nvDash 

⋪ \ntriangleleft ⋬ \ntrianglelefteq 

⊈ \nsubseteq ⊊ \subsetneq 

\varsubsetneq \subsetneqq 

\varsubsetneqq ≯ \ngtr 

\ngeq ≱ \ngeqslant 

\ngeqq \gneq 

≩ \gneqq \gvertneqq 

⋧ \gnsim \gnapprox 

⊁ \nsucc \nsucceq 

\nsucceq \succnsim 

⋩ \succnapprox ≇ \ncong 

\nshortparallel ∦ \nparallel 

\nvDash \nVDash 

⋫ \ntriangleright ⋭ \ntrianglerighteq 

⊉ \nsupseteq \nsupseteqq 

⊋ \supsetneq \varsupsetneq 

\supsetneqq \varsupsetneqq 

Table H.26 Miscellaneous A M S symbols 

\hbar ℏ \hslash 

△ \vartriangle ▽ \triangledown 

\square ♦ \lozenge 

\circledS ∠ \angle 

∡ \measuredangle ∄ \nexists 

\mho \Finv 

\Game k \Bbbk 

\backprime \varnothing 

\blacktriangle \blacktriangledown 

\blacksquare \blacklozenge 

\bigstar ∢ \sphericalangle 

∁ \complement ∂ \eth 

\diagup \diagdown


✲ ❄ h v ❄ 

\voffset 

✻ 

❄ 

✻ \headheight 

❄ 


❄ \headsep 


Rev. 

marg. 

note 

Marg. 

note, 

even 

pages 

✛ ✲ 

❄ 

\textheight 

❄ 

✍☞ \oddsidemargin 

▲ \evensidemargin 

✲ ✛ \hoffset 

✻ 

Head 

First line of text 

Body 

\textfraction×\textheight 

Min. space for text 

when floats are present 

❄ 

\topfraction×\textheight 

Max. space for floats at top 

✛ ✲ 

\textwidth 

Foot 



\marginparsep ✲ 

\bottomfraction×\textheight 

Max. space for floats at bottom 

\footskip 

Figure H.1 Single column page format 

(3.2.5), p. 48 – (4.10.6), p. 118 – (7.3), p. 172. 

❄ 

✻ 

❄ 

✻ 

❄ 

Normal 

marg. 

note 

✻ 

❄ 

✛ 

☎ 

✲ 

✛ 

Marg. 

note, 

odd 

pages

✲ ❄ h v ❄ 

\voffset 

✻ 

❄ 

✻ \headheight 

❄ 


❄ \headsep 


Marg. 

note, 

left 

column 

✛ ✲ 

❄First 

line of text 


Head 

Left column Right column 

✛ \columnwidth ✲ 

\textheight 

✛ ✲ 

\textwidth 

❄ 

✍☞ \oddsidemargin 

▲ \evensidemargin 

✲ ✛ \hoffset 

✲ ✛ \columnseprule 

(line thickness) 

✲✛ 

\columnsep 

Foot 

✻ 

\dbltopfraction× 

\textheight 

Max. space for floats 

extending over 

two columns 

❄ 



\marginparsep ✲ 

Reserved space for floats 

within one column as for 

the single column format 

\footskip 

Figure H.2 Double column page format 

(3.2.5), p. 48 – (3.1.1), p. 40 – (4.10.6), p. 118 – (7.3), p. 172. 

❄ 

Marg. 

note, 

right 

column 

✻ 

❄ 

✛ 

☎ 

✲ 

✛


Remarks on the page format figures 

The reference margins in the L AT E X processing are shifted from the logical 

margins by the amounts \hoffset and \voffset. These in turn are 

displaced from the physical margins by h and v in the DVI driver. The 

default values for \hoffset and \voffset are 0 pt, so that the reference 

margins are equal to the logical ones. The usual values for h and v are 

1 inch. Thus the logical page margins on the left and at the top are shifted 

1 inch from the physical edge of the paper. The user may alter this by 

changing the values of \hoffset and \voffset. 

L AT E X 2ε recognizes the parameters \paperwidth and \paperheight 

which contain the full dimensions of the paper, including the 1 inch 

margins. These are set by the paper size option to the class specification. 

The parameter \footheight was specified but never used in L AT E X 2.09; 

it has been dropped from L AT E X 2ε. 

Preceding text 

✻ 


\labelsep 

✛ ✲ 

❄ 

✛ ✲ 

Label 

\rightmargin 

✛ ✲ ✛ ✲ 

Item 1 

✛ 

\labelwidth 

✲ \itemindent Paragraph 1 

\leftmargin 

\listparindent ✻\parsep 

✛ ✲ ❄ 

Item 1 

Paragraph 2 

Label 

✻\itemsep 

+ \parsep 

❄ 

Item 2 

✻ 


❄ 

Following text 

Figure H.3 Format of the list environment (4.4.2) – p. 75. 

Note 1: The default values for the three parameters \itemindent, 

\listparindent, \rightmargin are 0 pt. 

Note 2: The default values in the trivlist environment are 0 pt for 

\itemindent, \leftmargin, \rightmargin, and \labelwidth; on 

the other hand, \parsep and \listparindent are assigned the 

respective values of \parskip and \parindent.

Bibliography 

Abrahams P. W., with Hargreaves K. A. and Berry K. (1990). T E X for the 

Impatient. Reading MA: Addison-Wesley 

Beccari, C. (1997). Typesetting mathematics for science and technology 

according to ISO 31/XI. TUGboat, 18(1), 39–48 

Botway L. and Biemesderfer C. (1985). L AT E X Command Summary. Providence 

RI: T E X Users Group 

Eijkhout V. (1992). T E X by Topic, a T E Xnician’s Reference. Harlow: Addison- 

Wesley 

Flynn P. (1995). HTML & T E X: Making them sweat. TUGboat, 16(2), 146–150 

Haralambous Y. and Rahtz S. (1995). L AT E X, hypertext and PDF, or the entry 

of T E X into the world of hypertext. TUGboat, 16(2), 162–173 

Goossens M. and Saarela J. (1995). From L AT E X to HTML and back. TUGboat, 

16(2), 174–214 

Goossens M., Mittelbach F. and Samarin A. (1994). The L AT E X Companion. 

Reading MA: Addison-Wesley 

Goossens M., Rahtz S. and Mittelbach F. (1997). The L AT E X Graphics Companion. 


Goossens M. and Rahtz S. (1999). The L AT E X Web Companion. Reading MA: 


Knuth D. E. (1986a). The T E Xbook, Computers and Typesetting, Vol. A. 


Knuth D. E. (1986b). T E X: The Program, Computers and Typesetting, Vol. B. 


Knuth D. E. (1986c). The METAFONT book, Computers and Typesetting, 

Vol. C. Reading MA: Addison-Wesley 

Knuth D. E. (1986d). METAFONT: The Program, Computers and Typesetting, 

Vol. D. Reading MA: Addison-Wesley 

605

606 BIBLIOGRAPHY 

Knuth D. E. (1986e). Computer Modern Typefaces, Computers and Typesetting, 

Vol. E. Reading MA: Addison-Wesley 

Kopka H. (1994). L AT E X, Band 1, eine Einführung. Bonn: Addison-Wesley 

Kopka H. (1995). L AT E X, Band 2, Ergänzungen, mit einer Einführung in 

METAFONT. Bonn: Addison-Wesley 

Kopka H. (1997). L AT E X, Band 3, Erweiterungen. Bonn: Addison-Wesley 

Lamport L. (1985). L AT E X—A Document Preparation System. Reading MA: 


Lamport L. (1994). L AT E X—A Document Preparation System, 2nd edn. for 

L AT E X 2ε. Reading MA: Addison-Wesley 

Marchal, B. XML by Example. Indianapolis IN: Que-Programming 

Merz T. (1997). PostScript & Acrobat/PDF. Berlin: Springer-Verlag 

Merz T. (1998). Web Publishing with Acrobat/PDF. Berlin: Springer-Verlag 

Rokicki T. (1985). Packed (PK) font file format. TUGboat, 6(3), 115–20 

Reckdahl, K. (1996a). Using EPS graphics in L AT E X 2ε documents, part 1: 

The graphics and graphicx packages. TUGboat, 17(1), 43–53 

Reckdahl K. (1996b). Using EPS graphics in L AT E X 2ε documents, part 2: 

Floating figures, boxed figures, captions, and math in figures. TUGboat, 

17(3), 288–310 

Samuel A. L. (1985). First Grade T E X: A Beginner’s T E X Manual. Providence, 

RI: T E X Users Group 

Schwarz N. (1990). Introduction to T E X. Reading MA: Addison-Wesley 

Snow W. (1992). T E X for the Beginner. Reading MA: Addison-Wesley 

Sojka P., Thành H. T. and Zlatuˇska J. The joy of T E X2PDF—Acrobatics with 

an alternative to DVI format. TUGboat, 17(3), 244–251 

Spivak M. (1986). The Joy of T E X . Providence RI: American Mathematical 

Society 

Taylor P. (1996). Computer typesetting or electronic publishing? New 

trends in scientific publication. TUGboat, 17(4), 367–381 

Urban M. (1986). An Introduction to L AT E X . Providence RI: T E X Users Group 

Williamson, H. A. XML from A to Z. Redmond WA: Redmond Technology 

Press

Index 

For purposes of alphabetization, the backslash character \ is ignored at the start 

of an entry. Otherwise, the ordering is by the ascii sequence. 

Bold page numbers indicate the place where the command or concept is introduced, 

explained, or defined. Slanted page numbers refer to the Command 

Summary, Appendix H. 

The keyword index is set up with main and two sub-entries. If a keyword cannot 

be found as a main entry, one should try to find it as a sub-entry to some more 

general term. Such major topics are 

A M S-L AT E X, bibliographic database, bibliography, box, command, command 

(user-defined), cross-reference, error messages, exercises, file types, float, 

fonts, footnote, formula, hyphenation, L AT E X, L AT E X counters, letter, line 

breaking, lists, package, page breaking, page formatting, page numbering, 

page style, picture, PostScript, programming, sectioning, seminar, slides, 

spacing, style parameter, symbols, tabbing, table, table examples, T E X, text. 

