LaTeX/Print version

Permission is granted to copy, distribute, and/or modify this document under the terms of the Creative Commons Attribution-ShareAlike 3.0 Unported License.


If you have saved this file to your computer, click on a link in the contents to go to that section.

Getting Started
  1. Introduction
  2. Installation
  3. Installing Extra Packages
  4. Basics
Common Elements
  1. Document Structure
  2. Text Formatting
  3. Paragraph Formatting
  4. Colors
  5. Fonts
  6. List Structures
  7. Special Characters
  8. Internationalization
  9. Rotations
  10. Tables
  11. Title creation
  12. Page Layout
  13. Importing Graphics
  14. Floats, Figures and Captions
  15. Footnotes and Margin Notes
  16. Hyperlinks
  17. Labels and Cross-referencing
  1. Errors and Warnings
  2. Lenghts
  3. Counters
  4. Boxes
  5. Rules and Struts
Technical Texts
  1. Mathematics
  2. Advanced Mathematics
  3. Theorems
  4. Chemical Graphics
  5. Algorithms
  6. Source Code Listings
  7. Linguistics
Special Pages
  1. Indexing
  2. Glossary
  3. Bibliography Management
  4. More Bibliographies
Special Documents
  1. Letters
  2. Presentations
  3. Teacher's Corner
  4. Curriculum Vitae
Creating Graphics
  1. Introducing Procedural Graphics
  2. MetaPost
  3. Picture
  4. PGF/TikZ
  5. PSTricks
  6. Xy-pic
  7. Creating 3D graphics
  1. Macros
  2. Plain TeX
  3. Creating Packages
  4. Themes
  1. Modular Documents
  2. Collaborative Writing of LaTeX Documents
  3. Export To Other Formats
Help and Recommendations
  1. FAQ
  2. Tips and Tricks
  1. Authors
  2. Links
  3. Package Reference
  4. Sample LaTeX documents
  5. Index
  6. Command Glossary


What is TeX?

TeX is a language created by Donald Knuth to typeset documents attractively and consistently. Knuth started writing the TeX typesetting engine in 1977 to explore the potential of the digital printing equipment that was beginning to infiltrate the publishing industry at that time, in the hope that he could reverse the trend of deteriorating typographical quality that he saw affecting his own books and articles. While TeX is a programming language in the sense that it is Turing complete, its main job is to serve as a markup language for describing how your document should look. The fine control TeX offers over document structure and formatting makes it a powerful and formidable tool. TeX is renowned for being extremely stable, for running on many different kinds of computers, and for being virtually bug free. The version numbers of TeX are converging toward the mathematical constant  , with the current version number being 3.1415926.

The name TeX is intended by its developer to be /'tɛx/, /x/ being the velar fricative, the final consonant of loch and Bach. (Donald E. Knuth, The TeXbook) The letters of the name are meant to represent the capital Greek letters tau, epsilon, and chi, as TeX is an abbreviation of τέχνη (ΤΕΧΝΗ – technē), Greek for both "art" and "craft", which is also the root word of technical. English speakers often pronounce it /'tɛk/, like the first syllable of technical.

The tools TeX offers "out of the box" are relatively primitive, and learning how to perform common tasks can require a significant time investment. Fortunately, document preparation systems based on TeX, consisting of collections of pre-built commands and macros, do exist. These systems save time by automating certain repetitive tasks; however, this convenience comes at the cost of complete design flexibility. One of the most popular macro packages is called LaTeX.

What is LaTeX?

LaTeX (pronounced either "Lah-tech" or "Lay-tech") is a set of macros for TeX created by Leslie Lamport. Its purpose is to simplify the TeX typesetting, especially for documents containing mathematical formulae. Within the typesetting system, its name is formatted as LaTeX.

TeX is both a typographical and logical markup language, and one has to take account of both issues when writing a TeX document. In creating LaTeX, Lamport's aim was to split those two aspects. A typesetter can make a template and the writers can focus on LaTeX logical markup despite perhaps not know anything about typesetting.

In addition to the commands and options LaTeX offers, many other authors have contributed extensions, called packages or styles, which you can use for your documents. Many of these are bundled with most TeX/LaTeX software distributions; more can be found in the Comprehensive TeX Archive Network (CTAN).

Why should I use LaTeX?

Most readers will be familiar with WYSIWYG (What You See Is What You Get) typesetting systems such as LibreOffice Writer, Microsoft Word, or Google Docs. Using LaTeX is fundamentally different from using these other programs—instead of seeing your document as it comes together, you describe how you want it to look using commands in a text file, then run that file through the LaTeX program to build the result. While this has the disadvantage of needing to pause your work and take multiple steps to see what your document looks like, there are many advantages to using LaTeX:

  • You can concentrate purely on the structure and contents of the document. LaTeX will automatically ensure that the typography of your document—fonts, text sizes, line heights, and other layout considerations—are consistent according to the rules you set.
  • In LaTeX, the document structure is visible to the user, and can be easily copied to another document. In WYSIWYG applications it is often not obvious how a certain formatting was produced, and it might be impossible to copy it directly for use in another document.
  • Indexes, footnotes, citations and references are generated easily and automatically.
  • Mathematical formulae can be easily typeset. (Quality mathematics was one of the original motivations of TeX.)
  • Since the document source is plain text,
    • Document sources can be read and understood with any text editor, unlike the complex binary and XML formats used with WYSIWYG programs.
    • Tables, figures, equations, etc. can be generated programmatically with any language.
    • Changes can be easily tracked with version control software.
  • Some academic journals only accept or strongly recommend submissions in the form of LaTeX documents. Publishers offer LaTeX templates.

When the source file is processed by the LaTeX program, or engine, it can produce documents in several formats. LaTeX natively supports DVI and PDF, but by using other software you can easily create PostScript, PNG, JPEG, etc.

Terms regarding TeX

Document preparation systems

LaTeX is a document preparation system based on TeX. So the system is the combination of the language and the macros.


TeX distributions are collections of packages and programs (compilers, fonts, and macro packages) that enable you to typeset without having to manually fetch files and configure things.


An engine is an executable that can turn your source code into a printable output format. The engine by itself only handles the syntax. It also needs to load fonts and macros to fully understand the source code and generate output properly. The engine will determine what kind of source code it can read, and what format it can output (usually DVI or PDF).

All in all, distributions are an easy way to install what you need to use the engines and the systems you want. Distributions usually target specific operating systems. You can use different systems on different engines, but sometimes there are restrictions. Code written for TeX, LaTeX or ConTeXt are (mostly) not compatible. Additionally, engine-specific code (like font for XeTeX) may not be compiled by every engine.

When searching for information on LaTeX, you might also stumble upon XeTeX, ConTeXt, LuaTeX or other names with a -TeX suffix. Let's recap most of the terms in this table.

Systems Descriptions
AMSTeX A legacy TeX macro-based document preparation system used by the American Mathematical Society (AMS) from 1982 to 1985. It evolved into the AMS-LaTeX collection which includes the amsmath package used in nearly every LaTeX document as well as mutliple AMS publication layout standards (document classes).
ConTeXt A TeX macro-based document preparation system designed by Hans Hagen and Ton Otten of Pragma ADE in the Netherlands around 1991. It is compatible with the pdfTeX, XeTeX and LuaTeX engines.

ConTeXt assumes the content author (writer of the document’s text) and the style author (designer of the document’s layout and appearance) are the same. It has a consistent and easy to understand syntax that provides the author with the tools and freedom necessary to produce a document with any desired layout. In cases where there are no standards to follow, ConTeXt provides creative freedom at the expense of required additional effort. ConTeXt excels at producing high-quality works with creative flair, such as textbooks and literature with artistically distinctive layouts.

LaTeX A TeX macro-based document preparation system designed by Leslie Lamport.

LaTeX assumes the content author and style author are different people. This allows authors (researchers, students, etc.) to concentrate on content and forget about design while allowing publishers (journals, graduate departments, etc.) to enforce institutional standards. Separation of content and design comes with the costs of package management, a less consistent syntax, and added complexity (compared to ConTeXt) if an author wishes to deviate from the layout designer's specification (documentclass). LaTeX excels at producing high-quality academic documents that conform to publication requirements, such as journal articles and theses.

MetaFont A high-quality font system designed by Donald Knuth along with TeX.
MetaPost A descriptive vector graphics language based on MetaFont.
TeX The original language designed by Donald Knuth.
Texinfo A TeX macro-based document preparation system designed by Richard Stallman that specializes in producing technical documentation (software manuals).
Engines Descriptions
xetex, xelatex a TeX engine which supports Unicode input and .ttf and .otf fonts. See Fonts.
luatex, lualatex A TeX engine with embedded Lua support, aiming at making TeX internals more flexible. Like XeTeX, supports Unicode input and modern font files.
pdftex, pdflatex Generates PDF output.
tex, latex The "original" TeX engine. Generates DVI output.
TeX Distributions Descriptions
MacTeX A TeX Live based distribution targetting Mac OS X.
MiKTeX A TeX distribution for Windows.
TeX Live A cross-platform TeX distribution.

What's next?

In the next chapter we discuss installing LaTeX on your system. Then we will typeset our first LaTeX file.

Learning more

One of the most frustrating things beginners and even advanced users might encounter using LaTeX is the difficulty of changing the look of your documents. While WYSIWYG programs make it trivial to change fonts and layouts, LaTeX requires you to learn new commands and packages to do so. Subsequent chapters will cover many common use cases, but know that this book is only scratching the surface.

Coming from a community of typography enthusiasts, most LaTeX packages contain excellent documentation. This should be your first step if you have questions—if a package's manual has not been installed on your machine as part of your TeX distribution, it can be found on CTAN.

Other useful resources include:


If this is the first time you are trying out LaTeX, you don't even need to install anything. For quick testing purpose you may just create a user account with an online LaTeX editor such as Overleaf, and continue this tutorial in the next chapter. These websites offer collaborative editing capabilities while allowing you to experiment with LaTeX syntax — without having to bother with installing and configuring a distribution and an editor. When you later feel that you would benefit from having a standalone LaTeX installation, you can return to this chapter and follow the instructions below.

LaTeX is not a program by itself; it is a document preparation system along with a language. Using LaTeX requires a series of tools. Acquiring them manually would result in downloading and installing multiple programs in order to have a suitable computer system that can be used to create LaTeX output, such as PDFs. TeX Distributions help the user in this way, in that it is a single step installation process that provides (almost) everything.

At a minimum, you'll need a TeX distribution, a good text editor and a DVI or PDF viewer. More specifically, the basic requirement is to have a TeX compiler (which is used to generate output files from source), fonts, and the LaTeX macro set. Optional, and recommended installations include an attractive editor to write LaTeX source documents (this is probably where you will spend most of your time), and a bibliographic management program to manage references if you use them a lot.


TeX and LaTeX are available for most computer platforms, since they were programmed to be very portable. They are most commonly installed using a distribution, such as TeX Live, MiKTeX, or MacTeX. TeX distributions are collections of packages and programs (compilers, fonts, and macro packages) that enable you to typeset without having to manually fetch files and configure things. LaTeX is just a set of macro packages built for TeX.

The recommended distributions for each of the major operating systems are:

  • TeX Live is a major TeX distribution for *BSD, GNU/Linux, Mac OS X and Windows.
  • MiKTeX is a Windows-specific distribution which can be installed on Windows or GNU/Linux.
  • MacTeX is a Mac OS-specific distribution based on TeX Live.

These, however, do not necessarily include an editor. You might be interested in other programs that are not part of the distribution, which will help you in writing and preparing TeX and LaTeX files.

*BSD and GNU/Linux

In the past, the most common distribution used to be teTeX. As of May 2006 teTeX is no longer actively maintained and its former maintainer Thomas Esser recommended TeX Live as the replacement.[1]

The easy way to get TeX Live is to use the package manager or portage tree coming with your operating system. Usually it comes as several packages, with some of them being essential, other optional. The core TeX Live packages should be around 200-300 MB.