! (MakeIndex), 226, 507 

\!, 145, 269, 507 

!‘, 24, 507 

" (BIBT E X), 311, 508 

" (MakeIndex), 227, 508 

", 23, 508 

\" (¨ accent), 24, 508 

##, 204, 508 

\#, 23, 425, 508 

#, 17, 187, 425, 429, 508 

$$, 120 

\$, 17, 23, 27, 28, 413, 508 

$, 17, 120, 413, 425, 426, 508 

\%, 23, 27, 28, 118, 508 

%, 17, 22, 60, 111, 118, 508 

\&, 23, 508 

&, 17, 97, 139, 272, 274, 425, 508 

\’ (tabbing), 83, 508 

’’, 23 

’, 23 

\’ (´ accent), 24, 508 

( (BIBT E X), 312, 509 

$, 119, 416, 425, 509 

(, 132, 509 

( (picture), 289, 509 

) (BIBT E X), 312, 509 

$, 119, 416, 509 

), 132, 509 

) (picture), 289, 509 

\+ (tabbing), 82, 84, 420, 509 

\,, 29, 123, 144, 145, 269, 509 

\- (tabbing), 82, 84, 420, 509 

\- (hyphenation), 35, 40, 427, 434, 

509 

-, 23, 509 

--, 23, 509 

---, 23, 509 

\. (˙ accent), 24, 509 

\/, 26, 509 

/, 132 

\:, 145, 269, 509 

\;, 145, 269, 509 

\< (tabbing), 82, 415, 420, 510 

\= (tabbing), 81, 82, 83, 419, 510 

607

608 INDEX 

\= (¯ accent), 24, 510 

\> (tabbing), 81, 82–4, 420, 510 

?‘, 24, 510 

@ (BIBT E X), 312–14, 510 

@ (MakeIndex), 226, 510 

\@, 29, 510 

@-expression, 96, 101, 110 

\@date, 363 

\@evenfoot, 362, 455 

\@evenhead, 362, 455 

\@for, 450 

\@ifnextchar, 449 

\@ifstar, 449 

\@ifundefined, 449 

\@latexerr, 439 

\@namedef, 449 

\@nameuse, 449, 458 

\@oddfoot, 362, 455 

\@oddhead, 362, 455 

\@onlypreamble, 455 

\@seccntformat, 457 

\@startsection, 456 

\@warning, 439 

\[, 120, 124, 416, 425, 510, 549 

[, 18, 132, 426, 510 

\\, 31, 32, 53, 67, 68, 81, 84, 97, 99, 

104, 271, 297, 352, 419, 

425, 434, 510 

\\*, 31, 279, 510 

\], 120, 124, 416, 510 

], 18, 132, 510 

\ˆ, 23 

ˆ, 17, 121, 123, 424, 511 

\ˆ (ˆ accent), 24, 511 

\_, 23, 414, 511 

_, 17, 121, 123, 413, 424, 511 

\‘ (tabbing), 83, 511 

‘, 23 

‘‘, 23 

\‘ (` accent), 24, 511 

{, 17, 18, 312, 425, 511 

\{, 23, 120, 132, 511 

|, 126, 132, 270, 511 

| (MakeIndex), 227, 511 

\|, 126, 127, 132, 270, 511 

\| (table), 96 

}, 17, 18, 312, 425, 426, 428, 511 

\}, 23, 120, 132, 511 

˜, 17, 22, 28, 511 

\˜ (˜ accent), 24, 511 

10pt option, 38, 61 

11pt option, 38, 61 

12pt option, 38, 61, 456 

8r.enc, 235, 236 

\ , 22, 28, 507 

\a’ (tabbing), 83, 512 

a4paper option, 38, 359, 447, 452, 

456 

a5paper option, 38 

\a= (tabbing), 83, 512 

\a‘ (tabbing), 83, 512 

\AA, 24, 512 

\aa, 24, 512 

abbrv bibliography style, 311 

abbrvnat bibliography style, 221, 

311 

\abovedisplayshortskip, 149, 512 

\abovedisplayskip, 149, 512 

Abrahams, P. W., 605 

abstract, 55 

abstract environment, 55, 516 

\abstractname, 460, 512 

accents 

in tabbing environment, 83 

in other languages, 24 

mathematical, 129 

Acrobat, 16, 236, 237, 242, 247, 249 

Acrobat Distiller, 237, 239 

\Acrobatmenu (hyperref), 247, 342 

\acute, 129, 263, 512 

\Acute (A M S), 263, 512 

\addButton (pdfscreen), 342 

\addcontentsline, 59, 60, 174, 

427, 436, 512 

\adddialect (babel), 255 

\addlanguage (babel), 255 

\address, 351, 352, 353, 356, 512 

\addtime (slides), 327, 512 

\addtoartlength (seminar), 332 

\addtocontents, 59, 60, 174, 427, 

436, 513 

\addtocounter, 21, 110, 115, 182, 

183, 418, 513 

\addtolength, 184, 200, 453, 513 

\addtoslidelength (seminar), 332 

\addtoslidereset (seminar), 337

\addvspace, 185, 513 

Adobe Systems Inc., 25, 231, 236, 

249, 387, 504 

\AE, 24, 513 

\ae, 24, 513 

Afrikaans, 252 

afterpage package, 171, 394 

\aleph, 127, 513 

align environment (A M S), 270, 

274, 516 

align page style (seminar), 335 

alignat environment (A M S), 270, 

275, 516 

aligned environment (A M S), 276, 

516 

\allinethickness (eepic), 306 

\allowdisplaybreaks (A M S), 278, 

513 

alltt environment, 111, 392, 517 

alltt package, 111, 392 

allversions* environment 

(seminar), 336 

\Alph, 73, 114, 183, 278, 513 

\alph, 73, 114, 183, 191, 192, 513 

alpha bibliography style, 311 

\alpha, 125, 513 

alphabetic page numbering, 45 

\alsoname, 460, 513 

\amalg, 126, 513 

American Mathematical Society, 

151, 257 

American spelling, 195 

amsart class, 258 

amsbook class, 258 

amsbsy package, 144, 192, 258–9 

amscd package, 258, 263, 282 

amsfndoc.tex, 257 

amsfonts package, 126, 284–5, 

422, 494 

A M S-L AT E X, 257–85 

\Acute, 263, 512 

align environment, 270, 274, 

516 

alignat environment, 270, 

275, 516 

aligned environment, 276, 

516 

\allowdisplaybreaks, 278, 

513 

INDEX 609 

arrows, extended, 262 

\Bar, 263, 515 

\binom, 264, 525 

binomials, 264 

Bmatrix environment, 266, 

517 

bmatrix environment, 266, 

517 

\boldsymbol, 144, 258, 526 

\boxed, 142, 270, 526 

\Breve, 263, 526 

cases environment, 277, 517 

CD environment, 282 

\cfrac, 265, 527 

\Check, 263, 528 

CM fonts, 283 

commutative diagrams, 282 

Cyrillic fonts, 283, 497 

\dbinom, 265, 531 

\ddddot, 263, 532 

\dddot, 263, 532 

\Ddot, 263, 532 

\DeclareMathOperator, 268, 

282, 534 

\dfrac, 264, 537 

\displaybreak, 278, 538 

\Dot, 263, 539 

dots, 263 

\dots, 263, 539 

\dotsb, 264, 539 

\dotsc, 264, 539 

\dotsi, 264, 539 

\dotsm, 264, 539 

\eqref, 278, 541 

equation* environment, 272 

falign environment, 270, 

275, 518 

fonts, 283–5 

fractions, 264 

continued, 265 

user-defined, 265 

function names, 267 

defining, 268 

gather environment, 270, 

273, 519 

gathered environment, 276, 

519 

\genfrac, 265, 545 

\Grave, 263, 546

610 INDEX 

\Hat, 263, 546 

\hdotsfor, 267 

\idotsint, 260, 548 

\iiiint, 260, 551 

\iiint, 260, 551 

\iint, 260, 551 

\intertext, 259, 550 

\leftroot, 269, 554 

limits 

multiline, 261 

special, 261 

\lVert, 270, 557 

\lvert, 270, 557 

math symbols, 285 

\mathbb, 284 

matrix, 266 

matrix environment, 266, 519 

\medspace, 269, 560 

\mod, 268, 560 

\mspace, 269, 560 

multiline equations, 270–7 

page breaks, 278 

multiple integrals, 260 

multline environment, 270, 

271, 520 

\multlinegap, 271, 561 

\negmedspace, 269, 561 

\negthickspace, 269, 561 

\negthinspace, 269, 561 

\notag, 271, 565 

\numberwithin, 277, 565 

options, 279 

\overleftarrow, 262, 566 

\overleftrightarrow, 262, 

566 

\overrightarrow, 262, 567 

\overset, 262, 567 

pmatrix environment, 266, 

521 

\pmb, 258, 570 

\pod, 268, 570 

\raisetag, 271 

roots, 269 

\rVert, 270, 576 

\rvert, 270, 576 

\shoveleft, 271 

\shoveright, 271 

\sideset, 262, 579 

smallmatrix environment, 

267 

\smash, 269, 580 

split environment, 270, 272, 

276, 521 

subarray environment, 261, 

522 

subequations environment, 

277, 522 

\substack, 261, 581 

\tag, 271, 573, 583 

\tbinom, 265, 583 

\text, 259, 583 

\tfrac, 264, 586 

theorems, 280 

\theoremstyle, 281 

\thickspace, 269, 586 

\thinspace, 269, 586 

\Tilde, 263, 587 

\underleftarrow, 262, 589 

\underleftrightarrow, 262, 

589 

\underrightarrow, 262, 589 

\underset, 262, 590 

upright references, 282 

\uproot, 269, 590 

\varinjlim, 268, 591 

\varliminf, 268, 591 

\varlimsup, 268, 591 

\varprojlim, 268, 591 

\Vec, 263, 592 

vertical bars, 270 

Vmatrix environment, 266, 

524 

vmatrix environment, 266, 

524 

\xleftarrow, 263, 594 

\xrightarrow, 263, 594 

amsldoc.tex, 257, 282 

amsmath package, 192, 258 

amsopn package, 258, 267–8 

amsproc class, 258 

amssymb package, 284, 285 

A M S-T E X, 257 

amstex format, 257 

amstext package, 258–9 

amsthm package, 81, 258, 280–2, 

395 

\and, 53, 513

\angle, 127, 513 

ANSI coding, 26, 462 

apalike package, 220 

appendix, 57 

appendix environment, 516 

\appendix, 57 

\appendixname, 460, 513 

Apple Macintosh coding, 26, 462 

\approx, 126, 514 

Arabic page numbering, 45 

\arabic, 73, 75, 77, 78, 114, 183, 

191, 277, 514 

\arc (eepic), 306 

\arccos, 128, 514 

\arcsin, 128, 514 

\arctan, 128, 514 

\arg, 128, 514 

argument, command, 18 

array environment, 95, 107, 134, 

146, 147, 149, 150, 266, 

394, 417, 418, 425, 428, 

429, 516 

array package, 107, 394 

\arraycolsep, 98, 149, 514 

\arrayrulewidth, 98, 514 

arrays, 133–6 

\arraystretch, 98, 514 

arrows, 127 

extended (A M S), 262 

Arseneau, Donald, 112, 224 

article class, 37, 41, 55, 113, 131, 

173, 191, 218, 225, 277, 

391, 397, 454, 459, 471 

article.cls, 397, 430 

article.sty (L AT E X 2.09), 430 

\articlemag (seminar), 332 

ascii, 5, 14, 17, 228, 231, 382, 487 

\askforoverwritefalse (DocStrip), 

465 

\ast, 126, 514 

\asymp, 126, 514 

\AtBeginDocument, 444, 472, 514 

\AtEndDocument, 444, 514 

\AtEndOfClass, 444, 515 

\AtEndOfPackage, 444, 515 

\atop (T E X), 137, 146, 147, 189 

attributes, font, 64–5, 368–72 

defaults, 372 

\author, 52, 53, 54, 69, 431, 515 

INDEX 611 

.aux file, 209, 210, 214, 396, 397, 

418, 427, 430, 436 

auxiliary file, 214 

avant package, 234 

AvantGarde, PostScript font, 234, 

504 

axioms, 80 

b5paper option, 38 

\b ( ¯ accent), 24, 515 

babel package, 252–6 

babel.def (babel), 253 

babel, multilingual L AT E X, 216, 

252–6, 385, 460 

background package, 345–6 

\backgroundcolor (pdfscreen), 

342 

\backmatter, 57, 210, 515 

\backslash, 127, 132, 515 

Bahasa, 252 

bakoma fonts, 506 

bakomaextra.map, 506 

\bar, 129, 263, 515 

\Bar (A M S), 263, 515 

Barroca, Leonor, 159 

baseline, 46, 63 

\baselineskip, 63, 362, 369, 515 

\baselinestretch, 46, 47, 63, 515 

Basque, 252 

\batchinput (DocStrip), 466 

.bbl file, 219, 224, 310, 397 

Beccari, Claudio, 605 

\begin, 19, 64, 416–18, 516 

\belowdisplayshortskip, 149, 524 

\belowdisplayskip, 149, 524 

Berry naming scheme, 504 

Berry, Karl, 381, 504, 605 

\beta, 125, 524 

\bf (L AT E X 2.09), 65, 367, 485 

\bfdefault, 233, 372, 524 

\bfseries, 19, 20, 64, 371, 416, 524 

\bgadd (background), 346 

\bgaddcenter (background), 346 

\bgclear (background), 346 

.bib file, 309, 397 

\bibitem, 217, 219–21, 322, 395, 

429–31, 524 

bibliographic database, 219, 309–21 

abbreviations, 320

612 INDEX 

creating, 311 

cross-referencing, 316, 317 

entry type, 311, 312, 314 

field, 312 

abbreviations, 319 

ignored, 312 

list of, 314, 316 

names, 317, 318 

optional, 312, 314 

required, 312, 314 

special formats, 317 

titles, 318, 319 

journal abbreviations, 322 

structure, 311 

template, 311, 320 

bibliography, 216 

abbrv, 311 

abbrvnat, 221, 311 

alpha, 311 

author–year style, 220–4 

BIBT E X, 217, 219, 224, 309–21 

customizing, 223, 321–2 

entry, 217 

format, see style 

multiple, 224 

numerical, 219 

open style, 39 

plain, 219, 310 

plainnat, 219, 221, 311 

reference in text, 219, 310 

style, 219, 221, 310 

unsrt, 310 

unsrtnat, 221, 311 

\bibliography, 210, 219, 224, 309, 

310, 397, 524 

\bibliographystyle, 219, 221, 

224, 310, 397, 524 

\bibname, 460, 525 

BIBT E X, 15, 217, 219, 224, 309–21, 

382, 388, 397, 430 

writing styles for, 321 

Biemesderfer, C., 605 

\Big (T E X), 148, 525 

\big (T E X), 148, 525 

\bigcap, 128, 525 

\bigcirc, 126, 525 

\bigcup, 128, 525 

\Bigg (T E X), 148, 525 

\bigg (T E X), 148, 525 

\bigodot, 128, 525 

\bigoplus, 128, 525 

\bigotimes, 128, 525 

\bigskip, 33, 525 

\bigskipamount, 525 

\bigsqcup, 128, 525 

\bigtriangledown, 126, 525 

\bigtriangleup, 126, 525 

\biguplus, 128, 525 

\bigvee, 128, 525 

\bigwedge, 128, 525 

binary operator symbols, 125 

\binom (A M S), 264, 525 

binomial coefficient, 137, 264 

blank, 11, 22 

after command, 19, 22, 27, 186 

at beginning of line, 22, 29 

forced, 22 

multiple, 11, 22 

protected, 22, 29 

rubber, 30 

suppression of, 22 

blank line for new paragraph, 22 

.blg file, 397 

Blue Sky Research, 382, 506 

bm package, 144, 394 

\bm, 144, 394 

Bmatrix environment (A M S), 266, 

517 

bmatrix environment (A M S), 266, 

517 

\bmod, 129, 268, 526 

body, 47 

bold face in formulas, 143, 394 

\boldmath, 143, 144, 150, 151, 373, 

394, 433, 526 

\boldsymbol (A M S), 144, 258, 526 

book class, 37, 41, 55, 57, 113, 130, 

173, 183, 187, 191, 218, 

224, 225, 277, 391, 459 

bookman package, 234 

Bookman, PostScript font, 234, 504 

\boolean (ifthen), 194, 195, 360, 

451, 453 

\bot, 127, 526 

\botfigrule, 173, 526 

bottom margin, 48 

\bottombuttons (pdfscreen), 342 

\bottomfraction, 172, 526

ottomnumber, 171, 526 

Botway, L., 605 

\bowtie, 126, 526 

box, 85–94 

calling, 87 

framed, 86 

lowering, 87 

LR, 85, 87 

nested, 92 

paragraph, 85 

parbox, 88 

positioning 

horizontal, 86 

vertical, 88, 89, 92 

raising, 87 

rule, 85, 91 

saving, 87 

style parameters, 93 

T E X primitive, 85 

vertical, 88 

\Box, 126, 127, 526 

boxed formula, 142, 270 

\boxed (A M S), 142, 270, 526 

bp (big point), 21, 155, 158 

Braams, Johannes, 252, 464 

\brace (T E X), 192 

\brack (T E X), 192 

bracket symbol, see symbols 

Breton, 252 

\breve, 129, 263, 526 

\Breve (A M S), 263, 526 

British spelling, 195 

bsr-interpolated.map, 506 

bsr.map, 506 

.bst file, 220, 221, 310, 321, 397 

btxdoc.tex, 321 

btxhak.tex, 321 

Bulgarian, 252 

\bullet, 126, 526 

\c (¸ accent), 24, 526 

calc package, 394 

calligraphic letters, 125, 284 

\cap, 126, 527 

\caption, 60, 108, 169, 173, 174, 

177, 178, 182, 213, 427, 

436, 527 

\captionslang (non-standard), 255, 

460, 527 

INDEX 613 

Carlisle, David, 154, 159, 193, 215, 

330, 480 

cases environment (A M S), 277, 

517 

Catalan, 252, 462 

catalogue.html, 13 

\cbinput (chapterbib), 224 

cbunit environment (chapterbib), 

224 

\cc, 353, 527 

cc (cicero), 21 

\ccname, 353, 460, 527 

CD environment (A M S), 282 

\cdot, 126, 527 

\cdots, 123, 134, 137, 263, 527 

center environment, 67, 69, 79, 98, 

176, 289, 516 

centered text, 67 

centering and indenting, 67–9 

\centering, 67, 69, 79, 178, 458, 

527 

\centerline (T E X), 67, 176, 408, 527 

\centerslidefalse (seminar), 334 

\centerslidetrue (seminar), 334 

centertags option, 272, 279 

.cfg file, 253, 384, 385, 397, 472 

\cfoot (fancyhdr), 44 

\cfrac (A M S), 265, 527 

\changes (doc), 470 

chapter counter, 57, 181, 183 

chapter opening page, 39 

\chapter, 44, 55, 58, 449, 527 

\chapter*, 55, 527 

chapterbib package, 224 

\chaptername, 460, 528 

character spacing, see spacing 

character specification, 66 

decimal, 66 

hexadecimal, 66 

octal, 66 

\CharacterTable (doc), 471 

\chead (fancyhdr), 44 

\check, 129, 263, 528 

\Check (A M S), 263, 528 

\CheckCommand, 430, 445, 528 

\CheckCommand*, 445 

\CheckSum (doc), 471 

chemical formulas, 143 

Chen, Pehong, 228

614 INDEX 

\chi, 125, 528 

chicago package, 220 

Chinese, 6 

\choose (T E X), 137, 146, 147, 189, 

192 

\circ, 126, 528 

\circle, 416 

\circle (eepic), 305 

\circle (picture), 295, 299, 300, 

431, 528 

\circle* (eepic), 305 

\circle* (picture), 295, 528 

\cite, 219, 310, 312, 314, 395, 429, 

430, 528 

\citeauthor (natbib), 222 

\citep (natbib), 219, 221, 310, 528 

\Citet (natbib), 222 

\citet (natbib), 219, 221, 310, 528 

\citeyear (natbib), 222 

\ClassError, 446, 528 

classes.dtx, 463 

\ClassInfo, 447, 529 

\ClassWarning, 446, 529 

\ClassWarningNoLine, 446, 529 

Clausen, Joern, 25 

\cleardoublepage, 34, 171, 420, 

529 

\clearpage, 34, 84, 171, 209, 420, 

428, 529 

forbidden, 84 

\cline, 97, 102, 106, 529 

.clo file, 384, 385, 397 

\closing, 353, 529 

.cls file, 384, 385, 398, 484 

\clubsuit, 127, 529 

cm (centimeter), 21 

cm-super fonts, 506 

\CodelineIndex (doc), 469 

\CodelineNumbered (doc), 470 

Codepages, IBM, 26, 462 

color package, 153, 166, 324, 325, 

335, 338, 345, 439 

\color (color), 167, 335, 529 

color.cfg, 166 

\colorbox (color), 167, 335, 529 

colors, 166–8 

column breaking, 34 

column separation 

two-column pages, 40 

column separation symbol, 97, 139 

\columnsep, 40, 51, 52, 395, 530 

\columnseprule, 40, 52, 395, 530 

\columnwidth, 51, 173 


*-form, 18 

distinguishing from text, 11 

followed by blank, 19, 186 

invisible, 201 

levels, 438 

multi-character, 17 

name, 17 

single character, 17 

syntax, 18 

two-character, 17 

command characters, printing of, 

23 

command, user-defined, 185–92 

examples 

T E X as L AT E X, 189 

equation numbering, 190 

footnotes, 190 

framing text, 189 

followed by blank, 186 

for math and text, 186 

general comments, 200 

nested, 204 

order of, 203 

scope of, 202 

global, 202 

local, 202 

storing, 200 

unwanted spacing, 201 

with arguments, 187, 188 

calling, 188 

with optional argument, 189 

without arguments, 185, 187 

calling, 186 

comment, 118 

character, 118 

comment environment (verbatim), 

111, 118, 396 

commenting out, 118 

compatibility mode, 392, 398, 418, 

484 

Comprehensive T E X Archive 

Network, see CTAN 

compressed graphics files, 166

Computer Modern fonts, 8, 64, 283, 

369, 488–97, 499 

PostScript version, 235, 505 

conditional text, 193 

config.cms, 235 

config.ps, 233, 235 

configuration file, local 

color, 166 

graphics, 164 

hyperref, 246 

hyphen, 253, 385 

L AT E X installation, 384 

ltxdoc, 397, 447, 472 

pdfscreen, 343 

seminar, 339, 340 

configuring L AT E X, 384 

\cong, 126, 530 

constant, see formula 

\contentsline, 530 

\contentsname, 459, 530 

continued fraction, 146, 149, 265 

continuing dots, 123, 263 

\coprod, 128, 530 

copyright sign, 23, 503 

\copyright, 23, 503, 530 

\copyrightspace, 391 

\cornersize (fancybox), 94 

\cos, 128, 530 

\cosh, 128, 530 

\cot, 128, 530 

\coth, 128, 530 

counter value, 183 

multiple printing, 183 

printing, 183 

counter, L AT E X, see L AT E X counter 

counter, user-defined, 182 

auto. reset, 182 

changing, 182 

creation, 182 

incrementing, 182 

setting, 182 

Courier, PostScript font, 234, 379, 

504 

Croatian, 252 

cross-reference, 213–16 

external document, 215 

to bibliography, 219 

to counters, 182, 213 

INDEX 615 

to equations, 131, 139, 213, 

278 

to figures, 177, 213 

to lists, 213 

to pages, 213 

to sections, 56, 213 

to tables, 177, 213 

to text, 213 

to theorems, 213 

variable, 215 

\csc, 128, 530 

CTAN, 381, 389 

\cup, 126, 530 

\CurrentOption, 359, 443, 530 

cyracc.def, 284 

Cyrillic fonts, 6, 283, 379, 497 

transliteration, 284 

Czech, 252 

\d ( . accent), 24, 531 

\dag, 23, 531 

dagger, 23 

\dagger, 126, 531 

Dahlgren, Mats, 178 

Daly, Patrick W., 220, 321 

Danish, 252 

DANTE, German-speaking T E X 

Users, 251 

dash, 23 

as hyphen, 23 

as minus sign, 23 

\dashbox (picture), 291, 292, 300, 

531 

dashjoin environment (epic), 302, 

304 

\dashline (epic), 302, 303 

\dashlinestretch (epic), 305 

\dashv, 126, 127, 531 

database, see bibliographic 

database 

\date, 52, 54, 69, 353, 358, 363, 531 

\datelang (non-standard), 255, 461, 

531 

.dat file, 253 

\day (T E X), 27, 460 

\dbinom (A M S), 265, 531 

\dblfigrule, 173, 531 

\dblfloatpagefraction, 172, 531 

\dblfloatsep, 172, 532

616 INDEX 

\dbltextfloatsep, 172, 532 

\dbltopfraction, 172, 532 

dbltopnumber, 172, 532 

DC fonts, see EC fonts 

dcolumn package, 107, 394 

dd (didôt), 21 

\ddag, 23, 532 

\ddagger, 126, 532 

\ddddot (A M S), 263, 532 

\dddot (A M S), 263, 532 

\ddot, 129, 263, 532 

\Ddot (A M S), 263, 532 

\ddots, 123, 532 

de Boer, Ingo H., 15 

DEC Multinational coding, 462 

declaration, 20 

local, 20 

scope of, 20 

\DeclareBoldMathCommand (bm), 

394 

\DeclareErrorFont, 376, 423, 424, 

532 

\DeclareFixedFont, 66, 361, 373, 

533 

\DeclareFontEncoding, 376, 378, 

379, 533 

\DeclareFontEncodingDefaults, 

376, 533 

\DeclareFontFamily, 376, 378, 

423, 533 

\DeclareFontShape, 377, 378, 423, 

533 

\DeclareFontSubstitution, 376, 

533 

\DeclareGraphicsExtensions 

(graphics), 165, 533 

\DeclareGraphicsRule 

(graphics), 165, 533 

\DeclareInputMath (inputenc), 

461 

\DeclareInputText (inputenc), 

461 

\DeclareMathAccent, 375, 423, 534 

\DeclareMathAlphabet, 191, 374, 

375, 416, 423, 534 

\DeclareMathDelimiter, 375, 534 

\DeclareMathOperator (A M S), 

268, 282, 534 

\DeclareMathRadical, 375, 534 

\DeclareMathSizes, 375, 534 

\DeclareMathSymbol, 375, 534 

\DeclareMathVersion, 374, 422, 

423, 535 

\DeclareOldFontCommand, 373, 535 

\DeclareOption, 359, 421, 442, 535 

\DeclareOption*, 359, 443, 535 

\declarepostamble (DocStrip), 465 

\declarepreamble (DocStrip), 465 

\DeclareRobustCommand, 445, 535 

\DeclareRobustCommand*, 445, 535 

\DeclareSymbolFont, 374, 423, 

424, 535 

\DeclareSymbolFontAlphabet, 

375, 424, 536 

\DeclareTextAccent, 379, 536 

\DeclareTextAccentDefault, 380, 

536 

\DeclareTextCommand, 379, 417, 

536 

\DeclareTextCommandDefault, 

380, 536 

\DeclareTextComposite, 379, 422, 

536 

\DeclareTextCompositeCommand, 

379, 536 

\DeclareTextFontCommand, 373, 

536 

\DeclareTextSymbol, 379, 536 

\DeclareTextSymbolDefault, 380, 

537 

\def (T E X), 439, 445, 450 

.def file, 153, 253, 376, 378, 379, 

385, 392, 398 

\definecolor (color), 167, 345, 

537 

\deg, 128, 537 

dehyphn.tex, 256 

dehypht.tex, 256 

delarray package, 108, 394 

\DeleteShortVerb (shortvrb), 

111, 393, 468, 537 

\Delta, 125, 537 

\delta, 125, 537 

\depth, 86, 90, 537 

\DescribeEnvironment (doc), 468 

\DescribeMacro (doc), 468 

description environment, 69, 70, 

74, 79, 230, 419, 420, 517

design size, 501 

\det, 128, 129, 280, 537 

determinant, 133–6 

\dfrac (A M S), 264, 537 

\DH, 503, 537 

\dh, 503, 538 

diacritical marks, see accents 

\Diamond, 126, 127, 538 

\diamond, 126, 538 

\diamondsuit, 127, 538 

\dim, 128, 538 

\DisableCrossrefs (doc), 470 

\discretionary, 35, 538 

\displaybreak (A M S), 278, 538 

displayed formula, 119 

displayed text 

centered, 67 

indented, 67 

left or right justified, 67 

nested, 68 

displaymath environment, 120, 

517 

\displaystyle, 142, 146, 150, 264, 

538 

\div, 126, 538 

\DJ, 503, 538 

\dj, 503, 538 

doc package, 392, 399, 441, 467–73 

doc.dtx, 468 

DocBook DTD, 479 

\DocInput, 467 

DocStrip, 321, 398, 464–7 

docstrip.dtx, 467 

docstrip.tex, 385, 391, 465 

document class, 37 

options, 120 

document environment, 12, 517 

Document Type Definition, 479 

document, major subdivisions, 

52–7 

documentation 

browser, 383 

integrated, 463, 467 

with doc, 467–73 

documentation of packages, 13 

\documentclass, 12, 37, 41, 398, 

416, 422, 443, 538 

\documentstyle (L AT E X 2.09), 418, 

422, 483 

INDEX 617 

dollar sign, 23, 27 

\DoNotIndex (doc), 470 

\dot, 129, 263, 539 

\Dot (A M S), 263, 539 

\doteq, 126, 539 

\dotfill, 30, 84, 86, 135, 539 

\dots, 124, 539 

\dots (A M S), 263, 539 

\dotsb (A M S), 264, 539 

\dotsc (A M S), 264, 539 

\dotsi (A M S), 264, 539 

\dotsm (A M S), 264, 539 

dottedjoin environment (epic), 

302, 304 

\dottedline (epic), 302, 303 

double dagger, 23 

\doublebox (fancybox), 94, 539 

\doublerulesep, 98, 539 

\Downarrow, 127, 132, 539 

\downarrow, 127, 132, 539 

Downes, Michael, 257 

draft option, 39 

Drakos, Nikos, 476 

drawjoin environment (epic), 302, 

304 

\drawline (epic), 161, 302, 303 

\drawlinestretch (epic), 305 

driver, printer, 15, 153, 389, 391, 

398, 453, 488, 497 

.drv file, 463, 464 

DTD, 479 

DocBook, 479 

TEI, 479, 480 

.dtx file, 13, 384, 389, 398, 463, 465 

Duchier, Denys, 464 

Dutch, 252 

.dvi file, 14, 27, 153, 161, 236, 305, 

393, 403, 477 

dvipdf, graphics option, 154 

dvipdfm, PDF driver, 162, 237, 305, 

346, 387, 389 

dvipdfm, graphics option, 154, 237 

dvips, graphics option, 154 

dvips, printer driver, 15, 162, 167, 

168, 231, 236, 305, 332, 

333, 335, 337, 387, 389, 

476 

dvipsone, graphics option, 154 

dviwin, graphics option, 154

618 INDEX 

e.tex, 394, 413 

EC fonts, 393, 499–503 

calling, 501 

character assignments, 500 

special character commands, 

503 

with NFSS, 379 

\edef (T E X), 450 

eepic package, 305–6 

eepicemu package, 305 

Eijkhout, Victor, 450, 605 

electron volt, symbol, 30, 192 

electronic documents, 340, 475 

electronic projection, 323 

electronic publishing, 236 

\ell, 127, 539 

\ellipse (eepic), 306 

ellipsis, 123 

\else (T E X), 451 

em dash, 23 

\em, 20, 62, 65, 433, 485, 539 

em, 21 

emacs, 382 

email addresses, 112 

\emblema (pdfscreen), 341 

\emph, 62, 65, 372, 540 

empty page style, 42, 327, 335, 352, 

452 

\emptyset, 127, 540 

emT E X for DOS, 382 

emtex, graphics option, 154 

en dash, 23 

\EnableCrossrefs (doc), 470 

encapsulated PostScript, see 

PostScript 

\encl, 353, 540 

\enclname, 353, 460, 540 

encoding, see font 

encoding commands in NFSS, 379 

encoding scheme 

OML, 491, 495 

OMS, 491, 495 

OMX, 491, 496 

OT1, 490, 492, 506 

OT2, 283, 497 

T1, 500, 501, 503, 506 

TS1, 502, 506 

encoding, font attribute, 368 

\encodingdefault, 372, 501 

end of line, as a blank, 22 

\end, 19, 64, 416, 540 

\endbatchfile (DocStrip), 465 

\end{document}, 12, 517 

\endfirsthead (longtable), 108, 

540 

\endfoot (longtable), 108, 540 

\endhead (longtable), 108, 540 

\endinput (T E X), 451 

\endlastfoot (longtable), 108, 

540 

\endpostamble (DocStrip), 465 

\endpreamble (DocStrip), 465 

English, 252 

english.ldf, 255 

\enlargethispage, 34, 419, 540 

\enlargethispage*, 34, 540 

\ensuremath, 186, 187, 188, 418, 

450, 541 

enumerate environment, 69, 70, 

71, 72, 74, 79, 213, 394, 

419, 420, 517 

enumerate package, 74, 394 

enumn counter, 73, 181 