If your *BSD or GNU/Linux distribution does not have the TeX Live packages, you should report a wish to the bug tracking system. In that case you will need to download TeX Live yourself and run the installer by hand.

You may wish to install the content of TeX Live more selectively. See below.

Mac OS X

Mac OS X users may use MacTeX, a TeX Live-based distribution supporting TeX, LaTeX, AMSTeX, ConTeXt, XeTeX and many other core packages. Download MacTeX.pkg on the MacTeX page, unzip it and follow the instructions. Further information for Mac OS X users can be found on the TeX on Mac OS X Wiki.

Since Mac OS X is also a Unix-based system, TeX Live is naturally available through MacPorts and Fink. Homebrew users should use the official MacTeX installer because of the unique directory structure used by TeX Live. Further information for Mac OS X users can be found on the TeX on Mac OS X Wiki.

Microsoft Windows

Microsoft Windows users can install MiKTeX onto their computer. It has an easy installer that takes care of setting up the environment and downloading core packages. Both the basic and the complete LaTeX systems are provided, with the distribution offering advanced features such as automatic installation of packages and simple interfaces to modify settings (e.g., default paper sizes).[2]

There is also a port of TeX Live available for Windows. For more, see TeX Live on Windows.

Custom installation with TeX Live

This section targets users who want fine-grained control over their TeX distribution, like an installation with a minimum of disk space usage. If not needed, the user may feel free to jump to the next section.

Picky users may wish to have more control over their installation. Common distributions might be tedious for the user caring about disk space. In fact, MikTeX and MacTeX and packaged TeX Live features hundreds of LaTeX packages, most of them which you will never use. Most Unix with a package manager will offer TeX Live as a set of several big packages, and you often have to install 300–400 MB for a functional system.

TeX Live features a manual installation with a lot of possible customizations. You can get the network installer at This installer allows you to select precisely the packages you want to install. As a result, you may have everything you need for less than 100 MB. TeX Live is then managed through its own package manager, tlmgr. It will let you configure the distributions, install or remove extra packages and so on.

You will need a Unix-based operating system for the following. Mac OS X, GNU/Linux or *BSD are fine. It may work for Windows but the process must be quite different.

TeX Live groups features and packages into different concepts:

  • Collections are groups of packages that can always be installed individually, except for the Essential programs and files collection. You can install collections at any time.
  • Installation Schemes group collections and packages. Schemes can only be used at installation time. You can select only one scheme at a time.

Minimal installation

We will give you general guidelines to install a minimal TeX distribution (i.e., only for plain TeX).

  1. Download the installer at and extract it to a temporary folder.
  2. Open a terminal in the extracted folder and log in as root.
  3. Change the umask permissions to 022 (user read/write/execute, group/others read/execute only) to make sure other users will have read-only access to the installed distribution.
# umask 022
  1. Launch install-tl.
  2. Select the minimal scheme (plain only).
  3. You may want to change the directory options. For example you may want to hide your personal macro folder which is located at TEXMFHOME. It is ~/texmf by default. Replace it by ~/.texmf to hide it.
  4. Now the options:
    1. use letter size instead of A4 by default: mostly for users from the USA.
    2. allow execution of restricted list of programs via \write18: it is recommended to select it for security reasons. Otherwise it allows the TeX engines to call any external program. You may still configure the list afterwards.
    3. create all format files: targetting a minimal disk space, the best choice depends on whether there is only one user on the system, then deselecting it is better, otherwise select it. From the help menu: "If this option is set, format files are created for system-wide use by the installer. Otherwise they will be created automatically when needed. In the latter case format files are stored in user's directory trees and in some cases have to be re-created when new packages are installed."
    4. install macro/font doc tree: useful if you are a developer, but very space consuming. Turn it off if you want to save space.
    5. install macro/font source tree: same as above.
    6. create symlinks to standard directories: symlinks are fine by default, change it if you know what you are doing.
  5. Select portable installation if you install the distribution to an optical disc, or any kind of external media. Leave to default for a traditional installation on the system hard drive.

At this point it should display

1 collections out of 85, disk space required: 40 MB

or a similar space usage.

You can now proceed to installation: start installation to hard disk.

Don't forget to add the binaries to your PATH as it's noticed at the end of the installation procedure.

First test

In a terminal write

$ tex '\empty Hello world!\bye'
$ pdftex '\empty Hello world!\bye'

You should get a DVI or a PDF file accordingly.


Formerly, TeX distributions used to be configured with the texconfig tool from the teTeX distribution. TeX Live still features this tool, but recommends using its own tool instead: tlmgr. Tlmgr’s functionality completely subsumes texconfig.[1]

List current installation options:

tlmgr option

You can change the install options:

tlmgr option srcfiles 1
tlmgr option docfiles 0
tlmgr paper letter

See the TLMGR(1) man page for more details on its usage. If you did not install the documents as told previously, you can still access the tlmgr man page with

tlmgr help

Installing LaTeX

Now we have a running plain TeX environment, let's install the base packages for LaTeX.

# tlmgr install latex latex-bin latexconfig latex-fonts

In this case you can omit latexconfig latex-fonts as they are auto-resolved dependencies to LaTeX. Note that tlmgr resolves some dependencies, but not all. You may need to install dependencies manually. Thankfully this is rarely too cumbersome.

Other interesting packages:

# tlmgr install amsmath babel carlisle ec geometry graphics hyperref lm  marvosym oberdiek parskip graphics-def url
amsmath The essentials for math typesetting.
babel Internationalization support.
carlisle Bundle package required for some babel features.
ec Required for T1 encoding.
geometry For page layout.
graphics The essentials to import graphics.
htlatex Includes TeX4ht used in (LA )TeX to HTML (and XML and more) convertion.
hyperref PDF bookmarks, PDF followable links, link style, TOC links, etc.
lm One of the best Computer Modern style font available for several font encodings (such as T1).
marvosym Several symbols, such as the official euro.
oberdiek Bundle package required for some geometry features.
parskip Let you configure paragraph breaks and indents properly.
graphics-def Required for some graphics features.
url Required for some hyperref features.

If you installed a package you do not need anymore, use

# tlmgr remove <package>


If you are using Babel for non-English documents, you need to install the hyphenation patterns for every language you are going to use. They are all packaged individually. For instance, use

# tlmgr install hyphen-{finnish,sanskrit}

for finnish and sanskrit hyphenation patterns.

Note that if you have been using another TeX distribution beforehand, you may still have hyphenation cache stored in you home folder. You need to remove it so that the new packages are taken into account. The TeX Live cache is usually stored in the ~/.texliveYYYY folder (YYYY stands for the year). You may safely remove this folder as it contains only generated data. TeX compilers will re-generate the cache accordingly on next compilation.


By default TeX Live will install in /usr/local/texlive. The distribution is quite proper as it will not write any file outside its folder, except for the cache (like font cache, hyphenation patters, etc.). By default,

  • the system cache goes in /var/lib/texmf;
  • the user cache goes in ~/.texliveYYYY.

Therefore TeX Live can be installed and uninstalled safely by removing the aforementioned folders.

Still, TeX Live provides a more convenient way to do this:

# tlmgr uninstall

You may still have to wipe out the folders if you put untracked files in them.


TeX and LaTeX source documents (and its related auxiliary files) are all plain-text files, and can be opened and modified in almost any text editor. You should use a text editor (e.g. Notepad), not a word processor (e.g., Microsoft Word, LibreOffice Writer). Dedicated LaTeX editors are more useful than generic plain text editors, because they are usually equipped with the autocomplete feature for commands, spelling and error checking and other handy macros.


Microsoft Word can accept LaTeX through Equation Editor, but it is not a full-fledged LaTeX editor.



Emacs is a general purpose, extensible text processing system. Advanced users can program it (in elisp) to make Emacs the best LaTeX environment that will fit their needs. In turn beginners may prefer to use it in combination with AUCTeX and Reftex (extensions that may be installed into the Emacs program). Depending on the configuration, Emacs can provide a complete LaTeX editing environment with auto-completion, spell-checking, a complete set of keyboard shortcuts, view of table of contents, document preview and many other features.


Gedit with gedit-latex-plugin is also worth trying out for users of GNOME. GEdit is a cross-platform application for Windows, Mac, and Linux


Screenshot of Gummi.

Gummi is a LaTeX editor for Linux, which compiles the output of pdflatex in real-time and shows it on the right half of the screen[3].



LyX is a popular document preparation system for Windows, Linux and Mac OS. It provides a graphical interface to LaTeX, including several popular packages. It contains formula and table editors and shows visual clues of the final document on the screen — which enables users to write LaTeX documents without worrying about the actual syntax. LyX calls this a What You See Is What You Mean (WYSIWYM) approach, since the screen only shows the structure and an approximation of the output.[4]

LyX saves a document in its own markup, from which LaTeX code can then be generated. The user is mostly isolated from the LaTeX code and is not in complete control of it, and for that reason LyX is generally not considered as a proper LaTeX editor. However, since it uses LaTeX as its underlying system, knowledge of how LaTeX works can also be useful to a LyX user. In addition, if one wants to implement a feature that is not supported in the GUI, then the use of LaTeX code may be required.


TeXmaker is a cross-platform editor that is very similar to Kile in both features and user interface. It is also equipped with its own PDF viewer as well.


TeXstudio is a cross-platform open source LaTeX editor forked from Texmaker.


Screenshot of TeXworks on Ubuntu 12.10.

TeXworks is a dedicated TeX editor that is included in MiKTeX and TeX Live. It was developed with the idea that a simple interface is better than a cluttered one, and thus to make it easier for the beginners of LaTeX to write their own documents. TeXworks originally came about precisely because a math professor wanted his students to have a better initial experience with LaTeX.

You can install TeXworks with the package manager of your Linux distribution or choose it as an install option in the Windows or Mac installer.


Vim is another general purpose text editor for a wide variety of platforms including UNIX, Mac OS X and Windows. A variety of extensions exist including LaTeX Box and Vim-LaTeX.

*BSD and GNU/Linux-only


Screenshot of Kile.

Kile is a LaTeX editor for KDE (cross platform), providing a powerful GUI for editing multiple documents and compiling them with many different TeX compilers. Kile is based on Kate editor, has a quick access toolbar for symbols, document structure viewer, a console and customizable build options. Kile can be run in all operating systems that can run KDE.


GNOME-LaTeX is another text editor for Linux (GNOME).

Mac OS X-only


TeXShop, the model for the TeXworks editor and previewer, is for Mac OS and is bundled with the MacTeX distribution. It uses multiple windows, one for editing the source, one for the preview, and one as a console for error messages. It offers one-click updating of the preview and allows easy crossfinding between the code and the preview by using CMD-click along with many features to make editing and typesetting TeX source easier.


TeXnicle is a free editor for Mac OS that includes the ability to perform live updates. It includes a code library for the swift insertion of code and the ability to execute detailed word counts on documents. It also performs code highlighting and the editing window is customisable, permitting the user to select the font, colour, background colour of the editing environment. It is in active development.


Archimedes is an easy-to-use LaTeX and Markdown editor designed from the ground up for Mac OS X. It includes a built-in LaTeX library, code completion support, live previews, macro support, integration with sharing services, and PDF and HTML export options. Archimedes's Magic Type feature lets users insert mathematical symbols just by drawing them on their MacBook's trackpad or Magic Trackpad.


Texpad is an integrated editor and viewer for Mac OS with a companion app for iOS devices. Similar to TeXShop, Texpad requires a working MacTeX distribution to function, however it can also support other distributions side-by-side with MacTex. It offers numerous features including templates, outline viewing, auto-completion, spell checking, customizable syntax highlighting, to-do list integration, code snippets, Markdown integration, multi-lingual support, and a Mac OS native user interface. Although Texpad offers a free evaluation period, the unlocked version is a paid download.





TeXnicCenter is a popular free and open source LaTeX editor for Windows. It also has a similar user interface to TeXmaker and Kile.


WinEdt is a powerful and versatile text editor with strong predisposition towards creation of LaTeX/TeX documents for Windows. It has been designed and configured to integrate with TeX Systems such as MiTeX or TeX Live. Its built-in macro helps in compiling the LaTeX source to the WYSIWYG-like DVI or PDF or PS and also in exporting the document to other mark-up languages as HTML or XML.



Online solutions

To get started without needing to install anything, you can use a web-hosted service featuring a full TeX distribution and a web LaTeX editor.

  • Authorea is an integrated online framework for the creation of technical documents in collaboration. Authorea's frontend allows one to enter text in LaTeX or Markdown, as well as figures, and equations (in LaTeX or MathML). Authorea's versioning control system is entirely based on Git (as every article is a Git repository).
  • CoCalc is a collaborative online workplace for computations, which also offers an editor for LaTeX documents.
  • Overleaf is a secure, easy to use online LaTeX editor with integrated rapid preview - like Etherpad for LaTeX. One can start writing by creating a free account, and share the link or add collaborators to the projects before publishing it through their platform. It supports real time preview, Rich Text mode (a partial WYSIWYG mode with math expressions, ordered/unordered lists, sectional titles and figures in rendered form), bibliographies and custom styles. Since July 2017, ShareLaTeX is now part of Overleaf.[5][6]
  • Verbosus is a professional online LaTeX Editor that supports collaboration with other users and is free to use. Merge conflicts can easily resolved by using a built-in merge tool that uses an implementation of the diff-algorithm to generate information required for a successful merge.

Bibliography management

Bibliography files (*.bib) are most easily edited and modified using a management system. These graphical user interfaces all feature a database form, where information is entered for each reference item, and the resulting text file can be used directly by BibTeX.


Screenshot of JabRef.

Mac OS X-only

Screenshot of BibDesk
  • BibDesk is a bibliography manager based on a BibTeX file. It imports references from the internet and makes it easy to organize references using tags and categories[7].


Finally, you will need a viewer for the files to view LaTeX outputs. By default, LaTeX saves the final document as a .dvi (Device independent file format), but you will rarely want it to, as DVI files do not contain embedded fonts — not to mention that many document viewers are unable to open them.

In most scenarios, you will use a LaTeX compiler like pdflatex to produce a PDF file directly, or a tool like dvi2pdf to convert the DVI file to PDF format. Then you can view the result with any PDF viewer.

Practically all LaTeX distributions have a DVI viewer for viewing the default output of latex, and also tools such as dvi2pdf for converting the result automatically to PDF and PS formats.

The following is a list of the various PDF viewers available on the web:

Tables and graphics tools

LaTeX is a document preparation system above all else: it does not aim at being a spreadsheet tool nor a vector graphics tool.

If LaTeX can render beautiful tables in a dynamic and flexible manner, it will not handle the handy features you could get with a spreadsheet like dynamic cells and calculus. Other tools are better at that. The ideal solution is to combine the strength of both tools: build your dynamic table with a spreadsheet, and export it to LaTeX to get a beautiful table seamlessly integrated to your document. See Tables for more details.

The graphics topic is a bit different since it is possible to write procedural graphics from within your LaTeX document. Procedural graphics produce state-of-the-art results that integrates perfectly to LaTeX (e.g. no font change), but have a steep learning curve and require a lot of time to draw.

For easier and quicker drawings, you may want to use a WYSIWYG tool (e.g., Adobe Photoshop, Canva) and export the result to a vector format like PDF. The drawback is that it will contrast in style with the rest of your document (e.g., font, size, color). Some tools have the capability to export to LaTeX, which will partially solve this issue. See Importing Graphics for more details.


  1. a b teTeX Home Page (Retrieved August 22, 2020)
  2. MikTeX — All Downloads
  3. Gummi
  4. LyX
  5. ShareLaTeX Joins Overleaf
  6. The Definitive, Non-Technical Introduction to LaTeX: Overleaf
  7. BibDesk

Installing Extra Packages

Add-on features for LaTeX are known as packages. Dozens of these are pre-installed with LaTeX and can be used in your documents immediately. They should all be stored in subdirectories of texmf/tex/latex named after each package. The directory name "texmf" stands for “TEX and METAFONT”. To find out what other packages are available and what they do, you should use the CTAN search page which includes a link to Graham Williams' comprehensive package catalogue.

A package is a file or collection of files containing extra LaTeX commands and programming which add new styling features or modify those already existing. There are two main file types: class files with .cls extension, and style files with .sty extension. There may be ancillary files as well. When you try to typeset a document which requires a package which is not installed on your system, LaTeX will warn you with an error message that it is missing. You can download updates to packages you already have (both the ones that were installed along with your version of LaTeX as well as ones you added). There is no limit to the number of packages you can have installed on your computer (apart from disk space!), but there is a configurable limit to the number that can be used inside any one LaTeX document at the same time, although it depends on how big each package is. In practice there is no problem in having even a couple of dozen packages active.

Most LaTeX installations come with a large set of pre-installed style packages, so you can use the package manager of the TeX distribution or the one on your system to manage them. See the automatic installation. But many more are available on the net. The main place to look for style packages on the Internet is CTAN. Once you have identified a package you need that is not in your distribution, use the indexes on any CTAN server to find the package you need and the directory where it can be downloaded from. See the manual installation.

Automatic installation

If on an operating system with a package manager or a portage tree, you can often find packages in repositories.

With MikTeX there is a package manager that allows you to pick the package you want individually. As a convenient feature, upon the compilation of a file requiring non-installed packages, MikTeX will automatically prompt to install the missing ones.

With TeX Live, it is common to have the distribution packed into a few big packages. For example, to install something related to internationalization, you might have to install a package like texlive-lang. With TeX Live manually installed, use tlmgr to manage packages individually.

tlmgr install <package1> <package2> ...
tlmgr remove <package1> <package2> ...

The use of tlmgr is covered in the Installation chapter.

If you cannot find the wanted package with any of the previous methods, see the manual installation.

Instructions for specific operating systems

On Ubuntu, with releases such as Trusty, you can use texlive and texlive-extra packages, e.g. texlive-full, texlive-latex-extra, texlive-math-extra, texlive-plain-extra, texlive-bibtex-extra, texlive-generic-extra, and language packages, which are all available here on the Ubuntu packages site, as well as here for Trusty updates. You can install these packages with sudo apt-get install <insert package name here>.

Manual installation

Downloading packages

What you need to look for is usually two files, one ending in .dtx and the other in .ins. The first is a DOCTeX file, which combines the package program and its documentation in a single file. The second is the installation routine (much smaller). You must always download both files. If the two files are not there, it means one of two things:

  • Either the package is part of a much larger bundle which you shouldn't normally update unless you change LaTeX version of LaTeX;
  • or it's an older or relatively simple package written by an author who did not use a .dtx file.

Download the package files to a temporary directory. There will often be a readme.txt with a brief description of the package. You should of course read this file first.

Installing a package

There are five steps to installing a LaTeX package. (These steps can also be used on the pieces of a complicated package you wrote yourself; in this case, skip straight to Step 3.)

1. Extract the files Run LaTeX on the .ins file. That is, open the file in your editor and process it as if it were a LaTeX document (which it is), or if you prefer, type latex followed by the .ins filename in a command window in your temporary directory. This will extract all the files needed from the .dtx file (which is why you must have both of them present in the temporary directory). Note down or print the names of the files created if there are a lot of them (read the log file if you want to see their names again).

2. Create the documentation Run LaTeX on the .dtx file. You might need to run it twice or more, to get the cross-references right (just like any other LaTeX document). This will create a .dvi file of documentation explaining what the package is for and how to use it. If you prefer to create PDF then run pdfLaTeX instead. If you created a .idx as well, it means that the document contains an index, too. If you want the index to be created properly, follow the steps in the indexing section. Sometimes you will see that a .glo (glossary) file has been produced. Run the following command instead:

makeindex -s -o name.gls name.glo

3. Install the files While the documentation is printing, move or copy the package files from your temporary directory to the right place[s] in your TeX local installation directory tree. Packages installed by hand should always be placed in your "local" directory tree, not in the directory tree containing all the pre-installed packages. This is done to a) prevent your new package accidentally overwriting files in the main TeX directories; and b) avoid your newly-installed files being overwritten when you next update your version of TeX.

For a TDS(TeX Directory Structure)-conformant system, your "local installation directory tree" is a folder and its subfolders. The outermost folder should probably be called texmf-local/ or texmf/. Its location depends on your system:

  • MacTeX: Users/username/Library/texmf/.
  • Unix-type systems: Usually ~/texmf/. If you use TexMaker on Ubuntu 18 it may be in /usr/share/texmf/
  • MikTeX: Your local directory tree can be any folder you like, as long as you then register it as a user-managed texmf directory (see

The "right place" sometimes causes confusion, especially if your TeX installation is old or does not conform to the TeX Directory Structure(TDS). For a TDS-conformant system, the "right place" for a LaTeX .sty file is a suitably-named subdirectory of texmf/tex/latex/. "Suitably-named" means sensible and meaningful (and probably short). For a package like paralist, for example, I'd call the directory texmf/tex/latex/paralist.

Often there is just a .sty file to move, but in the case of complex packages there may be more, and they may belong in different locations. For example, new BibTeX packages or font packages will typically have several files to install. This is why it is a good idea to create a sub-directory for the package rather than dump the files into misc along with other unrelated stuff. If there are configuration or other files, read the documentation to find out if there is a special or preferred location to move them to.

Where to put files from packages
Type Directory (under texmf/ or texmf-local/) Description
.afm fonts/afm/foundry/typeface Adobe Font Metrics for Type 1 fonts
.bib bibtex/bib/bibliography BibTeX bibliography
.bst bibtex/bst/packagename BibTeX style
.cls tex/latex/base Document class file
.dvi doc package documentation
.enc fonts/enc Font encoding
.fd tex/latex/mfnfss Font Definition files for METAFONT fonts
.fd tex/latex/psnfss Font Definition files for PostScript Type 1 fonts
.map fonts/map Font mapping files
.mf fonts/source/public/typeface METAFONT outline
.pdf doc package documentation
.pfb fonts/type1/foundry/typeface PostScript Type 1 outline
.sty tex/latex/packagename Style file: the normal package content
.tex doc TeX source for package documentation
.tex tex/plain/packagename Plain TeX macro files
.tfm fonts/tfm/foundry/typeface TeX Font Metrics for METAFONT and Type 1 fonts
.ttf fonts/truetype/foundry/typeface TrueType font
.vf fonts/vf/foundry/typeface TeX virtual fonts
others tex/latex/packagename other types of file unless instructed otherwise

For most fonts on CTAN, the foundry is public.

4. Update your index Finally, run your TeX indexer program to update the package database. This program comes with every modern version of TeX and has various names depending on the LaTeX distribution you use. (Read the documentation that came with your installation to find out which it is, or consult

  • teTeX, TeX Live, fpTeX: texhash
  • web2c: mktexlsr
  • MacTeX: MacTeX appears to do this for you.
  • MikTeX: initexmf --update-fndb (or use the GUI)
  • MiKTeX 2.7 or later versions, installed on Windows XP through Windows 7: Start -> All Programs -> MikTex -> Settings. In Windows 8 use the keyword Settings and choose the option of Settings with the MiKTex logo. In Settings menu choose the first tab and click on Refresh FNDB-button (MikTex will then check the Program Files directory and update the list of File Name DataBase). After that just verify by clicking 'OK'.

5. Update font maps If your package installed any TrueType or Type 1 fonts, you need to update the font mapping files in addition to updating the index. Your package author should have included a .map file for the fonts. The map updating program is usually some variant on updmap, depending on your distribution:

  • TeX Live and MacTeX: updmap --enable (if you installed the files in a personal tree) or updmap-sys --enable (if you installed the files in a system directory).
  • MikTeX: Run initexmf --edit-config-file updmap, add the line "Map to the file that opens, then run initexmf --mkmaps.