environment, 19, 516 

command name as, 20, 517 

global, 20 

math, 119 

nameless, 20, 62, 63 

environment environment (doc), 

469 

environment, user-defined, 195–200 

general comments, 200 

scope of, 202 

global, 202 

local, 202 


with arguments, 198, 199 

calling, 198 

with optional argument, 199 

without arguments, 196, 197 

calling, 196 

epic package, 302–5 

.eps file, 155 

epsf package, 159 

\epsf (epsfig), 159 

epsfig package, 159 

\epsfig (epsfig), 159 

\epsfxsize (epsfig), 159

\epsfysize (epsfig), 159 

\epsilon, 125, 541 

eqnarray environment, 120, 138, 

142, 213, 270, 420, 517 

eqnarray* environment, 120, 138, 

142, 270, 518 

\eqref (A M S), 278, 541 

\equal (ifthen), 194, 363 

equation, see formula 

equation counter, 181, 277 

equation environment, 120, 124, 

213, 271, 276, 518 

equation number, 120, 271 

changing hierarchy, 277 

right or left, 39, 120, 131, 271 

subnumbered, 277 

user-defined, 191 

vertically centered, 142, 272 

equation* environment (A M S), 

272 

equations, set of, 133–6 

\equiv, 126, 541 

error messages 

basic structure for T E X, 403 

continue program, 402 

emergency stop, 413 

error indicator, 402 

error line, 402, 405 

from L AT E X, 404–7 

from T E X, 401–4 

from T E X macros, 407 

going to deeper levels, 404 

list 

hard-to-find errors, 435 

L AT E X font errors, 422–4 

L AT E X font warnings, 433 

L AT E X general errors, 415–20 

L AT E X general warnings, 

429–31 

L AT E X package errors, 421–2 

L AT E X package warnings, 432 

T E X, 424–9 

T E X warnings, 433–5 

mathematical, 413, 414 

multi-file, 414 

propagation, 409–10 

stop program 

with I\stop, 403, 410 

with editor, 403 

INDEX 619 

with X, 403, 410 

unknown file name, 412, 417 

user response 

call editor, 403 

continue program, 402 

correction, 403 

deletion, 403 

help, 403 

recommendation, 407 

errorcontextlines counter, 404 

eso-pic package, 346 

Esperanto, 252, 462 

Esser, Thomas, 381 

Estonian, 252 

\eta, 125, 541 

eucal package, 284–5 

eufrak package, 285 

\EUR (europs), 25, 541 

euro symbol, 24–5, 387, 462, 503 

PostScript fonts, 25, 387 

\euro (eurosym), 25, 28, 541 

\EURofc (europs), 25, 541 

eurofont.exe, 387 

European Commission, 25 

European paper sizes, 12 

europs package, 25, 387 

eurosans package, 25, 387 

eurosym package, 25, 28 

\evensidemargin, 48, 49, 361, 453, 

541 

ex, 21 

\ExecuteOptions, 165, 443, 453, 

541 

executivepaper option, 38 

exercises 

Chapter 2, 27, 28 

Chapter 3, 40, 45, 47–9, 54, 57, 

58, 60 

Chapter 4, 68, 69, 72, 74, 78, 

84, 93, 103, 105, 110, 111, 

116, 118 

Chapter 5, 121, 124, 129, 130, 

132, 135–7, 141, 149, 150 

Chapter 6, 157, 159, 168 

Chapter 8, 183, 187, 192, 200 

Chapter 9, 208, 212, 218, 219 

Chapter 13, 290, 293, 295, 

297, 301 

\exists, 127, 541

620 INDEX 

\exp, 128, 541 

\expandafter (T E X), 451 

explicit names, changing, 459 

exponents, 121 

exscale package, 392 

Extended Computer fonts, see EC 

fonts 

\externaldocument (xr), 215, 243 

\extracolsep, 96, 110, 455, 541 

\extrarowheight (array), 107 

\extraslang (non-standard), 255 

\extraslideheight (seminar), 336 

\f@baselineskip, 376 

\f@encoding, 376 

\f@family, 376 

\f@series, 376 

\f@shape, 376 

\f@size, 376 

Fairbairns, Robin, 47 

falign environment (A M S), 270, 

275, 518 

family, font attribute, 64, 368 

\familydefault, 372 

fancy page style, 43 

fancybox package, 94–5, 334, 338, 

345 

\fancyfoot (fancyhdr), 44 

fancyhdr package, 43–5, 249, 335, 

454 

\fancyhead (fancyhdr), 44 

\fancyhf (fancyhdr), 44 

\fancypage (fancybox), 95, 542 

\fancypagestyle (fancyhdr), 45 

\fancyput (fancybox), 338 

\fbox, 86, 92–4, 115, 142, 334, 542 

\fbox (picture), 298 

\fboxrule, 93, 94, 164, 334, 542 

\fboxsep, 93, 94, 164, 298, 334, 542 

\fcolorbox (color), 167, 335, 542 

.fdd file, 398 

.fd file, 233, 234, 283, 376, 378, 

385, 398, 424, 490 

\fi (T E X), 451 

figure, see float 

figure counter, 181 

figure caption, see float, caption 

figure environment, 60, 108, 169, 

173, 182, 213, 417, 419, 

430, 518 

figure* environment, 169, 518 

\figurename, 459, 542 

figures, list of, 59 

file 

including, 448 

inputting safely, 447 

listing, 447 

transcript, 14, 50, 208, 211, 

399, 401, 447 

file transfer protocol, see FTP 

file types 

.aux, 209, 210, 214, 396, 397, 

418, 427, 430, 436 

.bbl, 219, 224, 310, 397 

.bib, 309, 397 

.blg, 397 

.bst, 220, 221, 310, 321, 397 

.cfg, 253, 384, 385, 397, 472 

.clo, 384, 385, 397 

.cls, 384, 385, 398, 484 

.dat, 253 

.def, 153, 253, 376, 378, 379, 

385, 392, 398 

.drv, 463, 464 

.dtx, 13, 384, 389, 398, 463, 

465 

.dvi, 14, 27, 153, 161, 236, 

305, 393, 403, 477 

.eps, 155 

.fdd, 398 

.fd, 233, 234, 283, 376, 378, 

385, 398, 424, 490 

.fmt, 398 

.gif, 476 

.glo, 230, 398, 470 

.gls, 399, 470 

.gz, 166 

.html, 476 

.idv, 478 

.idx, 226–8, 399, 470 

.ilg, 399 

.ind, 228, 399 

.ins, 389, 399, 463 

.ist, 385, 399, 470 

.jpeg, 162 

.jpg, 162 

.ldf, 253 

.lof, 59, 179, 399, 436

.log, 14, 246, 399, 401, 403, 

414, 429 

.lot, 59, 399, 436 

.map, 234 

.mbs, 321, 322 

.mf, 389, 400, 488, 498, 506 

.pbm, 476 

.pdf, 162, 237 

.pfa, 389 

.pfb, 234, 389, 506 

.pk, 233, 234, 236, 388, 400, 

488, 498, 504 

.png, 162 

.ppm, 478 

.ps, 232, 236 

.sty, 9, 12, 41, 384, 385, 463, 

484 

.tcx, 481 

.tex, 14, 17, 208, 209, 391, 

396, 400, 403, 436 

.tfm, 234, 400, 487, 498, 503, 

506 

.tiff, 162 

.tif, 162 

.toc, 59, 400, 436 

.txt, 384 

.vf, 234, 236, 400, 488, 504 

.zip, 166 

\file (DocStrip), 465 

filecontents environment, 432, 

448, 518 

filecontents* environment, 449, 

518 

\filedate (doc), 471 

fileerr.dtx, 394 

\fileinfo (doc), 471 

\filename (doc), 471 

\fileversion (doc), 471 

\fill, 22, 30, 32, 96, 110, 184, 455, 

542 

filler, with dots, rules, spacing, 30 

final option, 39 

\Finale (doc), 469, 473 

Finnish, 252 

\firsthline (array), 107 

firstpage page style, 362 

fixed length, 21 

flafter package, 171, 392 

\flat, 127, 542 

INDEX 621 

fleqn option, 39, 120, 124, 149, 

271, 280 

float, 110, 169–77 

caption, 173 

width, 174 

causing buffer overflow, 428 

double columns, 169 

examples, 109, 174–7 

figure title, 173 

head and foot text, 109 

numbering, 173 

placement, 109, 169–71 

style parameter, 171 

table title, 173 

float package, 179–80 

floatfig package, see floatflt 

floatflt package, 178–9 

floatingfigure environment, 178 

floatingtable environment, 179 

\floatname (float), 179 

\floatpagefraction, 172, 542 

\floatplacement (float), 180 

\floatsep, 172, 542 

\floatstyle (float), 180 

\flushbottom, 48, 542 

flushleft environment, 67, 79, 

518 

flushright environment, 67, 79, 

518 

\flushright, 117 

Flynn, Peter, 605 

.fmt file, 398 

\fnsymbol, 114, 183, 190, 543 

font 

slides class, 324 

amsfonts, 283–5 

attribute, 64–5, 368–72 

defaults, 372 

encoding, 368 

family, 64, 368 

internals, 376 

series, 64, 368 

shape, 64, 368 

size, 64, 369 

bitmaps, 488 

character assignment 

cmcsc10 (small caps), 492 

cmex10 (var. symbols), 496 

cmmi10 (math text), 495

622 INDEX 

cmr10 (standard), 492 

cmsy10 (math symbols), 495 

cmti10 (text italic), 493 

cmtt10 (typewriter), 493 

EC fonts, 500 

TC fonts, 502 

wncyr10 (Cyrillic), 497 

classification, 489 

commands, 65, 370 

Computer Modern, 8, 64, 283, 

369, 488–97, 499 

Cyrillic, 283, 379, 497 

decorative, 494 

design size, 501 

EC, 393, 499–503 

calling, 501 

with NFSS, 379 

encoding, 487 

euro, PostScript, 25, 387 

eurosym, 25 

extended, 498–503 

Extended Computer, see EC 

families, 489 

file names, 490 

L AT E X, 65, 494 

L AT E X 2.09 commands, 485 

line spacing for additional, 562 

loading, 66 

logos, 494 

magnification, 66 

math, 491 

symbols, 374 

math alphabets, 133, 258, 373 

NFSS, 367–72, 376–9 

pixel, 8 

PostScript, 233–6, 378, 503–5 

standard, 65, 367 

standard line spacing, 63 

TC, 24, 393, 501 

text, 490 

TrueType, 238 

type 1, 8, 25, 231, 238, 387, 

498, 503 

type 3, 8, 234, 238 

virtual, 488, 503 

font definition files, 236, 376, 398 

font environment, 64 

font size, 62 

changing, 62, 367 

class option, 37 

declarations, 63 

in formulas, 146 

standard line spacing, 63 

font size commands (L AT E X 2.09), 

485 

font style commands (L AT E X 2.09), 

485 

fontenc package, 378–9, 392, 501 

\fontencoding, 368, 370, 543 

\fontfamily, 368, 369, 370, 543 

fontmath.cfg, 385 

fontmath.ltx, 385 

\fontseries, 368, 369, 370, 543 

\fontshape, 368, 369, 370, 372, 543 

\fontsize, 368, 369, 370, 543 

fontsmpl package, 394 

fonttext.cfg, 385 

fonttext.ltx, 385 

foot, 42, 44, 47, 50, 85, 249, 335, 

347, 362, 435, 453 

customizing, 43, 454 

\footheight (L AT E X 2.09), 604 

footnote, 112–16 

change markers, 114 

forbidden mode, 114 

in minipages, 113, 115 

in tables, 116 

marker without text, 114 

non-standard, 113 

standard, 113 

forbidden, 113 

standard marker, 113 


text external to forbidden 

mode, 114 

footnote counter, 113, 181, 190 

\footnote, 112, 113, 114, 116, 543 

\footnotemark, 114, 115, 543 

\footnoterule, 118, 543 

\footnotesep, 117, 544 

\footnotesize, 63, 371, 428, 544 

\footnotetext, 110, 114, 115, 544 

\footrulewidth (fancyhdr), 44 

\footskip, 48, 50, 453, 544 

\forall, 127, 544 

\foreignlanguage (babel), 254, 

544 

format file, 6, 384, 398, 440

formula, 119 

arrays, 133–6, 266 

blanks within, 120 

bold face in, 143, 258, 394 

chemical, 143 

constants, 120 

continued fraction, 146, 149, 

265 

continuing dots, 123, 263 

displayed, 119 

centered, 39, 120, 280 

flush left, 39, 120, 280 

indented, 40 

ellipsis, 123 

exponent, 121 

extra T E X commands, 137 

fine-tuning, 145–8 

bracket sizing, 148 

font size, 146–7 

horizontal spacing, 145, 269 

font size in, 146 

detailed, 147 

simplified, 146 

fraction, 122 

framed, 142 

function names, 128, 267 

Greek letters, 125 


in text, 119 

index, 121 

integral, 123 

limits 

for 2 sizes, 128 

multiline, 261 

positioning, 123, 128 

with functions, 129 

lowering in, 121 

math style parameters, 148 

matrix, 133–6, 266 

multiline, 120, 138–41 

bracket sizing, 140, 148 

breaking, 138 

centered, 138 

column separation, 138 

left part, 140 

numbered, 139 

page breaks, 278 

with A M S, 270–7 

numbered, 120 

INDEX 623 

overbrace, 136 

overlining, 136 

raising in, 121 

root, 122, 269 

side-by-side, 142 

subnumbered, 277 

summation, 123 

symbols in, see symbols 

text within, 133, 259 

underbrace, 136 

underlining, 136 

variables, 120 

fpT E X for Windows, 381, 385 

\frac, 122, 146, 189, 264, 544 

fractions, 122 

with A M S, 264 

fragile commands, 440, 444 

\frame, 299, 300, 544 

\framebox, 86, 90, 93, 94, 106, 189, 

272, 300, 544 

\framebox (picture), 291, 292, 298, 

544 

framed formulas, see boxed 

formula 

framed table, 100 

framed text, 86 

French, 252 

french package, 251 

\frenchspacing, 29, 545 

frhyphen.tex, 256 

\from (DocStrip), 465 

\fromname, 363 

\frontmatter, 57, 210, 545 

\frown, 126, 545 

ftnright package, 394 

FTP, 475 

fullpage package, 452, 465 

function names, 128 

defining, 268 


\fussy, 36, 545 

Galician, 252, 462 

\Gamma, 125, 545 

\gamma, 125, 545 

gather environment (A M S), 270, 

273, 519 

gathered environment (A M S), 

276, 519

624 INDEX 

Gaulle, Bernard, 251 

\gcd, 128, 129, 280, 545 

\gdef (T E X), 450 

\ge, 126, 545 

\generate (DocStrip), 465 

\genfrac (A M S), 265, 545 

geometry package, 49–51, 452 

\geq, 126, 545 

German, 252 

spelling reform, 35 

german package, 251, 252, 360 

\GetFileInfo (doc), 441, 471, 472 

\gets, 127, 545 

\gg, 126, 545 

gglo.ist, 399, 470 

Ghostscript, 236, 476 

GhostView, 164, 232, 239 

GIF images, 476, 478 

.gif file, 476 

gind.ist, 399, 470 

Girou, Denis, 330 

.glo file, 230, 398, 470 

glossary, 230 

\glossary, 230, 398, 420, 428, 545 

\glossaryentry, 230, 398, 545 

\glossaryname, 460 

.gls file, 399, 470 

Goossens, M., 605 

graphics 

importing files, 154 

rotation, 156 

scaling, 156 

graphics package, 153, 155–7 

graphics.cfg, 164 

\graphicspath (graphics), 165 

graphicx package, 157–9 

graphpap package, 301, 392 

\graphpaper, 392 

\graphpaper (graphpap), 301, 546 

\grave, 129, 263, 546 

\Grave (A M S), 263, 546 

Greek, 252 

Greek letters, 125 

grfguide.tex, 154 

\grid (epic), 302, 303 

\guillemotleft, 503, 546 

\guillemotright, 503, 546 

\guilsinglleft, 503, 546 

\guilsinglright, 503, 546 