The reason this process has not been automated widely is that there are still thousands of installations which do not conform to the TDS, such as old shared Unix systems and some Microsoft Windows systems, so there is no way for an installation program to guess where to put the files: you have to know this. There are also systems where the owner, user, or installer has chosen not to follow the recommended TDS directory structure, or is unable to do so for political or security reasons (such as a shared system where the user cannot write to a protected directory). The reason for having the texmf-local directory (called texmf.local on some systems) is to provide a place for local modifications or personal updates, especially if you are a user on a shared or managed system (Unix, Linux, VMS, Windows NT/2000/XP, etc.) where you may not have write-access to the main TeX installation directory tree. You can also have a personal texmf subdirectory in your own login directory. Your installation must be configured to look in these directories first, however, so that any updates to standard packages will be found there before the superseded copies in the main texmf tree. All modern TeX installations should do this anyway, but if not, you can edit texmf/web2c/texmf.cnf yourself.

Checking package status

The universal way to check if a file is available to TeX compilers is the command-line tool kpsewhich.

$ kpsewhich tikz

kpsewhich will actually search for files only, not for packages. It returns the path to the file. For more details on a specific package use the command-line tool tlmgr (TeX Live only):

tlmgr info <package>

The tlmgr tool has lot more options. To consult the documentation:

tlmgr help

Package documentation

To find out what commands a package provides (and thus how to use it), you need to read the documentation. In the texmf/doc subdirectory of your installation there should be directories full of .dvi files, one for every package installed. This location is distribution-specific, but is typically found in:

Distribution Path
MacTeX /Library/TeX/Documentation/texmf-doc/latex
MiKTeX %MIKTEX_DIR%\doc\latex
TeX Live $TEXMFDIST/doc/latex

Generally, most of the packages are in the latex subdirectory, although other packages (such as BibTeX and font packages) are found in other subdirectories in doc. The documentation directories have the same name of the package (e.g. amsmath), which generally have one or more relevant documents in a variety of formats (dvi, txt, pdf, etc.). The documents generally have the same name as the package, but there are exceptions (for example, the documentation for amsmath is found at latex/amsmath/amsdoc.dvi). If your installation procedure has not installed the documentation, the DVI files can all be downloaded from CTAN. Before using a package, you should read the documentation carefully, especially the subsection usually called "User Interface", which describes the commands the package makes available. You cannot just guess and hope it will work: you have to read it and find out.

You can usually automatically open any installed package documentation with the texdoc command:

texdoc <package-name>

External resources

The best way to look for LaTeX packages is the already mentioned CTAN: Search. Additional resources form The TeX Catalogue Online:

See Also


This tutorial is aimed at getting familiar with the bare bones of LaTeX.

Before starting, ensure you have LaTeX installed on your computer (see Installation for instructions of what you will need).

  • We will first have a look at the LaTeX syntax.
  • We will create our first LaTeX document.
  • Then we will take you through how to feed this file through the LaTeX system to produce quality output, such as postscript or PDF.
  • Finally we will have a look at the file names and types.

The LaTeX syntax

When using LaTeX, you write a plain text file which describes the document's structure and presentation. LaTeX converts this source text, combined with markup, into a typeset document. For the purpose of analogy, web pages work in a similar way: HTML is used to describe the document, which is then rendered into on-screen output - with different colours, fonts, sizes, etc. - by your browser.

You can create an input file for LaTeX with any text editor. A minimal example looks something like the following (the commands will be explained later):


Hello world!


LaTeX normalises spaces in its input files so that whitespace characters, such as a space or a tab, are treated uniformly as space. Several consecutive spaces are treated as one, space opening a line is generally ignored, and a single line break also yields space. More line breaks (empty lines) define the end of a paragraph. An example of applying these rules is presented below: the left-hand side shows the user's input (.tex), while the right-hand side depicts the rendered output (.dvi, .pdf, .ps).

It does not matter whether you
enter one or several             spaces
after a word.

An empty line starts a new

It does not matter whether you enter one or several spaces after a word.

An empty line starts a new paragraph.

Reserved Characters

The following symbols are reserved characters that either have a special meaning under LaTeX or are unavailable in all the fonts. If you enter them directly in your text, they will normally not print but rather make LaTeX do things you did not intend.

# $ % ^ & _ { } ~ \

As you will see, these characters can be used in your documents all the same by adding a prefix backslash:

\# \$ \% \^{} \& \_ \{ \} \~{} \textbackslash{}

In some circumstances, the square bracket characters [ ] can also be considered as reserved characters, as they are used to give optional parameters to some commands. If you want to print these directly after some command, like in this situation: \command [text] it will fail, as [text] will be considered as an option given to \command. You can achieve the correct output this way: \command {} [text].

The backslash character \ cannot be entered by adding another backslash in front of it, like so \\; this sequence is used for line breaking. For introducing a backslash in math mode, you can use \backslash instead.

The commands \~ and \^ produce respectively a tilde and a hat which is placed over the next letter. For example \~n gives ñ. That's why you need braces to specify there is no letter as argument. You can also use \textasciitilde and \textasciicircum to enter these characters; or other commands .

If you want to insert text that might contain several particular symbols (such as URIs), you can consider using the \verb command, which will be discussed later in the section on formatting. For source code, see Source Code Listings

The less-than < and greater-than > characters are the only visible ASCII characters (not reserved) that will not print correctly. See Special Characters for an explanation and a workaround.

Non-ASCII characters (e.g. accents, diacritics) can be typed in directly for most cases. However you must configure the document appropriately. The other symbols and many more can be printed with special commands as in mathematical formulae or as accents. We will tackle this issue in Special Characters.

LaTeX groups

Sometimes a certain state should be kept local, in other words its scope should be limited. This can be done by enclosing the part to be changed locally in curly braces. In certain occasions, using braces won't be possible. LaTeX provides \bgroup and \egroup to begin and end a group, respectively.

normal text {\itshape walzing \bfseries Wombat} more normal text

normal text \bgroup\itshape walzing \bfseries Wombat\egroup{} more normal text

Environments form an implicit group.

LaTeX environments

Environments in LaTeX have a role that is quite similar to commands, but they usually have effect on a wider part of the document. Their syntax is:

text to be influenced

Between the \begin and the \end you can put other commands and nested environments. The internal mechanism of environments defines a group, which makes its usage safe (no influence on the other parts of the document). In general, environments can accept arguments as well, but this feature is not commonly used and so it will be discussed in more advanced parts of the document.

Anything in LaTeX can be expressed in terms of commands and environments.

LaTeX commands

LaTeX commands are case sensitive, and take one of the following two formats:

  1. They start with a backslash \ and then have a name consisting of letters only.
    • Command names are terminated by a space, a number or any other non-letter.
  2. They consist of a backslash \ and exactly one non-letter.
    • Command names are terminated after that one non-letter.

Some commands need an argument, which has to be given between curly braces { } after the command name. Some commands support optional parameters, which are added after the command name in square brackets [ ]. The general syntax is:


Many LaTeX formatting commands come in pairs.

  1. An argument form command, where one of the arguments is the text to be formatted.
  2. A scope form command, where the formatting will be applied to all text after the command until the end of the current scope. That is, until the end of the current group or environment. This form may also be called a switch command. A scope form command might still have arguments, but the text to be formatted is not an argument. This form should almost never be called outside of any scope, otherwise it will apply on the rest of the document.

An argument form command will have one argument more than its corresponding scope form command, the extra argument being the text the command affects.


Emphasing text: \emph is an argument form command with one argument, the text to be emphasised. \em is the corresponding scope form command with no arguments.

\emph{emphasized text}, this part is normal % Correct.
{\em emphasized text}, this part is normal  % Correct.

\emph emphasized text, this part is normal  % Incorrect: command without argument.
\em{emphasized text}, this part is normal   % Incorrect: switch with argument.
\em emphasized text, this part is normal    % Dangerous: switch outside of any environment.

Coloring text: This example requires you to \usepackage{xcolor}. \textcolor is an argument form command with two arguments, the color and the text to be colored. \color is the corresponding scope form command with only one argument, the color.

By default, this text is black. \textcolor{red}{This is red text.} Back to black.

By default, this text is black. {\color{red}This is red text.} Back to black.


When LaTeX encounters a % character while processing an input file, it ignores the rest of the current line, the line break, and all whitespace at the beginning of the next line.

This can be used to write notes into the input file, which will not show up in the printed version.

This is an % stupid
% Better: instructive <----
example: Supercal%

This is an example: Supercalifragilisticexpialidocious

Note that the % character can be used to split long input lines that do not allow whitespace or line breaks, as with Supercalifragilisticexpialidocious above.

The core LaTeX language does not have a predefined syntax for commenting out regions spanning multiple lines. Refer to multiline comments for simple workarounds.

Our first document

Now we can create our first document. We will produce the absolute bare minimum that is needed in order to get some output; the well known Hello World! approach will be suitable here.

  • Open your favorite text-editor. vim, emacs, Notepad++, and other text editors will have syntax highlighting that will help to write your files.
  • Reproduce the following text in your editor. This is the LaTeX source.
% hello.tex - Our first LaTeX example!
Hello World!
  • Save your file as hello.tex.

When picking a name for your file, make sure it bears a .tex extension.

What does it all mean?

% hello.tex - Our first LaTeX example! The first line is a comment. This is because it begins with the percent symbol (%); when LaTeX sees this, it simply ignores the rest of the line. Comments are useful for people to annotate parts of the source file. For example, you could put information about the author and the date, or whatever you wish.
\documentclass{article} This line is a command and tells LaTeX to use the article document class. A document class file defines the formatting standard to follow, which in this case is the generic article format. Journals, university departments, etc. can provide these files to ensure publication standards are met. In many instances, the same document content can be reformatted for submission to a different publisher simply by substituting the required document class file. There are numerous generic document classes available to choose from if one is not provided.
\begin{document} This line is the beginning of the environment called document; it alerts LaTeX that content of the document is about to commence. Anything above this command is known generally to belong in the preamble.
Hello World! This was the only actual line containing real content - the text that we wanted displayed on the page.
\end{document} The document environment ends here. It tells LaTeX that the document source is complete, anything after this line will be ignored.

As we have said before, each of the LaTeX commands begins with a backslash (\). This is LaTeX's way of knowing that whenever it sees a backslash, to expect some commands. Comments are not classed as a command, since all they tell LaTeX is to ignore the line. Comments never affect the output of the document, provided there is no white space before the percent sign.

Building a document

We then feed our input file into a LaTeX engine, a program which generates our final document.

There are several LaTeX engines in modern use: lualatex, xelatex, and pdflatex. There are important differences between the three, but we'll discuss those elsewhere - any of them will work for building our first document.

Generating the document

LaTeX itself does not have a GUI, though some LaTeX installations feature a graphical front-end where you can click LaTeX into compiling your input file. Assuming you're not using one of those:

  1. Open a terminal and navigate to the directory containing your .tex file.
  2. Type the command: xelatex hello.tex (The .tex extension is not required, although you can include it if you wish.)
  3. Various bits of info about LaTeX and its progress will be displayed. If all went well, the last two lines displayed in the console will be:
Output written on hello.pdf (1 page).
Transcript written on hello.log.

This means that your source file has been processed and the resulting document is called hello.pdf. You can view it with any PDF viewer installed on your system.

In this instance, due to the simplicity of the file, you only need to run the LaTeX command once. However, if you begin to create complex documents, including bibliographies and cross-references, etc., LaTeX needs to be executed multiple times to resolve the references. This will be discussed in the future when it comes up.

Autobuild Systems

Compiling can be quite tricky as soon as you start working on more complex documents. A number of programs exist to automatically read in a LaTeX document and run the appropriate compilers the appropriate number of times. For example, latexmk can generate a PDF from most LaTeX files simply:

$ latexmk -pdf file.tex

Note that most editors will take care of it for you.

Historical versions of LaTeX

Both LaTeX and TeX were created many years before the Portable Document Format (PDF) existed, so the plain LaTeX engine, latex, emits DVI, a format designed by Donald Knuth for device-independent TeX output. This format has fallen out of general use, but can be converted into more common output formats using programs from your LaTeX distribution:

  • dvips converts .dvi files to .ps (PostScript).
  • dvipdf converts .dvi files to .pdf (dvipdfm is an improved version).

You might also find Ghostscript, a set of free and open-source tools for working with PostScript, useful. Its ps2pdf converts .ps files to .pdf, and pdf2ps does the reverse.

The following diagram shows the relationships between the LaTeX source code and the formats you can create from it:


The boxed red text represents the file formats, the blue text on the arrows represents the commands you have to use, the small dark green text under the boxes represents the image formats that are supported. Any time you pass through an arrow you lose some information, which might decrease the features of your document. Therefore, you should choose the shortest route to reach your target format. This is probably the most convenient way to obtain an output in your desired format anyway. Starting from a LaTeX source, the best way is to use only latex for a DVI output, or only pdflatex for a PDF output, converting to PostScript only when it is necessary to print the document.

Note that using latex to generate DVI output keeps you from using PDF-only features, such as hyperlinks and embedded fonts.

Chapter Export To Other Formats discusses more about exporting LaTeX source to other file formats.


Picking suitable filenames

Never, ever use directories (folders) or file names that contain spaces. Although your operating system probably supports them, some don't, and they will only cause grief and tears with TeX. Make filenames as short or as long as you wish, but strictly avoid spaces. Stick to lower-case letters without accents a-z, the digits 0-9, the hyphen (-), and only one full point or period (.) to separate the file extension (somewhat similar to the conventions for a good Web URL): it will let you refer to TeX files over the Web more easily and make your files more portable. Some operating systems do not distinguish between upper-case and lower-case letters, others do. Therefore it's best not to mix them.

Ancillary files

The TeX compilers are single-pass processes. It means that there is no way for a compiler to jump around the document, which would be useful for the table of contents and references. Indeed the compiler cannot guess at which page a specific section is going to be printed, so when the table of contents is printed before the upcoming sections, it cannot set the page numbers.

To circumvent this issue, many LaTeX commands which need to jump use ancillary files which usually have the same file name as the current document but a different extension. It stores temporary data into these files and use them for the next compilation. So to have an up-to-date table of contents, you need to compile the document twice. There is no need to re-compile if no section moved.

For example, the temporary file for the table of contents data is filename.toc.

None of these files contains unrecoverable information. It means you can delete them safely, compiling will regenerate them automatically.

When you work with various capabilities of LaTeX (index, glossaries, bibliographies, etc.) you will soon find yourself in a maze of files with various extensions and probably no clue. The following list explains the most common file types you might encounter when working with TeX:

Common file extensions in LaTeX
.aux A file that transports information from one compiler run to the next. Among other things, the .aux file is used to store information associated with cross-references.
.bbl Bibliography file output by BiBTeX and used by LaTeX
.bib Bibliography database file. (where you can store a list of full bibliographic citations)
.blg BiBTeX log file. (errors are logged here)
.bst BiBTeX style file.
.cls Class files define what your document looks like. They are selected with the \documentclass command.
.dtx Documented TeX. This is the main distribution format for LaTeX style files. If you process a .dtx file you get documented macro code of the LaTeX package contained in the .dtx file.
.ins The installer for the files contained in the matching .dtx file. If you download a LaTeX package from the net, you will normally get a .dtx and a .ins file. Run LaTeX on the .ins file to unpack the .dtx file.
.fd Font description file telling LaTeX about new fonts.
.dvi Device Independent File. This is the main result of a LaTeX compile run with latex. You can look at its content with a DVI previewer program or you can send it to a printer with dvips or a similar application.
.pdf Portable Document Format. This is the main result of a LaTeX compile run with pdflatex. You can look at its content or print it with any PDF viewer.
.log Gives a detailed account of what happened during the last compiler run.
.toc Stores all your section headers. It gets read in for the next compiler run and is used to produce the table of contents.
.lof This is like .toc but for the list of figures.
.lot And again the same for the list of tables.
.idx If your document contains an index. LaTeX stores all the words that go into the index in this file. Process this file with makeindex.
.ind The processed .idx file, ready for inclusion into your document on the next compile cycle.
.ilg Logfile telling what makeindex did.
.sty LaTeX Macro package. This is a file you can load into your LaTeX document using the \usepackage command.
.tex LaTeX or TeX input file. It can be compiled with latex.
.out hyperref package file, just one for the master file.

And what now?

Common Elements

See Document Structure and the Common Elements part for all the common features that belong to every type of document.

Non-English documents and special characters

LaTeX has some nice features for most languages in the world. You can tell LaTeX to follow typography rules of the target language, ease special characters input, and so on. See Special Characters and Internationalization.

Modular document

See Modular Documents for good recommendations about the way to organize big projects into multiple files.

Questions and Issues

We highly urge you to read the FAQ if you have issues about basic features, or if you want to read essential recommendations. For the more specific questions and issues, refer to the Tips and Tricks page.

Macros for the utmost efficiency

The full power of LaTeX resides in macros. They make your documents very dynamic and flexible. See the dedicated part.

Working in a team

See chapter Collaborative Writing of LaTeX Documents.

Document Structure

The main point of writing a text is to convey ideas, information, or knowledge to the reader. The reader will understand the text better if these ideas are well-structured, and will see and feel this structure much better if the typographical form reflects the logical and semantic structure of the content.

Given only the logical and semantical structure of a text, LaTeX derives the typographical form of the text according to the “rules” given in the document class file and in various style files. LaTeX allows users to structure their documents with a variety of hierarchical constructs, including chapters, sections, subsections and paragraphs.

Global structure

When LaTeX processes an input file, it expects it to follow a certain structure. Thus every input file must contain the commands



The area between \documentclass{...} and \begin{document} is called the preamble. It normally contains commands that affect the entire document.

After the preamble, the text of your document is enclosed between two commands which identify the beginning and end of the actual document:


You would put your text where the dots are. The reason for marking off the beginning of your text is that LaTeX allows you to insert extra setup specifications before it (where the blank line is in the example above: we'll be using this soon). The reason for marking off the end of your text is to provide a place for LaTeX to be programmed to do extra stuff automatically at the end of the document, like making an index.

A useful side-effect of marking the end of the document text is that you can store comments or temporary text underneath the \end{document} in the knowledge that LaTeX will never try to typeset them:



Document classes

When processing an input file, LaTeX needs to know which layout standard to use. Layouts standards are contained within 'class files' which have .cls as their filename extension.


Here, the class parameter for the command \documentclass specifies the .cls file to use for the document. It is recommended to put this declaration at the very beginning. The LaTeX distribution provides additional classes for other layouts, including letters and slides. It is also possible to create your own, as is often done by journal publishers, who simply provide you with their own class file, which tells LaTeX how to format your content. But we'll be happy with the standard article class for now. The options parameter customizes the behavior of the document class. The options have to be separated by commas.

Example: an input file for a LaTeX document could start with the line


which instructs LaTeX to typeset the document as an article with a base font size of 11 points, and to produce a layout suitable for double sided printing on A4 paper.

Here are some document classes that can be used with LaTeX:

Document Classes
article For articles in scientific journals, presentations, short reports, program documentation, invitations, ...
IEEEtran For articles with the IEEE Transactions format.
proc A class for proceedings based on the article class.
minimal It is as small as it can get. It only sets a page size and a base font. It is mainly used for debugging purposes.
report For longer reports containing several chapters, small books, thesis, ...
book For books.
slides For slides. The class uses big sans serif letters.
memoir For sensibly changing the output of the document. It is based on the book class, but you can create any kind of document with it [2]
letter For writing letters.
beamer For writing presentations (see LaTeX/Presentations).

Here is a comprehensive list of document classes.

The generic document classes that come with LaTeX offer some layout flexibility, which is why they have a lot of options in common. Non-generic classes (those provided by university departments or publication houses) may have different options than those shown below or no options at all. Normally, third-party classes come with their own documentation. The most common options for the generic document classes are listed in the following table:

Document Class Options
10pt, 11pt, 12pt Sets the size of the main font in the document. If no option is specified, 10pt is assumed.
a4paper, letterpaper,... Defines the paper size. The default size is letterpaper; However, many European distributions of TeX now come pre-set for A4, not Letter, and this is also true of all distributions of pdfLaTeX. Besides that, a5paper, b5paper, executivepaper, and legalpaper can be specified.
fleqn Typesets displayed formulas left-aligned instead of centered.
leqno Places the numbering of formulas on the left hand side instead of the right.
titlepage, notitlepage Specifies whether a new page should be started after the document title or not. The article class does not start a new page by default, while report and book do.
twocolumn Instructs LaTeX to typeset the document in two columns instead of one.
twoside, oneside Specifies whether double or single sided output should be generated. The classes article and report are single sided and the book class is double sided by default. Note that this option concerns the style of the document only. The option twoside does not tell the printer you use that it should actually make a two-sided printout.
landscape Changes the layout of the document to print in landscape mode.
openright, openany Makes chapters begin either only on right hand pages or on the next page available. This does not work with the article class, as it does not know about chapters. The report class by default starts chapters on the next page available and the book class starts them on right hand pages.
draft, final final is default. draft makes LaTeX indicate hyphenation and justification problems with a small square in the right-hand margin of the problem line so they can be located quickly by a human. It also suppresses the inclusion of images and shows only a frame where they would normally occur.

For example, if you want a report to be in 12pt type on A4, but printed one-sided in draft mode, you would use:



While writing your document, you will probably find that there are some areas where basic LaTeX cannot solve your problem. If you want to include graphics, colored text or source code from a file into your document, you need to enhance the capabilities of LaTeX. Such enhancements are called packages. Some packages come with the LaTeX base distribution. Others are provided separately. Modern TeX distributions come with a large number of packages pre-installed. The command to use a package is pretty simple: \usepackage:


command, where package is the name of the package and options is a list of keywords that trigger special features in the package. For example, to use the color package, which lets you typeset in colors, you would type:



You can pass several options to a package, each separated by a comma.


The document environment

Top matter

At the beginning of most documents there will be information about the document itself, such as the title and date, and also information about the authors, such as name, address, email etc. All of this type of information within LaTeX is collectively referred to as top matter. Although never explicitly specified (there is no \topmatter command) you are likely to encounter the term within LaTeX documentation.

A simple example:


\title{How to Structure a LaTeX Document}
\author{Andrew Roberts}
\date{December 2004}

The \title, \author, and \date commands are self-explanatory. You put the title, author name, and date in curly braces after the relevant command. The title and author are usually compulsory (at least if you want LaTeX to write the title automatically); if you omit the \date command, LaTeX uses today's date by default. You always finish the top matter with the \maketitle command, which tells LaTeX that it's complete and it can typeset the title according to the information you have provided and the class (style) you are using. If you omit \maketitle, the title will never be typeset.

Using this approach, you can only create a title with a fixed layout. If you want to create your title freely, see the Title Creation section. You should remember, however, that the goal of LaTeX is to leave formatting to the documentclass designer, and if you wish to submit your work to multiple publishers then you should avoid designing a custom title.


As most research papers have an abstract, there are predefined commands for telling LaTeX which part of the content makes up the abstract. This should appear in its logical order, therefore, after the top matter, but before the main sections of the body. This command is available for the document classes article and report, but not book.



Your abstract goes here...

By default, LaTeX will use the word "Abstract" as a title for your abstract. If you want to change it into anything else, e.g. "Executive Summary", add the following line before you begin the abstract environment:

\renewcommand{\abstractname}{Executive Summary}

Sectioning commands

The commands for inserting sections are fairly intuitive. Of course, certain commands are appropriate to different document classes. For example, a book has chapters but an article doesn't. Here are some of the structure commands found in simple.tex.