Guntermann, Klaus, 343 

Gurari, Eitan M., 477 

.gz file, 166 

gzip program, 166 

h.tex, 394, 413 

\H (˝ accent), 24, 546 

Haralambous, Y., 605 

Hargreaves, K. A., 605 

harvard package, 220 

\hat, 129, 546 

\Hat (A M S), 263, 546 

\hbar, 127, 546 

\hbox (T E X), 439 

\hdotsfor (A M S), 267 

head, 42, 44, 47, 50, 85, 335, 362, 

435, 453 


right page, 43 

\headheight, 48, 50, 361, 363, 453, 

455, 546 

heading, 43 

left page, 43 

setting values, 47 

headings page style, 42, 44, 55, 58, 

327, 335, 352, 363, 452 

\headrulewidth (fancyhdr), 44 

\headsep, 48, 50, 361, 363, 453, 

455, 546 

\headtoname, 359, 360, 363, 460, 

547 

\heartsuit, 127, 547 

Hebrew, 252 

\height, 86, 90, 547 

helvet package, 234 

Helvetica, PostScript font, 234, 324, 

379, 504 

here package, 178 

\hfill, 30, 32, 77, 84, 88, 89, 117, 

142, 149, 176, 362, 547 

hhline package, 395 

\hline, 97, 100, 102, 103, 106, 547 

\hoffset (T E X), 547, 604 

\hom, 128, 547 

\hookleftarrow, 127, 547 

\hookrightarrow, 127, 547 

\hpagecolor (background), 345 

\href (hyperref), 247 

\hrulefill, 30, 84, 91, 547

\hspace, 29, 30, 82, 96, 105, 106, 

110, 332, 547 

\hspace*, 29, 82, 547 

\hss (T E X), 408 

HTML, 3–5, 9, 476–8 

html package, 476 

.html file, 476 

\Huge, 63, 371, 548 

\huge, 63, 371, 548 

Hungarian, 252 

\hyperbaseurl (hyperref), 247 

\hyperdef (hyperref), 247 

\hyperimage (hyperref), 247 

\hyperlink (hyperref), 161, 247, 

349, 548 

hyperref package, 112, 161, 220, 

239–49, 330, 339, 340, 

344 

\hyperref (hyperref), 247 

hyperref.cfg, 246 

\hypersetup (hyperref), 241, 246, 

344, 548 

\hypertarget (hyperref), 247, 349, 

548 

hyphen, 23 

hyphen.cfg, 252, 256, 385 

babel, 253 

hyphen.ltx, 385 

hyphen.tex, 385 

hyphenation, 34, 427 

exception list, 35 

in other languages, 499 

manual, 35 

multilingual, 36, 251, 256 

other languages, 389 

overfull \hbox, 434 

turning off, 36 

with accents, 426 

\hyphenation, 36, 426, 427, 548 

\i, 24, 548 

Icelandic, 252 

\idotsint (A M S), 260, 548 

.idv file, 478 

idx.tex, 227, 385, 391 

.idx file, 226–8, 399, 470 

\if (T E X), 451 

\ifcase (T E X), 451, 461 

\iff, 127, 548 

INDEX 625 

\iffalse (T E X), 468 

\IfFileExists, 447, 548 

\iflanguage (babel), 254, 548 

ifthen package, 193–5, 359, 392, 

439, 452 

\ifthenelse (ifthen), 193, 195, 

360, 363, 439, 451, 453 

\iiiint (A M S), 260, 551 

\iiint (A M S), 260, 551 

\iint (A M S), 260, 551 

.ilg file, 399 

\Im, 127, 549 

\imageButton (pdfscreen), 342 

\imath, 127, 129, 549 

importing graphics files, 154–66 

compressed, 166 

configuring, 164 

problems, 161 

\in, 126, 549 

in (inch), 21 

\include, 182, 209, 210, 212, 215, 

224, 396, 397, 402, 411, 

412, 414, 415, 417, 549 

\includegraphics, 156, 159–61, 

163, 164, 169, 249, 332, 

346 

\includegraphics (graphics), 

155, 549 

\includegraphics (graphicx), 

157, 550 

\includegraphics* (graphics), 

156, 549 

\includegraphics* (graphicx), 

159 

\includeonly, 209, 210, 211, 215, 

328, 416, 550 

including one file in another, 448 

\indent, 46, 550 

indentfirst package, 46, 395 

index of keywords, 225–8 

\index, 225, 226, 228, 391, 392, 

399, 420, 428, 550 

\indexentry, 226, 399, 550 

\indexname, 225, 459, 550 

\indexspace, 225, 550 

.ind file, 228, 399 

indices, 121 

\inf, 128, 129, 280, 550 

\infty, 127, 550

626 INDEX 

initex, 6, 251, 253, 256, 384, 398 

input coding, 26, 461 

\input, 192, 201, 207, 208, 209, 

396, 402, 411, 412, 414, 

415, 442, 447, 551 

nested, 208 

inputenc package, 26, 392, 461–2 

\InputIfFileExists, 447, 551 

.ins file, 389, 399, 463 

install.txt, 384 

instr-l.tex, 257 

\int, 123, 128, 551 

integral sign, 123 

Internet, 9, 316, 390, 475, 590 

giving addresses, 112 

\intertext (A M S), 259, 550 

\intextsep, 172, 551 

intlimits option, 260, 279 

Inuit, 462 

invisible commands, 201 

\invisible (slides), 326, 551 

\iota, 125, 551 

Irish Gaelic, 252 

ISO math standards, 144 

ISO-Latin codings, 462 

\isodd (ifthen), 193 

.ist file, 385, 399, 470 

\it (L AT E X 2.09), 367, 373, 485 

Italian, 252 

italic correction, 62, 65, 487 

\itdefault, 372, 551 

\item, 69, 70, 73–5, 77, 78, 200, 214, 

225, 409, 416, 418, 419, 

551 

\item[option], 71, 72, 79, 551 

\itemindent, 77, 79, 551, 604 

itemize environment, 69, 70, 71, 

72, 74, 79, 419, 420, 519 

\itemsep, 76, 77, 552, 604 

\itshape, 64, 371, 372, 416, 458, 

552 

\j, 24, 552 

Japanese, 6 

\jmath, 127, 129, 552 

\Join, 127, 552 

\jot, 149, 552 

JPEG images, 162, 237, 238 

.jpeg file, 162 

.jpg file, 162 

\jput (epic), 302, 304 

justification, 28, 34 

\k (˛ accent), 503, 552 

\kappa, 125, 552 

\keepsilent (DocStrip), 465 

\ker, 128, 552 

keyboard input, 211 

keyboard symbols, 120 

\kill, 82, 552 

Knappen, Jörg, 499 

Kneser, Thomas, 178 

Knuth, Donald E., 6, 7, 119, 151, 

218, 381, 488, 494, 497, 

605 

Kopka, Helmut, 606 

Kwok, Conrad, 305 

\L, 24, 552 

\l, 24, 552 

label 

for bibliography, see 

bibliography entry 

for lists, see list labels 

\label, 56, 131, 139, 177, 182, 183, 

213, 247, 278, 395, 396, 

430, 431, 552 

\labelenumn, 73, 552 

\labelitemn, 73, 553 

\labelsep, 77, 200, 553, 604 

\labelwidth, 77, 79, 200, 553, 604 

lablst.tex, 214, 385, 391 

\Lambda, 125, 553 

\lambda, 125, 553 

Lamport, Leslie, 7, 8, 75, 193, 207, 

218, 228, 229, 321, 437, 

467, 606 

landscape environment (lscape), 

159 

landscape option, 38, 232 

\landscapeonly (seminar), 332 

\langle, 132, 270, 553 

language.dat, 252, 256, 385 

language.dat (babel), 253 

\language (T E X), 36, 251, 255, 256, 

553 

\languagename (babel), 254 

languages

accents and special letters, 24 

babel, 252 

Lappish, 462 

\LARGE, 63, 325, 371, 553 

\Large, 63, 371, 458, 553 

\large, 63, 367, 371, 553 

\lasthline (array), 107 

L AT E X 

compatibility with 2.09, 392, 

398, 418, 484 

files, 12, 396–400 

fonts, 61–6 

installing, 381–6 

logo, 18 

sources, 389 

version 3, 8 

L AT E X counters 

enumn, 73, 181 

chapter, 57, 181, 183 

equation, 181, 277 

errorcontextlines, 404 

figure, 181 

footnote, 113, 181, 190 

MaxMatrixCols (A M S), 266 

mpfootnote, 115, 181 

page, 45, 181, 183, 190, 195 

paragraph, 57, 181 

part, 57, 181 

secnumdepth, 56, 58, 458 

section, 57, 181, 183, 277 

subparagraph, 57, 181 

subsection, 57, 181 

subsubsection, 57, 181 

table, 181 

tocdepth, 58 

latex format, 14, 384, 398 

\LaTeX, 18, 440, 553 

latex.fmt, 385, 398 

latex.ltx, 384, 390, 399, 463 

latex.tex, 467 

L AT E X 2.09, 7, 9, 46, 65, 153, 159, 

324, 330, 373, 374, 392, 

422, 430, 440, 443, 445, 

483–6 

latex209.def, 398 

L AT E X2HTML translator, 476, 480 

L AT E X3 Project, 3, 8 

latexbug.tex, 385 

\LaTeXe, 19, 418, 553 

INDEX 627 

L AT E X error messages, see error 

messages 

latexsym package, 126, 285, 392, 

422, 494 

Latin, 252 

layout package, 47, 395 

\lceil, 132, 553 

.ldf file, 253 

\ldots, 123, 137, 263, 553 

\le, 126, 553 

\leadsto, 127, 553 

left margin, 48, 50 

two-sided, 48 

\left, 108, 131, 134, 137, 140, 553 

\Leftarrow, 127, 554 

\leftarrow, 127, 554 

\lefteqn, 140, 554 

\leftharpoondown, 127, 554 

\leftharpoonup, 127, 554 

\leftmargin, 76, 77, 79, 80, 554, 

604 

\leftmarginn, 80, 554 

\leftmark, 44 

\Leftrightarrow, 127, 554 

\leftrightarrow, 127, 554 

\leftroot (A M S), 269, 554 

legalpaper option, 38 

length, 21, 184 

0 value, 21 

fixed, 21 

rubber, see rubber lengths 

setting, see \setlength 

units of, 21 

zero, 21 

length, user-defined, 184 

\lengthtest (ifthen), 194 

\leq, 126, 554 

leqno option, 39, 120, 131, 271, 

272, 280 

\let (T E X), 451 

letter, 351–64 

closing, 353 



local, 356 

standard, 351–6 

distribution list, 353 

enclosures, 353 

house style, 356–64

628 INDEX 

L AT E X standard, 351–6 

letterhead, company, 356–64 

postscript, 353 

recipient name and address, 

352 

reference entry, 356 

return name and address, 351 

salutation, 353 

stickers, 356 

subject entry, 356 

letter class, 37, 351, 391, 460 

letter environment, 352, 356, 519 

letter.cls, 359 

letterhead, 352, 356 

letterpaper option, 38 

\lfloor, 132, 554 

\lfoot (fancyhdr), 44 

\lg, 128, 554 

\lhd, 126, 554 

\lhead (fancyhdr), 44 

ligatures, 26, 488 

breaking of, 26 

Cyrillic, 284 

\lim, 128, 129, 268, 280, 554 

\liminf, 128, 129, 280, 554 

\limits, 123, 128, 260, 280, 555 

\limsup, 128, 129, 280, 555 

line breaking 

discouraged, 32 

encouraged, 31 

forced, 31 

with justification, 31 

suppressed, 29, 32 

with centered text, 67 

with extra space, 31 

with left justification, 67 

with right justification, 67 

line width, 48, 50 

\line (eepic), 305 

\line (picture), 293, 294, 295, 416, 

431, 555 

line, black, see rules 

\linebreak, 31, 33, 434, 555 

\linethickness, 300, 555 

\linewidth, 47, 51, 158, 161, 332, 

555 

Lingnau, Anselm, 179 

list environment, 74, 77–80, 230, 

419, 420, 519 

list labels, 69 

changing style, 72 

in lists, 74 

list markers, see list labels 

list of references, see bibliography 

\listfigurename, 459, 555 

\listfiles, 208, 441, 442, 447, 555 

listings, 69–74 

\listof (float), 179 

\listoffigures, 59, 210, 399, 427, 

556 

\listoftables, 59, 210, 399, 427, 

556 

\listparindent, 76, 79, 556, 604 

lists, 74–80 

format diagram, 76 

nested, 71, 72, 79 

numbered, 75 

standard label, 74 

style parameters, 75 

trivial, 79 

user’s environment, 78 

\listtablename, 459, 556 

literal text, printing, 110, 393 

live.pdf, 383 

\ll, 126, 556 

\llap (T E X), 348 

\ln, 128, 556 

\LoadClass, 359, 398, 421, 422, 

442, 443, 556 

\LoadClassWithOptions, 442, 454, 

556 

\location, 352, 556 

.lof file, 59, 179, 399, 436 

\log, 128, 556 

.log file, 14, 246, 399, 401, 403, 

414, 429 

\Longleftarrow, 127, 557 

\longleftarrow, 127, 557 

\Longleftrightarrow, 127, 557 

\longleftrightarrow, 127, 557 

\longmapsto, 127, 557 

\Longrightarrow, 117, 127, 557 

\longrightarrow, 127, 557 

longtable environment 

(longtable), 108, 519 

longtable package, 108, 395 

.lot file, 59, 399, 436 

Lower Sorbian, 252

lowering text, see text 

lowering, in formulas, see formula 

\lq (T E X), 557 

LR box, 85, 87 

lrbox environment, 87, 418, 519 

lscape package, 159 

ltxdoc class, 392, 397, 447, 471 

ltxdoc.cfg, 472 

ltxdoc.dtx, 472 

\lVert (A M S), 270, 557 

\lvert (A M S), 270, 557 

LyX, 15 

Macintosh, see Apple 

macro environment (doc), 469 

macrocode environment (doc), 469 

\mainmatter, 57, 210, 557 

\makebox, 86, 90, 364, 557 

\makebox (picture), 161, 291, 292, 

299, 300, 557 

makebst.tex, 321 

\makeglossary, 230, 398, 416, 557 

makeidx package, 228, 392, 399, 

460 

MakeIndex, 15, 228–9, 382, 389, 

392, 399, 470, 550 

options, 229 

style file, 229 

\makeindex, 226, 228, 396, 399, 

416, 557 

makeindex.tex, 229 

\makelabel, 75, 77, 557 

\makelabels, 355, 557 

\MakeLowercase, 455, 557 

\MakeShortVerb (shortvrb), 111, 

393, 468, 558 

\maketitle, 52–4, 54, 69, 419, 431, 

558 

\MakeUppercase, 455, 457, 458, 558 

Maltese, 462 

Malyshev, Basil, 506 

.map file, 234 

\mapsto, 127, 558 

Marchal, B., 606 

margin, 48–50 

margin mark, 116 

marginal notes, 116–18 

\marginpar, 116, 117, 417, 419, 

420, 430, 558 

INDEX 629 

\marginparpush, 118, 558 

\marginparsep, 50, 118, 558 

\marginparwidth, 50, 118, 228, 558 

\margins (pdfscreen), 342 

\markboth, 43, 454, 558 

\markright, 43, 58, 454, 558 

markup, 4–5, 11, 14, 17, 197, 207 

logical, 5, 7, 61, 112, 456, 479 

typographical, 4, 28, 479 

markup languages, 4 

math accents, 129 

math alphabets, 133, 258, 373 

math environment, 119, 519 

math equation, see formula 

math fine-tuning, 145–8 

bracket sizing, 148 

font size, 146–7 



math formula, see formula 

math mode, 119 

math standards, 144 

math style parameters, 148 

math symbols, see symbols 

math variable, see formula 

math version, 373 

\mathalpha, 375 

\mathbb (A M S), 284 

\mathbf, 133, 143, 258, 373, 416, 

422, 558 

\mathbin, 375 

\mathcal, 125, 133, 258, 284, 373, 

559 

\mathclose, 375 

\mathfrak, 285 

\mathindent, 40, 120, 124, 142, 

149, 280, 559 

\mathit, 133, 258, 373, 416, 559 

MathML, 480 

\mathnormal, 125, 133, 373, 559 

\mathop, 375 

\mathopen, 375 

\mathord, 375 

\mathpunct, 375 

\mathrel, 375 

\mathring, 129, 559 

\mathrm, 123, 133, 143, 144, 258, 

373, 559 

mathscr option, 284

630 INDEX 

\mathscr, 284 

\mathsf, 133, 144, 258, 373, 559 

\mathtt, 133, 258, 373, 559 

\mathversion, 373, 433, 559 

matrix, 133–6 

small in text formulas, 138 