This chapter's content...

This section's content...

\subsection{Top Matter}
This subsection's content...

\subsubsection{Article Information}
This subsubsection's content...

Notice that you do not need to specify section numbers; LaTeX will sort that out for you. Also, for sections, you do not need to use \begin and \end commands to indicate which content belongs to a given block.

LaTeX provides 7 levels of depth for defining sections (see table below). Each section in this table is a subsection of the one above it.

Command Level Comment
\part{''part''} -1 not in letters
\chapter{''chapter''} 0 only books and reports
\section{''section''} 1 not in letters
\subsection{''subsection''} 2 not in letters
\subsubsection{''subsubsection''} 3 not in letters
\paragraph{''paragraph''} 4 not in letters
\subparagraph{''subparagraph''} 5 not in letters

All the titles of the sections are added automatically to the table of contents (if you decide to insert one). But if you make manual styling changes to your heading, for example a very long title, or some special line-breaks or unusual font-play, this would appear in the Table of Contents as well, which you almost certainly don't want. LaTeX allows you to give an optional extra version of the heading text which only gets used in the Table of Contents and any running heads, if they are in effect. This optional alternative heading goes in [square brackets] before the curly braces:

\section[Effect on staff turnover]{An analysis of the
effect of the revised recruitment policies on staff
turnover at divisional headquarters}

Section numbering

Numbering of the sections is performed automatically by LaTeX, so don't bother adding them explicitly, just insert the heading you want between the curly braces. Parts get roman numerals (Part I, Part II, etc.); chapters and sections get decimal numbering like this document, and appendices (which are just a special case of chapters, and share the same structure) are lettered (A, B, C, etc.).

You can change the depth to which section numbering occurs, so you can turn it off selectively. By default it is set to 3. If you only want parts, chapters, and sections numbered, not subsections or subsubsections etc., you can change the value of the secnumdepth counter using the \setcounter command, giving the depth level you wish. For example, if you want to change it to "1":


A related counter is tocdepth, which specifies what depth to take the Table of Contents to. It can be reset in exactly the same way as secnumdepth. For example:


To get an unnumbered section heading which does not go into the Table of Contents, follow the command name with an asterisk before the opening curly brace:


All the divisional commands from \part* to \subparagraph* have this "starred" version which can be used on special occasions for an unnumbered heading when the setting of secnumdepth would normally mean it would be numbered.

If you want the unnumbered section to be in the table of contents anyway, use package unnumberedtotoc [1]. It provides the command


which will take care of a proper header as well. \addpart and \addchap are also available. KOMA classes provide those commands by default.

If you don't want to use package unnumberedtotoc, you have to do everything by hand using \addcontentsline and \markright{} (or even \markboth{}{}).


Note that if you use PDF bookmarks you will need to add a phantom section so that hyperlinks will lead to the correct place in the document. The \phantomsection command is defined in the hyperref package, and is Commonly used like this:


For chapters you will also need to clear the page (this will also correct page numbering in the ToC):

\clearpage %or \cleardoublepage
\addcontentsline{toc}{chapter}{List of Figures}

Section number style

See Counters.

Ordinary paragraphs

Paragraphs of text come after section headings. Simply type the text and leave a blank line between paragraphs. The blank line means "start a new paragraph here": it does not mean you get a blank line in the typeset output. For formatting paragraph indents and spacing between paragraphs, refer to the Paragraph Formatting section.

Table of contents

All auto-numbered headings get entered in the Table of Contents (ToC) automatically. You don't have to print a ToC, but if you want to, just add the command \tableofcontents at the point where you want it printed (usually after the Abstract or Summary).

Entries for the ToC are recorded each time you process your document, and reproduced the next time you process it, so you need to re-run LaTeX one extra time to ensure that all ToC pagenumber references are correctly calculated. We've already seen how to use the optional argument to the sectioning commands to add text to the ToC which is slightly different from the one printed in the body of the document. It is also possible to add extra lines to the ToC, to force extra or unnumbered section headings to be included.

The commands \listoffigures and \listoftables work in exactly the same way as \tableofcontents to automatically list all your tables and figures. If you use them, they normally go after the \tableofcontents command. The \tableofcontents command normally shows only numbered section headings, and only down to the level defined by the tocdepth counter, but you can add extra entries with the \addcontentsline command. For example if you use an unnumbered section heading command to start a preliminary piece of text like a Foreword or Preface, you can write:


This will format an unnumbered ToC entry for "Preface" in the "subsection" style. You can use the same mechanism to add lines to the List of Figures or List of Tables by substituting lof or lot for toc. If the hyperref package is used and the link does not point to the correct chapter, the command \phantomsection in combination with \clearpage or \cleardoublepage can be used (see also Labels and Cross-referencing):

\addcontentsline{toc}{chapter}{List of Figures}

To change the title of the ToC, you have to paste this command \renewcommand{\contentsname}{<New table of contents title>} in your document preamble. The List of Figures (LoF) and List of Tables (LoT) names can be changed by replacing the \contentsname with \listfigurename for LoF and \listtablename for LoT.


The default ToC will list headings of level 3 and above. To change how deep the table of contents displays automatically the following command can be used in the preamble:


This will make the table of contents include everything down to paragraphs. The levels are defined above on this page. Note that this solution does not permit changing the depth dynamically.

You can change the depth of specific section type, which could be useful for PDF bookmarks (if you are using the hyperref package) :

\renewcommand*{\toclevel@chapter}{-1} % Put chapter depth at the same level as \part.
\renewcommand*{\toclevel@chapter}{0} % Put chapter depth back to its default value.

In order to further tune the display or the numbering of the table of contents, for instance if the appendix should be less detailed, you can make use of the tocvsec2 package (CTAN, doc).

Book structure

The standard LaTeX book class follows the same layout described above with some additions. By default a book will be two-sided, i.e. left and right margins will change according to the page number parity. Furthermore current chapter and section will be printed in the header.

If you do not make use of chapters, it is barely useful to use the book class.

Additionally the class provides macros to change the formatting of some places of the document. We will give you some advice on how to use them properly.[2]



% Introductory chapters
% ...

\chapter{First chapter}
% ...

\chapter{First Appendix}

\chapter{Last note}
  • The frontmatter chapters will not be numbered. Page numbers will be printed in roman numerals. Frontmatter is not supposed to have sections, so they will be numbered 0.n because there is no chapter numbering. Check the Counters chapter for a fix.
  • The mainmatter chapters works as usual. The command resets the page numbering. Page numbers will be printed in arabic numerals.
  • The \appendix macro can be used to indicate that following sections or chapters are to be numbered as appendices. Appendices can be used for the article class too:
\section{First Appendix}

Only use the \appendix macro once for all appendices.

  • The backmatter behaves like the frontmatter. It has the same issue with section numbering.

As a general rule you should avoid mixing the command order. Nonetheless all commands are optional, so you might consider using only a few.

Note that the special content like the table of contents is considered as an unnumbered chapter.

Page order

This is one traditional page order for books.

  1. Half-title
  2. Empty
  3. Title page
  4. Information (copyright notice, ISBN, etc.)
  5. Dedication if any, else empty
  6. Table of contents
  7. List of figures (can be in the backmatter too)
  8. Preface chapter
  1. Main topic
  1. Some subordinate chapters
  1. Bibliography
  2. Glossary / Index

Special pages

Comprehensive papers often feature special pages at the end, like indices, glossaries and bibliographies. Since this is quite a complex topic, we will give you details in the dedicated part Special Pages.


Any good research paper will have a complete list of references. LaTeX has two ways of inserting your references into a document:

  • you can embed them within the document itself. It's simpler, but it can be time-consuming if you are writing several papers about similar subjects so that you often have to cite the same books.
  • you can store them in an external BibTeX file and then link them via a command to your current document and use a BibTeX style to define how they appear. This way you can create a small database of the references you might use and simply link them, letting LaTeX work for you.

To learn how to add a bibliography to your document, see the Bibliography Management section.

Notes and references

Text Formatting

This section will guide you through text-formatting techniques. Formatting refers to most things to do with appearance including text style and spacing. Formatting may also refer to paragraph and page layout, here we will focus on the customization of words and sentences.

Writers use formatting techniques to differentiate textual elements from the rest of the text. The many ways in which writers wish to differentiate textual elements give rise to many formatting techniques. Italicization is often used to add emphasis to key words or phrases. Footnotes are useful for providing extra information or clarification without interrupting the main flow of the text. For these reasons, formatting is very important. However, it is also very easy to abuse, and a document that has been overdone can look and read worse than one with none at all.

LaTeX is so flexible that we will actually only skim the surface, as you can have much more control over the presentation of your document if you wish. Having said that, one of the purposes of LaTeX is to take away the stress of having to deal with the physical presentation yourself, so you need not get too carried away!


Line Spacing

If you want to use larger inter-line spacing in a document, you can change its value by putting the


command into the preamble of your document. Use \linespread{1.3} for "one and a half" line spacing, and \linespread{1.6} for "double" line spacing. Normally the lines are not spread, so the default line spread factor is 1. This may not be ideal in all situations: see .

The setspace package allows more fine-grained control over line spacing. To set "one and a half" line spacing document-wide, but not where it is usually unnecessary (e.g. footnotes, captions):


To change line spacing within the document, the setspace package provides the environments singlespace, onehalfspace, doublespace and spacing:

This paragraph has \\ default \\ line spacing.
  This paragraph has \\ double \\ line spacing.
  This paragraph has \\ huge gaps \\ between lines.

Non-breaking spaces

This essential feature is a bit unknown to newcomers, although it is available on most WYSIWYG document processors. A non-breaking space between two tokens (e.g. words, punctuation marks) prevents the processors from inserting a line break between them. Additionally, a non-breaking space cannot be enlarged. It is very important for a consistent reading.

LaTeX uses the '~' symbol as a non-breaking space. You would usually use non-breaking spaces for punctuation marks in some languages, for units and currencies, for initials, etc. In French typography, you would put a non-breaking space before all two-parts punctuation marks.



Sentence-spacing -- space between words and sentences

To get a straight right margin in the output, LaTeX inserts varying amounts of space between the words. By default, it follows traditional typesetting practice and inserts more space at the end of a sentence to assist the reader: an em-space rather than a word-space. This can be approximated on a typewriter or computer keyboard by pressing the space bar twice rather than once. After the invention of the typewriter, English practice was to press the spacebar twice between sentences (but not around various other punctuation), while French practice was to press the spacebar once (but then also again around various other punctuation).

The extra space added at the end of sentences is latterly considered typographically old-fashioned in current mass-market English language printing. The practice was discovered in the fifteenth century to be preferred by readers and remained the standard throughout the fifteenth to twentieth centuries. Typesetting technology changes around WWII encouraged mass-market publishers to increasingly use single spacing for cost/profit reasons, and in the late 1980s the then-innovative Macintosh DTP technology created a suddenly wide platform for the University of Chicago Press to reintroduce and evangelize William Morris's failed nineteenth century attempt to popularize the rejected early fifteenth century's close-set type as a Norm, but this time to people without industry or market knowledge. "French Spacing" was redefined at this time in the USA (only) as a pejorative term to describe its opposite: to mean em-spacing between sentences rather than word-spacing. LaTeX users should note that Donald Knuth created TeX specifically to correct what he described as the ugliness and unreadability of then-(70s, pre-MacintoshDTP)-machine-typesetting, that he later added the FrenchSpacing option not as default but as a lengthily-specified option, i.e. that original TeX assumed the user wanted reader-centric typesetting (em-spacing sentences not word-spacing), and that Donald Knuth explicitly coded the FrenchSpacing option to implement precisely the opposite of what current post-University-of-Chicago typographers declare it to mean despite him being "otherwise" regarded as extremely intelligent and extremely well informed.) Most modern typesetters treat the end of sentence space the same as the interword space. (See for example, Bringhurst's Elements of Typographic Style.)