matrix environment (A M S), 266, 

519 

\matrixput (epic), 302, 303 

Mattes, Eberhardt, 382 

\max, 128, 129, 280, 559 

MaxMatrixCols counter (A M S), 

266 

\mbox, 32, 86, 87, 90, 133, 144, 150, 

151, 426, 439, 559 

.mbs file, 321, 322 

\mddefault, 372, 559 

\mdseries, 64, 371, 560 

\medskip, 33, 560 

\medskipamount, 560 

\medspace (A M S), 269, 560 

merlin.mbs, 321 

Merz, Thomas, 606 

\MessageBreak, 446, 447, 560 

messages, error, see error messages 

messages, processing, see 

processing messages 

METAFONT, 6, 8, 233, 389, 488, 

494, 497–9, 501 

.mf file, 389, 400, 488, 498, 506 

\mho, 127, 560 

\mid, 126, 560 

MikT E X for Windows, 381, 386 

miktex.txt, 384 

\min, 128, 129, 280, 560 

minimal class, 392 

minipage environment, 88, 89, 90, 

92, 93, 113, 142, 200, 212, 

290, 292, 417, 419, 455, 

520 

minus sign, 23 

minus, 22, 47, 77 

Mittelbach, Frank, 8, 51, 178, 215, 

257, 282, 367, 464, 467, 

605 

mm (millimeter), 21 

\mod (A M S), 268, 560 

mode 

left-to-right, 86 

math, 119 

paragraph, 88 

\models, 126, 560 

monitor input/output, 211 

\month (T E X), 27, 460 

Moore, Ross, 307, 605 

\mp, 126, 560 

mpfootnote counter, 115, 181 

\mspace (A M S), 269, 560 

\mu, 125, 560 

multi-file text, 207 

merging, 207 

selective processing, 208 

multicol package, 51–2, 395 

multicols environment, 51, 395, 

520 

multicols* environment, 52 

\multicolumn, 97, 101–3, 106, 151, 

417, 418, 560 

multiline formula, 120, 138, 270–7 

multilingual L AT E X, 251–6 

\multiput (picture), 289, 290, 291, 

293, 295, 428, 561 

\multiputlist (epic), 302 

multline environment (A M S), 270, 

271, 520 

\multlinegap (A M S), 271, 561 

myheadings page style, 42, 45, 58, 

335, 452, 454 

\nabla, 127, 561 

name commands, 459 

\name, 351, 352, 353, 355, 356, 358, 

561 

nameless environment, 20, 62, 63 

namelimits option, 280 

natbib package, 219–24, 476 

natnotes.tex, 224 

\natural, 127, 561 

\nearrow, 127, 561 

\NeedsTeXFormat, 359, 421, 432, 

438, 440, 454, 561 

\neg, 127, 561 

negation symbols, 126 

\negmedspace (A M S), 269, 561 

\negthickspace (A M S), 269, 561 

\negthinspace (A M S), 269, 561 

\neq, 126, 561

netpbm library, 476 

network servers, 390 

New Font Selection Scheme, see 

NFSS, see NFSS 

\newboolean (ifthen), 194, 195, 

359, 451, 452, 561 

newcent package, 234 

NewCenturySchlbk, PostScript font, 

234, 504 

\newcolumntype (array), 107 

\newcommand, 25, 185, 202, 416, 

418, 425, 426, 438, 449, 

562 

\newcommand*, 445, 562 

\newcounter, 21, 75, 77, 182, 190, 

203, 416, 418, 562 

not with \include, 210 

\newenvironment, 78, 196, 202, 

416, 418, 425, 439, 562 

\newenvironment*, 445, 562 

\newfloat (float), 179 

\newfont, 66, 364, 373, 416, 562 

\newif (T E X), 451 

\newlength, 21, 185, 361, 416, 426, 

562 

\newline, 31, 32, 419, 434, 563 

\newpage, 33, 34, 60, 84, 108, 428, 

563 

forbidden, 84 

\newpagestyle (seminar), 335 

\newsavebox, 21, 87, 106, 199, 203, 

300, 361, 416, 426, 563 

\newslide (seminar), 331, 336 

\newslideframe (seminar), 334 

\newtheorem, 80, 81, 214, 395, 416, 

563 

\newtheorem (amsthm), 281 

\newtheorem* (amsthm), 281 

\newtheoremstyle (amsthm), 281 

NeXt coding, 462 

NFSS, 9, 64, 367–80, 398, 484, 489 

EC fonts, 379 

encoding commands, 379 

font attributes, 64–5, 368–72 

defaults, 372 

font definition, 376–9 

installing, 372–80 

nfssfont.tex, 385, 391 

\NG, 503, 563 

INDEX 631 

\ng, 503, 563 

\ni, 126, 563 

\nobottombuttons (pdfscreen), 

342 

\nobreak, 364 

\nocite, 310, 563 

\noexpand (T E X), 451 

\noextraslang (non-standard), 255 

\nofiles, 396, 397–400, 416, 436, 

563 

\noindent, 46, 563 

nointlimits option, 279 

\nolimits, 128, 280, 563 

\nolinebreak, 32, 33, 564 

nonamelimits option, 280 

\nonfrenchspacing, 29, 564 

\nonumber, 139, 142, 564 

\nopagebreak, 33, 564 

\normalcolor (color), 167, 564 

\normalfont, 65, 167, 371, 455, 

458, 564 

\normalmarginpar, 117, 564 

\normalsize, 63, 371, 412, 424, 

458, 564 

North Sami, 252 

Norwegian, 252 

nosumlimits option, 279 

\not, 126, 564 

\notag (A M S), 271, 565 

note environment (seminar), 337 

note environment (slides), 326, 

520 

\notin, 126, 565 

notitlepage option, 39 

\noxcomment (seminar), 337 

\nu, 125, 565 

null.tex, 413 

numbering in lists, 75 

\numberwithin (A M S), 277, 565 

\nwarrow, 127, 565 

\O, 24, 379, 565 

\o, 24, 565 

\oddsidemargin, 48, 50, 361, 453, 

565 

\odot, 126, 565 

\OE, 24, 565 

\oe, 24, 565 

ogonek accent, 503

632 INDEX 

\oint, 128, 565 

\Omega, 125, 565 

\omega, 125, 565 

\ominus, 126, 565 

OML encoding, 491, 495 

OMS encoding, 491, 495 

OMX encoding, 491, 496 

one-column page formatting 

for single pages, 51 

onecolumn option, 38 

\onecolumn, 51, 565 

oneside option, 38 

\OnlyDescription (doc), 468, 472 

\onlynotes (slides), 328, 565 

\onlyslides (seminar), 336 

\onlyslides (slides), 327, 565 

Oostrum, Piet van, 43, 454 

openany option, 39 

openbib option, 39 

\opening, 353, 355, 356, 363, 566 

openright option, 39 

\oplus, 126, 566 

\OptionNotUsed, 443, 566 

options 

defining, 442 

global, 42, 443 

local, 42, 444 

processing, 443 

to classes, 37 

to packages, 41 

\or (T E X), 451 

\originalTeX (babel), 255 

\oslash, 126, 566 

OT1 encoding, 490, 492, 506 

ot1cmr.fd, 378, 398 

ot1enc.def, 236, 378 

ot1ptm.fd, 236 

OT2 encoding, 283, 497 

otherlanguage environment 

(babel), 254 

otherlanguage* environment 

(babel), 254 

\otimes, 126, 566 

outline fonts, see type 1 fonts 

\oval (eepic), 305 

\oval (picture), 296, 297, 300, 431, 

566 

\Ovalbox (fancybox), 94, 566 

\ovalbox (fancybox), 94, 566 

\overbrace, 136, 566 

overbraces in formulas, 136 

Overfull \hbox warning, 433 

Overfull \vbox warning, 434 

overlapping text, 86, 160, 348 

overlay environment (slides), 

326, 521 

\overlay (pdfscreen), 342 

\overlay (seminar), 337 

\overleftarrow (A M S), 262, 566 

\overleftrightarrow (A M S), 262, 

566 

\overline, 136, 567 

overlining, 136 

\overrightarrow (A M S), 262, 567 

\overset (A M S), 262, 567 

oztex, graphics option, 154 

\P, 23, 567 

package, 8, 9, 12, 13, 41 

afterpage, 171, 394 

alltt, 111, 392 

amsbsy, 144, 192, 258–9 

amscd, 258, 263, 282 

amsfonts, 126, 284–5, 422, 

494 

amsmath, 192, 258 

amsopn, 258, 267–8 

amssymb, 284, 285 

amstext, 258–9 

amsthm, 81, 258, 280–2, 395 

apalike, 220 

array, 107, 394 

avant, 234 

babel, 252–6 

background, 345–6 

bm, 144, 394 

bookman, 234 

calc, 394 

chapterbib, 224 

chicago, 220 

color, 153, 166, 324, 325, 

335, 338, 345, 439 

dcolumn, 107, 394 

delarray, 108, 394 

doc, 392, 399, 441, 467–73 

documentation, 13 

eepic, 305–6 

eepicemu, 305

enumerate, 74, 394 

epic, 302–5 

epsf, 159 

epsfig, 159 

eso-pic, 346 

eucal, 284–5 

eufrak, 285 

europs, 25, 387 

eurosans, 25, 387 

eurosym, 25, 28 

exscale, 392 

fancybox, 94–5, 334, 338, 345 

fancyhdr, 43–5, 249, 335, 454 

flafter, 171, 392 

float, 179–80 

floatfig, see floatflt 

floatflt, 178–9 

fontenc, 378–9, 392, 501 

fontsmpl, 394 

french, 251 

ftnright, 394 

fullpage, 452, 465 

geometry, 49–51, 452 

german, 251, 252, 360 

graphics, 153, 155–7 

graphicx, 157–9 

graphpap, 301, 392 

harvard, 220 

helvet, 234 

here, 178 

hhline, 395 

html, 476 

hyperref, 112, 161, 220, 

239–49, 330, 339, 340, 

344 

ifthen, 193–5, 359, 392, 439, 

452 

indentfirst, 46, 395 

inputenc, 26, 392, 461–2 

latexsym, 126, 285, 392, 422, 

494 

layout, 47, 395 

longtable, 108, 395 

lscape, 159 

makeidx, 228, 392, 399, 460 

multicol, 51–2, 395 

natbib, 219–24, 476 

newcent, 234 

palatino, 234 

INDEX 633 

parskip, 46–7 

pause, 346–9 

pdfscreen, 340–3 

pp4link, 349 

pstricks, 308 

rotating, 159 

sempdftx, 340 

shortvrb, 111, 393 

showidx, 228, 392 

showkeys, 395 

somedefs, 395 


syntonly, 393 

t1enc, 393 

tabularx, 107, 395 

TeX4ht, 477 

textcomp, 24–5, 393, 503 

theorem, 81, 282, 395 

times, 234 

tracefnt, 393, 433 

upref, 282 

url, 112 

varioref, 215–16, 253, 396 

verbatim, 111, 118, 396 

with options, 41 

xr, 215, 243, 396 

xspace, 186, 396 

\PackageError, 439, 446, 567 

\PackageInfo, 447, 567 

packages, catalogue of, 13 

\PackageWarning, 439, 446, 567 

\PackageWarningNoLine, 446, 567 

page breaking, 33 

discouraged, 33 

encouraged, 33 

forced, 33 

with justification, 33 

two columns, 34 

two-sided, 34 

with tabbing, 84 

with figures and tables, 34 

page counter, 45, 181, 183, 190, 

195 

page format, 47 

diagrams, 48 

full page, 452–4 

one-column, 48, 51 

one-sided, 38 

setting values, 47

634 INDEX 

two-column, 38, 51 

two-sided, 38, 50 

with geometry, 49–51 

page numbering 

alphabetic, 45 

Arabic, 45 

changing, 45 

in heading, 42 

Roman numerals, 20, 45 

setting, 45 


page style 

align (seminar), 335 


empty, 42, 327, 335, 352, 452 

fancy, 43 

firstpage, 362 

headings, 42, 44, 55, 58, 327, 

335, 352, 363, 452 

myheadings, 42, 45, 58, 335, 

452, 454 

plain, 42, 45, 327, 335, 352, 

363, 452 

in letters, 352 

page transitions, PDF, 245, 344 

\pagebreak, 33, 34, 84, 178, 278, 

567 

no effect, 84 

\pagecolor (color), 167, 335, 338, 

567 

\PageIndex (doc), 469 

\pagename, 359, 360, 363, 460, 568 

\pagenumbering, 20, 21, 45, 58, 60, 

568 

\pageref, 56, 178, 209, 213, 214, 

247, 282, 395, 396, 430, 

431, 568 

\pageref* (hyperref), 247 

\pagestyle, 42, 58, 326, 327, 335, 

352, 452, 455, 568 

pagetrans.tex, 344 

pagination, see page numbering 

palatino package, 234 

Palatino, PostScript font, 234, 504 

\paneloverlay (pdfscreen), 343 

\panelwidth (pdfscreen), 342 

paper size, option, 38 

\paperheight, 48, 50, 452, 453, 

568, 604 

\paperwidth, 48, 50, 195, 452, 453, 

568, 604 

\par, 23, 364, 419, 426, 445, 568 

paragraph, 11 

box, 85, 88 

paragraph counter, 57, 181 

paragraph character, 23 

paragraph termination 

with blank line, 11 

with \par, 23 

\paragraph, 55, 568 

\paragraph*, 55, 569 

paragraphing 

extra line space, 11, 22, 46 

indentation, 11, 46 

after section command, 46 

setting, 20, 46 


\parallel, 126, 569 

\parbox, 88, 89, 90, 92, 142, 151, 

174, 290, 292, 362, 417, 

419, 439, 569 

\parindent, 20, 46, 79, 569 

\parsep, 76, 77, 79, 569, 604 

parskip package, 46–7 

\parskip, 22, 46, 79, 184, 364, 569, 

604 

part counter, 57, 181 

\part, 55, 569 

\part*, 55, 569 

\partial, 127, 569 

Partl, H., 46, 251 

\partname, 460, 569 

\partopsep, 75, 570, 604 

PassiveT E X, 480 

\PassOptionsToClass, 359, 443, 

570 

\PassOptionsToPackage, 443, 570 

Patashnik, Oren, 321 

\path, 568 

\path (eepic), 306 

\path (url), 112 

pause package, 346–9 

\pause (pause), 346 

\pausebuild (pause), 348 

\pausecolors (pause), 348 

\pausehighlight (pause), 348 

\pauselevel (pause), 347 

.pbm file, 476

pc (pica), 21 

pctex32, graphics option, 154 

pctexhp, graphics option, 154 

pctexps, graphics option, 154 

pctexwin, graphics option, 154 

PDF, 3, 8, 9, 16, 162–3, 231, 236–9, 

242–6, 307, 338, 339, 480, 

481, 498 

slide show, 323, 330, 333, 338 

with PostScript fonts, 234, 505 

with PPower4, 343 

.pdf file, 162, 237 

pdfL AT E X, 238, 324 

pdflatex format, 238, 385 

pdflatex.fmt, 385 

\pdfoutput, 238 

\pdfpageheight (pdfT E X), 339 

\pdfpagewidth (pdfT E X), 339 

pdfscreen package, 340–3 

pdfT E X, 6, 16, 238–9, 305, 346, 385, 

387, 389, 481 

graphics inclusion, 162–3 

with seminar, 339 

pdftex, graphics option, 154, 238 

pdftex.cfg, 238 

pdftex.dll, 386 

Perl, 476 

\perp, 126, 570 

.pfa file, 389 

.pfb file, 234, 389, 506 

\Phi, 125, 570 

\phi, 125, 570 

\Pi, 125, 570 

\pi, 125, 570 

\picsquare (epic), 302, 304 

picture 

arrow, 294 

allowed slopes, 295 

min. length, 295 

box, 291–3 

dashed, 291, 292 

examples, 291, 292 

framed, 291 

parbox inside, 292 

positioning argument, 291 

unframed, 291, 292 

with text, 291 

circle, 295 

filled, 295 

INDEX 635 

open, 295 

commands, 289–300 

coordinate system, 287 

curves, 299 

elements, 289 

arrow, 294 

box, 291–3 

circle, 295 

half circle, 297 

line, 294 

oval, 296 

positioning, 289 

quarter circles, 297 

repeated, 289 

text, 290 

floating, see float 

framed text, 298 

half circles, 297 

length unit, 287, 288 

limited command set, 289 

line, 294 

allowed slopes, 293, 294 

min. length, 294 

sloping, 293 

vertical, 293 

line thickness, 289, 300 

multiline text, 290 

ref. point, 290 

oval, 296 

examples, 296, 297 

positioning, 287 

positioning commands, 289 

quarter circles, 297 

saving, 300 

shifting, 301 

superimposing images, 160 

vertical text, 297 

x-axis, 287 

y-axis, 287 

picture environment, 93, 160, 

175, 288, 289, 295, 298, 

302, 428, 431, 521 

with offset, 301 

pixel fonts, 8 

.pk file, 233, 234, 236, 388, 400, 

488, 498, 504 

plain bibliography style, 219, 310 

plain format, 398

636 INDEX 

plain page style, 42, 45, 327, 335, 

352, 363, 452 

plain.fmt, 398 

plainnat bibliography style, 219, 

221, 311 

Plain T E X, 7, 398, 408 

plus, 22, 47, 77, 78 

\pm, 126, 570 

pmatrix environment (A M S), 266, 

521 

\pmb (A M S), 258, 570 

\pmod, 129, 268, 570 

PNG images, 162, 237, 238 

.png file, 162 

\pod (A M S), 268, 570 

Podar, Sunil, 302 

poetry, 68 

pointers, 127 

Polish, 252 

Popineau, Fabrice, 381 

\poptabs, 83, 84, 419, 570 

Portable Document Format, see PDF 

\portraitonly (seminar), 332 

Portuguese, 252 

\postamble (DocStrip), 465 

PostScript, 3, 15, 64, 154, 162, 163, 

231–6, 239, 308, 335, 337, 

338, 367, 370, 475, 498 

compressing, 166 

driver, 153, 154, 162, 165, 168, 

231 

encapsulated, 153, 155, 162–5, 

232 

euro fonts, 25, 387 

fonts, 233–6, 238, 239, 331, 

369, 372, 378, 389, 503–5 

installing, 234 

printer, 232, 236, 305 

T E X fonts, 235, 505 

under NFSS, 233, 378 

viewer, 164, 232, 305 

pound sign, 23, 27, 28, 570 

\pounds, 23, 27, 570 

pp4link package, 349 

.ppm file, 478 

PPower4, 343–9 

\Pr, 128, 129, 280, 570 

preamble, 12, 208 

\preamble (DocStrip), 465 

\prec, 126, 571 

\preceq, 126, 571 

\prefacename, 460 

preload.cfg, 385 

preload.ltx, 385 

presentation material, see slides 

\prime, 127, 571 

print environment (pdfscreen), 

343 

\PrintChanges (doc), 470 

printer driver, 15, 153, 389, 391, 

398, 453, 488, 497 

\PrintIndex (doc), 470 

\printindex, 210, 228, 399, 571 

\printlandscape (seminar), 333 

proc class, 391 

processing messages, 399, 414 

\ProcessOptions, 359, 443, 447, 

571 

\ProcessOptions*, 443, 444, 453, 

571 

\prod, 128, 137, 571 

programming L AT E X, 437–51 

deferred coding, 444 

durability, 440 

error messages, 446 

examples 

custom headline, 454–6 

customized letter, 359–64 

full page, 452–4 

sectioning, 456–9 

file checking, 440–2 

guidelines, 438 

loading files, 442 

options, 442–4 

portability, 437 

robust commands, 444 

warnings, 446 

proof environment (amsthm), 282 

\proofname, 460 

\propto, 126, 571 

\protect, 211, 426, 429, 440, 445, 

446, 535, 571 

protected space, 22, 29, 144 

protocol, see transcript file 

\providecommand, 187, 363, 439, 

449, 571 

\providecommand*, 445, 571

\ProvidesClass, 359, 416, 432, 

441, 454, 571 

\ProvidesFile, 442, 448, 472, 571 

\ProvidesPackage, 416, 432, 441, 

572 

\ProvideTextCommand, 379, 572 

\ProvideTextCommandDefault, 

380, 572 

\ps, 353, 572 

.ps file, 232, 236 

\psfig (epsfig), 159 

psfonts.map, 235 

\Psi, 125, 572 

\psi, 125, 572 

psnfss, PostScript support, 233 

PSTricks, 308 

pstricks package, 308 

pt (point), 21 

\ptsize (seminar), 331 

\pushtabs, 83, 84, 419, 572 

\put (picture), 161, 289, 290–9, 

302, 572 

\putfile (epic), 302, 304 

\qbezier, 299, 572 

\qed (amsthm), 282 

\qedsymbol (amsthm), 282 

\qquad, 30, 124, 269, 572 

\quad, 30, 124, 269, 273, 572 

quotation environment, 61, 67, 

68, 79, 521 

quotation marks, 23, 29 

French, 503 

quote environment, 19, 61, 67, 68, 

79, 197, 521 

\quotedblbase, 503, 573 

\quotesinglbase, 503, 573 

r.tex, 413 

\r (˚ accent), 24, 573 

Radhakrishnan, C. V., 340 

\raggedbottom, 48, 573 

\raggedleft, 67, 573 

\raggedright, 67, 573 

\raggedslides (seminar), 334 

Rahtz, Sebastian, 154, 159, 239, 

330, 382, 480, 605 

Raichle, Bernd, 251 

INDEX 637 

\raisebox, 87, 88, 92, 102, 104, 

362, 573 

\raisetag (A M S), 271 

raising text, see text 

raising, in formulas, see formula 

\rangle, 132, 270, 573 

\rceil, 132, 573 

\Re, 127, 573 

Reckdahl, Keith, 606 

\RecordChanges (doc), 470 

\ref, 56, 131, 139, 177, 182, 209, 

214, 247, 278, 282, 395, 

396, 430, 431, 573 

\ref* (hyperref), 247 

reference, cross, see cross-reference 

references, list of, see bibliography 

\reflectbox (graphics), 156, 573 

\refname, 460, 574 

\refstepcounter, 182, 574 

\reftext. . . (varioref), 216 

relational symbols, 126 

\relax (T E X), 451 

\renewcommand, 44, 185, 202, 415, 

425, 426, 438, 574 

\renewcommand*, 445, 574 

\renewenvironment, 196, 203, 415, 

425, 439, 574 

\renewenvironment*, 445, 574 

\renewpagestyle (seminar), 335 

report class, 37, 41, 55, 113, 130, 

173, 183, 187, 191, 218, 

224, 225, 277, 391, 459 

reqno option, 271, 272, 280 

\RequirePackage, 400, 421, 432, 

442, 574 

\RequirePackageWithOptions, 

442, 574 

\resizebox (graphics), 156, 574 

\resizebox* (graphics), 575 

\restylefoat (float), 180 

\reversemarginpar, 117, 575 

\rfloor, 132, 575 

\rfoot (fancyhdr), 44 

\rhd, 126, 575 

\rhead (fancyhdr), 44 

\rho, 125, 575 

\right, 108, 131, 134, 137, 140, 575 

\Rightarrow, 74, 127, 575 

\rightarrow, 127, 575

638 INDEX 

\rightharpoondown, 127, 575 

\rightharpoonup, 127, 575 

\rightleftharpoons, 127, 575 

\rightmargin, 76, 77, 79, 575, 604 

\rightmark, 44 

\rlap (T E X), 348 

\rm (L AT E X 2.09), 367, 485 

\rmdefault, 233, 372, 379, 575 

\rmfamily, 64, 371, 379, 575 

robust commands, 440, 445 

Rokicki, Tomas, 159, 162, 231, 606 

Roman page numbering, 45, 58, 60 

\Roman, 73, 75, 114, 183, 192, 575 

\roman, 73, 114, 183, 575 

Romanian, 252 

roots, 122, 269 

Rose, Kristoffer H., 307 

rotate environment (rotating), 

159 

\rotatebox (graphics), 156, 576 

\rotatebox (graphicx), 159 

\rotateheaderstrue (seminar), 

333 

rotating graphics and text, 156 

rotating package, 159 

Rowley, Chris, 8 

\rq (T E X), 576 

rubber lengths, 21, 22, 30, 32, 46, 

77, 85, 149, 173, 184, 457 

rule box, 85 

\rule, 18, 91, 106, 116, 173, 332, 

362, 364, 455, 576 

rules 

horizontal, 91 

vertical, 91 

Russian, 252, 283 

\rVert (A M S), 270, 576 

\rvert (A M S), 270, 576 

\S, 23, 576 

s.tex, 394, 413 

Saarela, J., 605 

Samarin, A., 605 

sample2e.tex, 385, 391 

Samuel, Arthur L., 606 

\savebox, 87, 93, 203, 576 

\savebox (picture), 300, 576 

\sb (T E X), 576 

\sbox, 87, 93, 106, 199, 361, 439, 

576 

\sc (L AT E X 2.09), 367, 485 

\scalebox (graphics), 156, 576 

scaling graphics and text, 156 

\scdefault, 372, 576 

Schenk, Christian, 381 

Schmidt, Walter, 25 

Schöpf, Rainer, 8, 257, 367 

Schrod, J., 251 

Schwarz, Norbert, 499, 606 

Scientific Workplace, 382 

Scottish Gaelic, 252 

screen environment (pdfscreen), 

343 

\screensize (pdfscreen), 342 

\scriptscriptstyle, 146, 375, 577 

\scriptsize, 63, 371, 577 

\scriptstyle, 146, 282, 375, 577 

\scshape, 64, 371, 577 

\searrow, 127, 577 

\sec, 128, 577 

secnumdepth counter, 56, 58, 458 

section counter, 57, 181, 183, 277 

\section, 44, 55, 58, 113, 455, 577 

\section*, 55, 577 

sectioning 

*-form, 55, 56 

hierarchy, 55 

numbering, 55 

depth of, 56 

secnumdepth, 57 


title, 55 

short form, 58 

sectioning commands, 55 

redefining, 456 

\see, 227, 228, 577 

\seealso, 227, 577 

\seename, 460, 577 

\selectfont, 369, 370, 372, 376, 

578 

selective file processing, 208 

controlling, 210, 212 

\selectlanguage (babel), 254, 256, 

578 

sem-user.tex, 330 

\semcm (seminar), 332 

\semin (seminar), 332

seminar 

\addtoartlength, 332 

\addtoslidelength, 332 

\addtoslidereset, 337 

\articlemag, 332 

\centerslidefalse, 334 

\centerslidetrue, 334 

\extraslideheight, 336 

\landscapeonly, 332 

\newpagestyle, 335 

\newslide, 331, 336 

\newslideframe, 334 

\noxcomment, 337 

\onlyslides, 336 

\overlay, 337 

\portraitonly, 332 

\printlandscape, 333 

\ptsize, 331 

\raggedslides, 334 

\renewpagestyle, 335 

\rotateheaderstrue, 333 

\semcm, 332 

\semin, 332 

\setartlength, 332 

\setslidelength, 332 

\slidebottommargin, 334 

\slidefootfont, 336 

\slideframe, 331, 334 

\slideframesep, 334 

\slideframewidth, 334 

\slideheadfont, 336 

\slideleftmargin, 334 

\slidepagestyle, 336 

\slidereset, 337 

\sliderightmargin, 334 

\slidesmag, 331 

\slidetopmargin, 334 

\thenote, 335, 337 

\theslide, 335, 337 

seminar class, 330–40 

seminar.con, 339, 340 

sempdftx package, 340 

sempdftx.sty, 340 

Serbian, 252 

series, font attribute, 64, 368 

\seriesdefault, 372 

servers, see network servers 

\setartlength (seminar), 332 

INDEX 639 

\setboolean (ifthen), 194, 359, 

451, 452, 578 

\setbox (T E X), 439 

\setcounter, 21, 182, 183, 203, 

418, 578 

\setlength, 20–2, 75, 83, 173, 184, 

200, 439, 578 

\SetMathAlphabet, 374, 423, 578 

\setminus, 126, 578 

\setslidelength (seminar), 332 

\SetSymbolFont, 374, 578 

\settime (slides), 327, 578 

\settodepth, 184, 579 

\settoheight, 184, 579 

\settowidth, 184, 189, 200, 579 

\sf (L AT E X 2.09), 367, 485 

\sf@size, 376 

\sfdefault, 233, 372, 379, 579 

\sffamily, 64, 66, 325, 371, 379, 

579 

\shadowbox (fancybox), 94, 335, 

579 

\shadowsize (fancybox), 94, 579 

shape, font attribute, 64, 368 

\shapedefault, 372 

\sharp, 127, 579 

shorthnd.tex, 322 

\shortstack, 297, 298, 299, 579 

shortvrb package, 111, 393 

\shoveleft (A M S), 271 

\shoveright (A M S), 271 

showidx package, 228, 392 

showkeys package, 395 

\sideset (A M S), 262, 579 

sideways environment (rotating), 

159 

\Sigma, 125, 579 

\sigma, 125, 579 

\signature, 351, 352, 353, 358, 579 

\sim, 126, 580 

\simeq, 126, 580 

Simonic, Aleksander, 15 

\sin, 128, 580 

\sinh, 128, 580 

size, font attribute, 64, 369 

\sl (L AT E X 2.09), 367, 485 

\sldefault, 372, 580 

slide environment (seminar), 331, 

332, 521

640 INDEX 

slide environment (slides), 325, 

521 

slide* environment (seminar), 

331, 332 

\slidebottommargin (seminar), 

334 

\slidefootfont (seminar), 336 

\slideframe (seminar), 331, 334 

\slideframesep (seminar), 334 

\slideframewidth (seminar), 334 

\slideheadfont (seminar), 336 

\slideleftmargin (seminar), 334 

\slidepagestyle (seminar), 336 

\slidereset (seminar), 337 

\sliderightmargin (seminar), 334 

slides, 323–40 

leading page, 325 

notes, 326 

overlays, 325–6 

page styles, 326 


special environments, 325–6 

timing notes, 327 

slides class, 324–30, 392 

\slidesmag (seminar), 331 

\slidetopmargin (seminar), 334 

SLIT E X, see slides 

\sloppy, 36, 88, 434, 580 

sloppypar environment, 36, 198, 

434, 521 

Slovakian, 252 

Slovenian, 252 

\slshape, 64, 78, 199, 363, 371, 580 

small environment, 110, 517 

\small, 19, 63, 371, 580 

small2e.tex, 385, 391 

smallmatrix environment (A M S), 

267 

\smallskip, 33, 580 

\smallskipamount, 580 

\smash (A M S), 269, 580 

\smile, 126, 580 

Snow, W., 606 

Sojka, Petr, 606 

somedefs package, 395 

source file, 5, 6 

source text, 4, 11, 14, 17 

\sp (T E X), 580 

space, see blank 

\space, 446 

spacing 

character, 29 

French spacing, 29 

horizontal, 29, 30 

line 

local change, 32 


manual adjustment, 28 

paragraph, 11, 22, 46 

unwanted, 201 

vertical, 31–3 

rubber, 32 

word, 28 

after period, 29 

after punctuation, 28 

\spadesuit, 127, 580 

Spanish, 252 

Spannagel, Christian, 343 

special characters, 23 

special letters in other languages, 

24 

\special (T E X), 153, 232, 305, 477 

spelling, American or British, 195 

Spivak, Michael, 257 

\spline (eepic), 306 

split environment (A M S), 270, 

272, 276, 521 

\sqcap, 126, 580 

\sqcup, 126, 580 

\sqrt, 122, 269, 580 

\sqsubset, 126, 581 

\sqsubseteq, 126, 581 

\sqsupset, 126, 581 

\sqsupseteq, 126, 581 

square root sign, 122 

\SS, 24, 576 

\ss, 24, 581 

\ssf@size, 376 

\stackrel, 137, 581 

\star, 126, 581 

\stepcounter, 110, 115, 182, 581 

\StopEventually (doc), 469 

\stretch, 184, 581 

strut, 92, 103, 105 

\strut, 298 

.sty file, 9, 12, 41, 384, 385, 463, 

484 

style files (L AT E X 2.09), 483

style parameter 

float, 171 

footnote, 117 

frame box, 93 

lists, 75 

marginal note, 118 

math, 148 

table, 98 

subarray environment (A M S), 

261, 522 

subequations environment 

(A M S), 277, 522 

\subitem, 225, 226, 581 

subparagraph counter, 57, 181 

\subparagraph, 55, 581 

\subparagraph*, 55, 581 

subscript, 121 

subsection counter, 57, 181 

\subsection, 44, 55, 181, 581 

\subsection*, 55, 581 

\subset, 126, 582 

\subseteq, 126, 582 

\substack (A M S), 261, 581 

\subsubitem, 225, 226, 582 

subsubsection counter, 57, 181 

\subsubsection, 55, 582 

\subsubsection*, 55, 582 

\succ, 126, 582 

\succeq, 126, 582 

\sum, 123, 128, 137, 582 

sumlimits option, 279 

summation sign, 123 

\sup, 128, 129, 280, 582 