The additional space after periods can be disabled with the command


which tells LaTeX not to insert more space after a period than after ordinary character. Frenchspacing can be turned off later in your document via the \nonfrenchspacing command.

If an author wishes to use the wider end-of-sentence spacing, care must be exercised so that punctuation marks are not misinterpreted as ends of sentences. TeX assumes that sentences end with periods, question marks or exclamation marks. Although if a period follows an uppercase letter, this is not taken as a sentence ending, since periods after uppercase letters normally occur in abbreviations. Any exception from these assumptions has to be specified by the author. A backslash before a space generates a space that will not be enlarged. A tilde ‘~’ character generates a non-breaking space. The command \@ before a period specifies that this period terminates a sentence even when it follows an uppercase letter. (If you are using \frenchspacing, then none of these exceptions need be specified.)

Stretched spaces

You can insert a horizontal stretched space with \hfill in a line so that the rest gets "pushed" toward the right margin. For instance this may be useful in the header.

Author Name \hfill \today

Similarly you can insert vertical stretched space with \vfill. It may be useful for special pages.


\section{My first section}
% ...

See Lengths for more details.

Manual spacing

The spaces between words and sentences, between paragraphs, sections, subsections, etc. is determined automatically by LaTeX. It is against LaTeX philosophy to insert spaces manually and will usually lead to bad formatting. Manual spacing is a matter of macro writing and package creation.

See Lengths for more details.


LaTeX hyphenates words whenever necessary. Hyphenation rules will vary for different languages. LaTeX supports only English by default, so if you want to have correct hyphenation rules for your desired language, see Internationalization.

If the hyphenation algorithm does not find the correct hyphenation points, you can remedy the situation by using the following commands to tell TeX about the exception. The command

\hyphenation{word list}

causes the words listed in the argument (separated by blanks) to be hyphenated only at the points marked by “-”. The argument of the command should only contain words built from normal letters, or rather characters that are considered to be normal letters by LaTeX. It is known that the hyphenation algorithm does not find all correct American English hyphenation points for several words. A log of known exceptions is published periodically in the TUGboat journal. (2012 list:

The hyphenation hints are stored for the language that is active when the hyphenation command occurs. This means that if you place a hyphenation command into the preamble of your document it will influence the English language hyphenation. If you place the command after the \begin{document} and you are using some package for national language support like babel, then the hyphenation hints will be active in the language activated through babel. The example below will allow “hyphenation” to be hyphenated as well as “Hyphenation”, and it prevents “FORTRAN”, “Fortran” and “fortran” from being hyphenated at all. No special characters or symbols are allowed in the argument. Example:

\hyphenation{FORTRAN Hy-phen-a-tion}

With babel, the recommended command to set hyphenation exceptions is \babelhyphenation. When LuaTeX is used, babel also allows to add new patterns and modify existing ones (with \babelpatterns), as well as to define non-standard rules (like ‘ff’ to ‘ff-f’ in some languages, or ranked hyphenation) to be applied without explicit mark-up (with \babelposthyphenation).

The command \- inserts a discretionary hyphen into a word. This also becomes the only point where hyphenation is allowed in this word. This command is especially useful for words containing special characters (e.g., accented characters), because LaTeX does not automatically hyphenate words containing special characters.

I think this is: su\-per\-cal\-%


LaTeX does not hyphenate compound words that contain a dash[1]. There are two packages that can add back flexibility. The hyphenat package supplies the \hyp command. This command typesets the dash and then subjects the constituent words to automatic hyphenation. After loading the package:


one should write, instead of electromagnetic-endioscopy:


The extdash package also offers features for controlling the hyphenation of compound words containing dashes — as opposed to the words themselves which it leaves to LaTeX. The shortcuts option enables a more compressed syntax:


Typical usage is as follows, assuming the compressed syntax. In both cases, LaTeX can break and hyphenate the constituent words, but in the latter case, it will not break after the L:


One or more words can be kept together on the one line with the standard LaTeX command:


This prevents hyphenation and causes its argument to be kept together under all circumstances. For example:

My phone number will change soon. It will be \mbox{0116 291 2319}.

\fbox is similar to \mbox, but in addition there will be a visible box drawn around the content.

To avoid hyphenation altogether, the penalty for hyphenation can be set to an extreme value:


You can change the degree to which LaTeX will hyphenate by changing the value of \tolerance=1000 and \hyphenpenalty=1000. You'll have to experiment with the values to achieve the desired effect. A document which has a low tolerance value will cause LaTeX not to tolerate uneven spacing between words, hyphenating words more frequently than in documents with higher tolerances. Also note that using a higher text width will decrease the probability of encountering badly hyphenated word. For example adding


will widen the text width and reduce the amount of margin overruns.


LaTeX treats left and right quotes as different entities. For single quotes, a grave accent, ` (on American keyboards, this symbol is found on the tilde key; adjacent to the number 1 key on most keyboards) gives a left quote mark, and an apostrophe, ' gives a right. For double quotes, simply double the symbols, and LaTeX will interpret them accordingly. (Don't use the " for right double quotes: when the babel package is used for some languages (e.g. German), the " is redefined to produce an umlaut accent; using " for right double quotes will either lead to bad spacing or it being used to produce an umlaut). On British keyboards, ' ` ' is left of the ' 1 ' key and shares the key with ' ¬ ', and sometimes ' ¦ ' or ' | '. The apostrophe (') key is to the right of the colon/semicolon key and shares it with the ' @ ' symbol.

To `quote' in LaTeX


To ``quote'' in LaTeX


To ``quote" in LaTeX


To ,,quote'' in LaTeX


,,German quotation marks``


<<French quotation marks>>


``Please press the `x' key.''


,,Proszę, naciśnij klawisz <<x>>''.


The right quote is also used for apostrophe in LaTeX without trouble.

For left bottom quote and European quoting style you need to use T1 font encoding enabled by:


See Fonts for more details on font encoding.

The package csquotes offers a multilingual solution to quotations, with integration to citation mechanisms offered by BibTeX. This package allows one for example to switch languages and quotation styles according to babel language selections.

Diacritics and accents

Most accents and diacritics may be inserted with direct keyboard input by configuring the preamble properly. For symbols unavailable on your keyboard, diacritics may be added to letters by placing special escaped metacharacters before the letter that requires the diacritic.

See Special Characters.

Margin misalignment

Some very long words, numbers or URLs may not be hyphenated properly and move far beyond the side margin. One solution for this problem is to use sloppypar environment, which tells LaTeX to adjust word spacing less strictly. As a result, some spaces between words may be a bit too large, but long words will be placed properly.

This is a paragraph with
then we have another bad thing
--- a long number 1234567890123456789.

This is a paragraph with
then we have another bad thing
--- a long number 1234567890123456789.


Another solution is to edit the text to avoid long words, numbers or URLs approaching the side margin.


Some letter combinations are typeset not just by setting the different letters one after the other, but by actually using special symbols (like "ff"), called ligatures. Ligatures can be prohibited by inserting {} or, if this does not work, {\kern0pt} between the two letters in question. This might be necessary with words built from two words. A classic example is shelfful:[2]

\Large Not shelfful\\
but shelf{}ful


If you are using LuaLaTeX, you can automate some of this work with the selnolig package.

Slash marks

When the slash character / is immediately preceded and/or followed by text without intervening white space, LaTeX does not allow a line break to occur between the slash and adjacent text. This behaviour was implemented because historically most occurrences of the slash within text were situations where the convention was to keep the surrounding text together on one line. Examples include

  • Dates, such as "1/1/2021".
  • Abbreviations of units, such as "mm/year".

Dates and unit abbreviations are relatively short, so the constraint that they must be kept together on a single line does not cause issues. However, the / character can also occur in longer expressions which might create "overfull" errors in output, causing text to overflow a margin. In these situations the / character should be replaced by "\slash", which allows the line to "break" after the slash mark if needed. Examples include

  • Words separated by slashes, such as "input/output" which should be typeset as "input\slash output", or Yes/No/Cancel. (Shorter examples like and/or could be safely left as not allowing a line break.)
  • Directory names in operating systems that use the slash as a separator, such as /home/texlion/Documents, which should be typeset as "/home\slash texlion\slash Documents". Note that the first of the 3 slashes is NOT replaced.

A word after / or \slash is not automatically hyphenated. This is a similar problem to non-hyphenation of words with a dash described under Hyphenation. One way to have both a line break and automatic hyphenation in both words is


Both / and \slash can be used with a zero \hspace like this. \slash includes a penalty to make a line break there less desirable. This combination can be made into a new slash macro if desired. The hyphenat package includes an \fshyp which will add a hyphen after the slash like "input/- output" if the line breaks there.


To change the font family, emphasize text, and other font-related issues, see Fonts.

Formatting macros

Even if you can easily change the output of your fonts using those commands, you're better off not using explicit commands like this, because they work in opposition to the basic idea of LaTeX, which is to separate the logical and visual markup of your document. This means that if you use the same font changing command in several places in order to typeset a special kind of information, you should use \newcommand to define a "logical wrapper command" for the font changing command.


Do not \oops{enter} this room,
it’s occupied by \oops{machines}
of unknown origin and purpose.

Do not enter this room, it’s occupied by machines of unknown origin and purpose.

This approach has the advantage that you can decide at some later stage that you want to use some visual representation of danger other than \textit, without having to wade through your document, identifying all the occurrences of \textit and then figuring out for each one whether it was used for pointing out danger or for some other reason.

See Macros for more details.

Text mode superscript and subscript

Sub and superscripting can be done quite easily using \textsubscript{} and \textsuperscript{}.


Michelangelo was born on March 6\textsuperscript{th}, 1475.


Note: A LaTeX version from 2015 or later, or the package fixltx2e, is needed to use text-mode subscripts in all contexts.[3]

Text figures ("old style" numerals)

Many typographers prefer to use titling figures, sometimes called lining figures, when numerals are interspersed with full caps, when they appear in tables, and when they appear in equations, using text figures elsewhere. LaTeX allows this usage through the \oldstylenums{} command:


Some fonts do not have text figures built in; the textcomp package attempts to remedy this by effectively generating text figures from the currently-selected font. Put \usepackage{textcomp} in your preamble. textcomp also allows you to use decimal points, properly formatted dollar signs, etc. within \oldstylenums{}.

One common use for text figures is in section, paragraph, and page numbers. These can be set to use text figures by placing some code in your preamble:


% Enclose everything in an \AtBeginDocument{}
  % Make \section{} use text figures
  \renewcommand{\thesection}{ \oldstylenums{\myTheSection} }

  % Make \paragraph{} use text figures
  \renewcommand{\theparagraph}{ \oldstylenums{\myTheParagraph} }

  % Make the page numbers in text figures
  \renewcommand{\thepage}{ \oldstylenums{\myThePage} }

Should you use additional sectioning or paragraphing commands, you may adapt the previous code listing to include them as well.


A subsequent use of the \pagenumbering command, e.g., \pagenumbering{arabic}, will reset the \thepage command back to the original. Thus, if you use the \pagenumbering command in your document, be sure to reinstate your \myThePage definition from the code above:

% without this, the \thepage command will not be in oldstyle (e.g., in your Table of Contents}
\renewcommand{\thepage}{ \oldstylenums{\myThePage} }

Dashes and hyphens

LaTeX knows four kinds of dashes: a hyphen (-), en dash (–), em dash (—), or a minus sign (−). You can access three of them with different numbers of consecutive dashes. The fourth sign is actually not a dash at all—it is the mathematical minus sign:

Hyphen: daughter-in-law, X-rated\\
En dash: pages 13--67\\
Em dash: yes---or no? \\
Minus sign: $0$, $1$ and $-1$


The names for these dashes are: ‘-’(-) hyphen , ‘--’(–) en-dash , ‘---’(—) em-dash and ‘ ’(−) minus sign. They have different purposes:

Input Output Purpose
- - inter-word
-- page range, 1–10
--- punctuation dash—like this
$-$ minus sign

Use \hyp{} macro from hyphenat package instead of hyphen if you want LaTeX to break compound words between lines.