superscript, 121 

\suppressfloats, 171, 175, 582 

\supset, 126, 582 

\supseteq, 126, 582 

\surd, 127, 582 

Sutor, Robert, 605 

\swapnumbers (amsthm), 281 

\swarrow, 127, 582 

Swedish, 252 

switch.def (babel), 253 

\symbol, 66, 88, 582 

symbols 

above one another, 137, 261 

alternative commands, 462 

arrows, 127, 262, 282 

binary operators, 125 

INDEX 641 

brackets, 131 

automatic size, 131 

invisible, 132 

manual sizing, 148 

multiline equations, 140, 

148, 276 

direct input, 26, 461 

integral, 123, 260 

keyboard, 120 

letters 

calligraphic, 125, 284 

Greek, 125 

limits, 128, 261 

miscellaneous, 127 

negation, 126 

pointers, 127 

relational, 126 

summation, 123 

two sizes, 128 

\syntaxonly (syntonly), 393 

syntonly package, 393 

T1 encoding, 500, 501, 503, 506 

t1enc package, 393 

t1enc.def, 398, 501 

\t ( accent), 24, 582 

tab stop, 81 

tabbing, 81–5 

accents in, 83 

additional commands, 83 

backwards, 82 

end of line, 81 

left margin, 82 

setting, 82 

nesting, 84 

page breaking, 84 

sample line, 82 

setting stops, 81 


tabbing environment, 81, 83, 84, 

415, 419, 420, 428, 522 

\tabbingsep, 83, 582 

\tabcolsep, 98, 106, 583 

table, 95–110 

column alignment, 101 

on decimals, 107 

construction, 95–8 

@-expression, 96, 101, 110 

column formatting, 96

642 INDEX 

column repetition, 96 


horizontal line, 97 

math arrays, 133–6 

merging columns, 97, 101 

paragraph column, 96, 104 

row end, 97 

vertical line, 96, 97 

vertical positioning, 95 

examples, see table examples 

extension packages, 106, 394 

floating, see float, 169 

footnotes inside, 116 

intercolumn spacing, 96 

math, 133–6 

multi-page, 108 

multipage, 108 

numbering, see float, caption 

setting width, 96, 107 


title, see float 

table counter, 181 

table environment, 60, 108–10, 

169, 173, 182, 213, 417, 

419, 430, 522 

table examples, 98–106 

@-expression, 110 

changing fonts, 102 

column alignment, 101 

column format, 99 

empty box, 105 

floating, 110 

framed, 100 

horizontal line, 100, 102 

math arrays, 133–6 

merging columns, 101 

paragraph column, 104 

line breaking, 105 

row termination, 99 

setting widths, 110 

text between columns, 101 

unframed, 99 

vertical line, 99, 102, 106 

with all table features, 106 

table of contents, 55, 58 

automatic entries, 58 

for figures, 59 

for tables, 59 

level of entry, 58 

tocdepth, 58 

manual entry, 59 

printing, 58 

table* environment, 169, 522 

\tablename, 255, 459, 583 

\tableofcontents, 58, 60, 210, 

396, 400, 427, 436, 583 

tables, list of, 59 

tabular environment, 10, 95, 99, 

102, 107, 116, 179, 297, 

362, 394, 417, 418, 425, 

428, 429, 522 

tabular* environment, 95, 107, 

110, 395, 455, 523 

\tabularnewline, 98, 105, 583 

tabularx environment, 107, 395 

tabularx package, 107, 395 

tabulator, see tabbing 

\tag (A M S), 271, 573, 583 

\tan, 128, 583 

\tanh, 128, 583 

\tau, 125, 583 

Taylor, Philip, 606 

\tbinom (A M S), 265, 583 

tbtags option, 272, 279 

TC fonts, 24, 501 

tcidvi, graphics option, 154 

.tcx file, 481 

TDS, T E X Directory Structure, 388 

techexplorer Hypermedia 

Browser, 481 

TEI DTD, 479, 480 

\telephone, 352, 356, 583 

testpage.tex, 385, 391 

teT E X for Unix, 381 

T E X 

logo, 19 

T E X, 6, 398 

implementations, 381 

T E X commands, 438 

adjusting bracket sizes, 148 

allowed, 450–1 

\atop, 137, 146, 147, 189 

avoiding, 439 

\Big, 148, 525 

\big, 148, 525 

\Bigg, 148, 525 

\bigg, 148, 525 

\brace, 192

\brack, 192 

\centerline, 67, 176, 408, 527 

\choose, 137, 146, 147, 189, 

192 

\day, 27, 460 

\def, 439, 445, 450 

\edef, 450 

\else, 451 

\endinput, 451 

\expandafter, 451 

\fi, 451 


\gdef, 450 

\hbox, 439 

\hoffset, 547, 604 

\hss, 408 

\if, 451 

\ifcase, 451, 461 

\iffalse, 468 

\language, 36, 251, 255, 256, 

553 

\let, 451 

\llap, 348 

\lq, 557 

\month, 27, 460 

\newif, 451 

\noexpand, 451 

\or, 451 

\relax, 451 

\rlap, 348 

\rq, 576 

\sb, 576 

\setbox, 439 

\sp, 580 

\special, 153, 232, 305, 477 

\vbox, 439 

\voffset, 593, 604 

\xdef, 450 

\year, 27, 460 

\TeX, 19, 583 

tex.dll, 386 

T E X4ht program, 477, 479, 480 

TeX4ht package, 477 

T E X error messages, see error 

messages 

.tex file, 14, 17, 208, 209, 391, 396, 

400, 403, 436 

T E XLive, 3, 13, 15, 160, 344, 381–9, 

478, 479, 506 

INDEX 643 

texmf.cnf, 386 

texsys.cfg, 384 

text 

commands in text, 11 

comments in, 118 

components, 11 

framed, 86 

in formulas, 133 

literal, printing, 110, 393 

lowering, 87 

math, 119 

merging, 207 

multiple files, 207 

overlapping, 86 

raising, 87 



vertically stacked, 297 

width calculation, 184 

text companion fonts, 24, 393, 501, 

506 

text editor, 4, 17 

text formula, 119 

text height, 48, 50 

\textsymbol, 463, 584 

\text (A M S), 259, 583 

\textbf, 65, 372, 373, 584 

\textcircled, 463, 584 

\textcolor (color), 167, 335, 584 

textcomp package, 24–5, 393, 503 

\textcompwordmark, 463, 584 

\texteuro, 25 

\texteuro (textcomp), 503 

\textfloatsep, 172, 584 

\textfraction, 172, 584 

\textheight, 48, 50, 332, 361, 434, 

452, 453, 584 

\textit, 65, 372, 585 

\textmd, 65, 372, 585 

\textnormal, 65, 372, 585 

\textquotedbl, 503, 585 

\textregistered, 463, 585 

\textrm, 65, 372, 585 

\textsc, 65, 372, 585 

\textsf, 25, 65, 372, 585 

\textsl, 44, 65, 372, 585 

\textstyle, 137, 146, 264, 375, 585 

\textsuperscript, 463, 585 

\texttrademark, 463, 585

644 INDEX 

\texttt, 65, 88, 372, 585 

\textup, 65, 372, 585 

Textures, 382 

textures, graphics option, 154 

\textwidth, 21, 48, 50, 184, 195, 

332, 361, 452, 453, 455, 

586 

\tf@size, 376 

.tfm file, 234, 400, 487, 498, 503, 

506 

\tfrac (A M S), 264, 586 

\TH, 503, 586 

\th, 503, 586 

Thành, Hàn Thê ´ , 16, 238, 606 

\thanks, 54, 424, 586 

\the, 586 

\thecounter, 183, 586 

thebibliography environment, 

217, 220, 221, 224, 228, 

309, 311, 419, 523 

\thechapter, 187 

\theequation, 191, 277 

\thefootnote, 113, 116, 190 

Theiling, Henrik, 25 

theindex environment, 225, 228, 

399, 523 

\thenote (seminar), 335, 337 

theorem environment, 80, 523 

theorem package, 81, 282, 395 

theorems, 80 


\theoremstyle (A M S), 281 

\thepage, 44, 183, 335, 363, 455 

\thesection, 183, 187, 277 

\theslide (seminar), 335, 337 

\thesubsection, 187 

\Theta, 125, 586 

\theta, 125, 586 

\Thicklines (eepic), 306 

\thicklines (picture), 94, 289, 

293, 300, 497, 586 

\thickspace (A M S), 269, 586 

\thinlines (picture), 94, 289, 300, 

494, 586 

\thinspace (A M S), 269, 586 

\thisfancypage (fancybox), 95, 

586 

\thispagestyle, 21, 43, 327, 335, 

363, 587 

thmtest.tex, 281 

.tif file, 162 

TIFF images, 162, 238 

.tiff file, 162 

\tilde, 129, 263, 587 

\Tilde (A M S), 263, 587 

times package, 234 

\times, 126, 587 

Times-Roman, PostScript font, 234, 

379, 504 

times.sty, 378 

\tiny, 63, 371, 587 

title 

of sections, 55 

of table, see float 

title page, 39, 52 

free format, 54 

standard format, 52 

\title, 52, 53, 54, 69, 419, 431, 587 

titlepage environment, 52, 54, 

69, 210, 523 

titlepage option, 39, 54, 55, 351 

\to, 127, 587 

\toaddress, 359, 363 

tocdepth counter, 58 

.toc file, 59, 400, 436 

\today, 27, 28, 36, 255, 363, 460, 

587 

\toname, 359, 360, 363 

top margin, 48, 50 

\top, 127, 587 

\topfigrule, 173, 587 

\topfraction, 172, 587 

\toplink (pp4link), 349 

\topmargin, 48, 50, 361, 453, 587 

topnumber, 171, 588 

\topsep, 75, 149, 588, 604 

\topskip, 48, 588 

\toptarget (pp4link), 349 

\totalheight, 86, 90, 588 

totalnumber, 172, 588 

tracefnt package, 393, 433 

transcript file, 14, 50, 399, 401 

transliteration, Cyrillic, 284 

\triangle, 127, 588 

\triangleleft, 126, 588 

\triangleright, 126, 588 

trivial lists, 79 

trivlist environment, 79, 523

truetex, graphics option, 154 

TrueType fonts, 238 

TS1 encoding, 502, 506 

\tt (L AT E X 2.09), 65, 367, 485 

\ttdefault, 233, 372, 379, 588 

\ttfamily, 64, 371, 379, 588 

TUG, T E X Users Group, 3, 382, 388 

Turkish, 252, 462 

turn environment (rotating), 159 

\turnbox (rotating), 159 

two-column page formatting, 38 

column dividing rule, 40 


for single pages, 51 

two-sided page formatting, 38, 50 

left margin, 48 

twocolumn option, 34, 38, 40, 51, 

116, 117, 351 

\twocolumn, 34, 51, 395, 431, 588 

twoside option, 34, 38, 43, 44, 49, 

50, 58, 116, 117 

.txt file, 384 

typefaces, see fonts 

\typein, 211, 212, 328, 589 

\typeout, 211, 589 

type 1 fonts, 8, 25, 231, 238, 387, 

498, 503 

type 3 fonts, 8, 234, 238 

\u (˘ accent), 24, 589 

ukhyphen.tex, 256 

Ukrainian, 252 

Umeki, Hideo, 49 

\unboldmath, 143, 373, 433, 589 

\underbrace, 136, 589 

underbraces in formulas, 136 

Underfull \hbox warning, 434 

Underfull \vbox warning, 435 

\underleftarrow (A M S), 262, 589 

\underleftrightarrow (A M S), 

262, 589 

\underline, 136, 362, 458, 589 


\underrightarrow (A M S), 262, 589 

\underset (A M S), 262, 590 

\unitlength (picture), 288, 289, 

290, 590 

units of length, 21 

Unix, 381 

INDEX 645 

\unlhd, 126, 590 

unpack.ins, 384 

\unrhd, 126, 590 

unsrt bibliography style, 310 

unsrtnat bibliography style, 221, 

311 

\Uparrow, 127, 132, 590 

\uparrow, 127, 132, 590 

\updefault, 372, 590 

\Updownarrow, 127, 590 

\updownarrow, 127, 590 

\uplus, 126, 590 

Upper Sorbian, 252 

upref package, 282 

\uproot (A M S), 269, 590 

\upshape, 64, 371, 590 

\Upsilon, 125, 590 

\upsilon, 125, 590 

Urban, Michael, 606 

url package, 112 

\url (url), 112, 590 

\urldef (url), 112 

\urlid (pdfscreen), 342 

\urlstyle (url), 112, 590 

\usebox, 87, 93, 106, 199, 300, 362, 

591 

\usecounter, 75, 77, 78, 591 

\usefont, 369, 591 

\usepackage, 12, 24, 41, 42, 201, 

222, 395, 412, 416, 418, 

421, 422, 432, 441, 591 

\usepostamble (DocStrip), 465 

\usepreamble (DocStrip), 465 

ushyphen.tex, 256 

\v (ˇ accent), 24, 591 

\value, 183, 191, 193, 195, 426, 591 

van Dongen, Marc, 344 

van Zandt, Timothy, 94, 308, 330 

\varepsilon, 125, 591 

variable, math, see formula 

\varinjlim (A M S), 268, 591 

varioref package, 215–16, 253, 

396 

\varliminf (A M S), 268, 591 

\varlimsup (A M S), 268, 591 

\varphi, 125, 591 

\varpi, 125, 591 

\varprojlim (A M S), 268, 591

646 INDEX 

\varrho, 125, 591 

\varsigma, 125, 591 

\vartheta, 125, 591 

\vbox (T E X), 439 

\vdash, 126, 127, 592 

\vdots, 123, 592 

\vec, 129, 144, 263, 592 

\Vec (A M S), 263, 592 

\vector (picture), 161, 295, 299, 

300, 416, 592 

\vee, 126, 592 

\verb, 110, 393, 420, 592 

\verb*, 110, 592 

verbatim environment, 110, 467, 

523 

in verbatim package, 111, 396 

verbatim package, 111, 118, 396 

verbatim* environment, 110, 523 

\verbatiminput (verbatim), 111, 

396 

verse, 68 

verse environment, 61, 68, 79, 523 

version, math, 373 

.vf file, 234, 236, 400, 488, 504 

\vfill, 32, 592 

viewgraphs, see slides 

virtual fonts, 488, 503 

\visible (slides), 326, 592 

\vline, 97, 106, 592 

Vmatrix environment (A M S), 266, 

524 

vmatrix environment (A M S), 266, 

524 

\voffset (T E X), 593, 604 

Volovich, Vladimir, 506 

\vpagecolor (background), 345 

\vpageref (varioref), 215, 396, 

593 

\vpagerefrange (varioref), 216 

\vref (varioref), 215, 396, 593 

\vrefrange (varioref), 216 

\vspace, 32, 92, 332, 364, 419, 455, 

593 

\vspace*, 32, 69, 173, 593 

vtex, graphics option, 154 

Washington, University, 283 

Web Consortium, 479 

Web2c T E X, 381 

web2ctex.txt, 384 

\wedge, 126, 593 

Welsh, 252 

what-you-see-is-what-you-get, 4, 

382 

\whiledo (ifthen), 193, 195, 593 

Wicks, Mark A., 162, 237 

\widehat, 593 

\widetilde, 594 

\width, 86, 90, 594 

Williams, Graham, 13 

Williamson, H. A., 606 

Windows, 381 

Windows coding, 26, 462 

windvi previewer, 154, 234, 305 

WinEdt, 15, 382 

WinShell, 15, 382 

Woliński, Marcin, 464 

Wooding, Mark, 464 

word division, see hyphenation 

word spacing, see spacing 

World Wide Web, 4, 8, 9, 475–82 

\wp, 127, 594 

\wr, 126, 594 

x.tex, 394, 413 

\xdef (T E X), 450 

xdvi previewer, 15, 154, 234, 305 

\Xi, 125, 594 

\xi, 125, 594 

\xleftarrow (A M S), 263, 594 

XML, 3, 5, 10, 478–81 

xmltex format, 481 

xmltex.tex, 480 

xr package, 215, 243, 396 

\xrightarrow (A M S), 263, 594 

XSL, 5, 479 

xspace package, 186, 396 

XY-pic, 307 

Y&Y, 382, 489, 506 

\year (T E X), 27, 460 

\zeta, 125, 594 

zip program, 166 

.zip file, 166 

Zlatuˇska, Jiˇrí, 606

Guide to LaTeX (4th Edition) (Tools and Techniques

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?