The commands \textendash and \textemdash are also used to produce en-dash (–), and em-dash (—), respectively.

Ellipsis (…)

A sequence of three dots is known as an ellipsis, which is commonly used to indicate omitted text. On a typewriter, a comma or a period takes the same amount of space as any other letter. In book printing, these characters occupy only a little space and are set very close to the preceding letter. Therefore, you cannot enter ‘ellipsis’ by just typing three dots, as the spacing would be wrong. Instead, there is a special command for these dots. It is called \ldots:

Not like this ... but like this:\\
New York, Tokyo, Budapest, \ldots


Alternatively, you can use the \textellipsis command which allows the spacing between the dots to vary.

Ready-made strings

There are some very simple LaTeX commands for typesetting special text strings:



To do:
Describe microtype package -- hanging punctuation, font expansion, additional kerning.

Notes and References

  1. hyphenat package documentation, p3
  2. Knuth, Donald. "Chapter 5: Grouping". The TeXbook. p. 19.

This page uses material from Andy Roberts' Getting to grips with LaTeX with permission from the author.

Paragraph Formatting

Altering the paragraph formatting is rarely necessary in academic writing. It is primarily used for formatting text in floats or for more exotic documents.

Paragraph alignment

Paragraphs in LaTeX are usually fully justified, i.e. flush with both the left and right margins. For whatever reason, should you wish to alter the justification of a paragraph, there are three environments at hand, and also LaTeX command equivalents.

Alignment Environment Command
Left justified flushleft \raggedright
Right justified flushright \raggedleft
Center center \centering

All text between the \begin and \end of the specified environment will be justified appropriately. The commands listed are for use within other environments. For example, p (paragraph) columns in tabular.

However, if you really need to disable one of the above commands locally (for example because you have to use some broken package), you can use the command \justifying from package ragged2e.

Paragraph indent

By default, the first paragraph after a heading follows the standard Anglo-American publishers' practice of no indentation. The size of subsequent paragraph indents is determined by a parameter called \parindent. The default length that this constant holds is set by the document class that you use. It is possible to override it by using the \setlength command. This will set paragraph indents to 1cm:

\setlength{\parindent}{1cm} % Default is 15pt.

Whitespace in LaTeX can also be made flexible (what Lamport calls "rubber" lengths). This means that values such as extra vertical space inserted before a paragraph \parskip can have a default dimension plus an amount of expansion minus an amount of contraction. This is useful on pages in complex documents where not every page may be an exact number of fixed-height lines long, so some give-and-take in vertical space is useful. You specify this in a \setlength command like this:

\setlength{\parskip}{1cm plus 4mm minus 3mm}

If you want to indent a paragraph that is not indented, you can use


at the beginning of the paragraph. Obviously, this will only have an effect when \parindent is not set to zero. If you want to indent the beginning of every section, you can use the indentfirst package: once loaded, the beginning of any chapter/section is indented by the usual paragraph indentation.

To create a non-indented paragraph, you can use


as the first command of the paragraph. This might come in handy when you start a document with body text and not with a sectioning command.

Be careful, however, if you decide to set the indent to zero, then it means you will need a vertical space between paragraphs in order to make them clear. The space between paragraphs is held in \parskip, which could be altered in a similar fashion as above. However, this parameter is used elsewhere too, such as in lists, which means you run the risk of making various parts of your document look very untidy by changing this setting. If you want to use the style of having no indentation with a space between paragraphs, use the parskip package, which does this for you, while making adjustments to the spacing of lists and other structures which use paragraph spacing, so they don't get too far apart. If you want both indent and break, use


To indent subsequent lines of a paragraph, use the TeX command \hangindent. (While the default behaviour is to apply the hanging indent after the first line, this may be changed with the \hangafter command.) An example follows.

\hangindent=0.7cm This paragraph has an extra indentation at the left.

The TeX commands \leftskip and \rightskip add additional space to the left and right sides of each line, allowing the formatting for subsequent paragraphs to differ from the overall document margins. This space is in addition to the indentation added by \parindent and \hangindent.

To change the indentation of the last line in a paragraph, use the TeX command \parfillskip.

Paragraph line break

Default style for \paragraph may seem odd in the first place, as it writes the following text next to the title. If you do not like it, use a class other than the traditional article/book, or use ConTeXt or PlainTeX. Hacking of the class in use is really not the way LaTeX is intended to be used, and you may encounter a lot of frustrating issues.

Anyway, let's analyse the problem. If you add a manual line break with \\, LaTeX will complain that

There's no line here to end.

Simply adding an empty space will do it:

\paragraph{Title} \hspace{0pt} \\

Alternatively, you can use the shorter, yet not completely equivalent syntax:

\paragraph{Title} ~\\

You can also change ${}$ (an empty inline mathematical expression) for \hspace{0pt} or ~ in the above makeshift.

Line spacing

To change line spacing in the whole document use the command \linespread covered in Text Formatting.

Alternatively, you can use the \usepackage{setspace} package, which is also covered in Text Formatting. This package provides the commands \doublespacing, \onehalfspacing, \singlespacing and \setstretch{baselinestretch}, which will specify the line spacing for all sections and paragraphs until another command is used. Furthermore, the package provides the following environments in order to change line spacing within the document but not document-wide:

  • doublespace: lines are double spaced;
  • onehalfspace: line spacing set to one-and-half spacing;
  • singlespace: normal line spacing;
  • spacing: customizable line spacing, e.g. \begin{spacing}{\baselinestretch} ... \end{spacing}.

See the section on customizing lists for information on how to change the line spacing in lists.

Manual breaks

LaTeX takes care of formatting, breaks included. You should avoid manual breaking as much as possible, for it could lead to very bad formatting.

Controlling the breaks should be reserved to macro and package writers. Here follows a quick reference.

\newline Breaks the line at the point of the command.
\\ Breaks the line at the point of the command; it is usually a shorter version of the previous command, but LaTeX sometimes redefines it for several environments.
\\[extra-space] Command \\ has an optional argument that specifies the amount of extra vertical space to be inserted before the next line. This amount can be negative.
\\* Breaks the line at the point of the command and additionally prohibits a page break after the forced line break. This command also features the vertical space as optional parameter.
\linebreak[number] Breaks the line at the point of the command. The number you provide as an argument represents the priority of the command in a range from 0 (allow but not encourage a line break) to 4 (do it anyway). LaTeX will try to produce the best line breaks possible. If it cannot, it will decide whether or not to include the linebreak according to the priority you have provided.
\break (TeX) If used in horizontal mode, this is equivalent to `\linebreak`, if used in vertical mode it is equivalent to `\pagebreak`.
\par (TeX) Ends the current paragraph. It is equivalent to leaving a blank line in the input.
  • If you use these comments to put a break in a section heading, the line will also be broken in the table of contents. To avoid such a division, you can use the \section[]{} command, which is described in sectioning commands.

The page breaks are covered in Page Layout. More details on manual spaces between paragraphs (such as \bigskip) can be found in Lengths.

Special paragraphs

Verbatim text

There are several ways to introduce text that won't be interpreted by the compiler. If you use the verbatim environment, everything input between the begin and end commands are processed as if by a typewriter. All spaces and new lines are reproduced as given, and the text is displayed in an appropriate fixed-width font. Any LaTeX command will be ignored and handled as plain text. This is ideal for typesetting program source code. Here is an example:

The verbatim environment
  simply reproduces every
 character you input,
including all  s p a c e s!


Note: once in the verbatim environment, the only command that will be recognized is \end{verbatim}. Any others will be output. The font size in the verbatim environment can be adjusted by placing a font size command before \begin{verbatim}. If this is a problem, you can use the alltt package instead, providing an environment with the same name:

Verbatim extended with the ability
to use normal commands.  Therefore, it
is possible to \emph{emphasize} words in
this environment, for example.


Remember to add \usepackage{alltt} to your preamble to use it though! Within the alltt environment, you can use the command \normalfont to get back the normal font. To write equations within the alltt enviroment, you can use \( and \) to enclose them, instead of the usual $.

When using \textbf{} inside the alltt enviroment, note that the standard font has no bold TT font. Txtfonts has bold fonts: just add \renewcommand{\ttdefault}{txtt} after \usepackage{alltt}.

If you just want to introduce a short verbatim phrase, you don't need to use the whole environment, but you have the \verb command:

\verb+my text+

The first character following \verb is the delimiter: here we have used "+", but you can use any character you like except *; \verb will print verbatim all the text after it until it finds the next delimiter. For example, the code:

\verb;\textbf{Hi mate!};

will print \textbf{Hi mate!}, ignoring the effect \textbf should have on text.

For more control over formatting, however, you can try the fancyvrb package, which provides a Verbatim environment (note the capital letter) which lets you draw a rule round the verbatim text, change the font size, and even have typographic effects inside the Verbatim environment. It can also be used in conjunction with the fancybox package and it can add reference line numbers (useful for chunks of data or programming), and it can even include entire external files.

To use verbatim in beamer, the frame needs to be made fragile: \begin{frame}[fragile] .

Typesetting URLs

One of either the hyperref or url packages provides the \url command, which properly typesets URLs, for example:

Go to \url{} for my website.

will show this URL exactly as typed (similar to the \verb command), but the \url command also performs a hyphenless break at punctuation characters (only in PDFLaTeX, not in plain LaTeX+ dvips). It was designed for Web URLs, so it understands their syntax and will never break midway through an unpunctuated word, only at slashes and full stops. Bear in mind, however, that spaces are forbidden in URLs, so using spaces in \url arguments will fail, as will using other non-URL-valid characters.

When using this command through the hyperref package, the URL is "clickable" in the PDF document, whereas it is not linked to the web when using only the url package. Also when using the hyperref package, to remove the border placed around a URL, insert pdfborder = {0 0 0 0} inside the \hypersetup{}. (Alternately pdfborder = {0 0 0} might work if the four zeroes do not.)

You can put the following code into your preamble to change the style, how URLs are displayed to the normal font:


See also Hyperlinks

Listing environment

This is also an extension of the verbatim environment provided by the moreverb package. The extra functionality it provides is that it can add line numbers alongside the text. The command is \begin{listing}[step]{first line}. The mandatory first line argument is for specifying which line the numbering should start at. The optional step is the step between numbered lines (the default is 1, which means every line will be numbered).

To use this environment, remember to add \usepackage{moreverb} to the document preamble.

Multiline comments

As we have seen, the only way LaTeX allows you to add comments is by using the special character %, that will comment out all the rest of the line after itself. This approach is really time-consuming if you want to insert long comments or just comment out a part of your document that you want to improve later, unless you're using an editor that automates this process. Alternatively, you can use the verbatim package, to be loaded in the preamble as usual:


(you can also use the comment package instead) you can use an environment called comment that will comment out everything within itself. Here is an example:

This is another
rather stupid,
but helpful
example for embedding
comments in your document.

This is another example for embedding comments in your document.

Note that this won’t work inside complex environments, like math for example. You may be wondering, why should I load a package called verbatim to have the possibility to add comments? The answer is straightforward: commented text is interpreted by the compiler just like verbatim text, the only difference is that verbatim text is introduced within the document, while the comment is just dropped.

Alternatively, you can define a \comment{} command, by adding the following to the document's preamble:


Then, to comment out text, simply do something like this:

\comment{This is a long comment and can extend over multiple lines, etc.} But it won't show.

But it won't show.

This approach can, however, produce unwanted spaces in the document, so it may work better to use


Then if you supply only one argument to \comment{}, this has the desired effect without producing extra spaces.

Another drawback is that content is still parsed and possibly expanded, so you cannot put anything you want in it (such as LaTeX commands).

Skipping parts of the source

A more robust way of making the TeX engine skip some part of the source is to use the TeX \iffalse-conditional. The typical use is

This we want to keep

\iffalse % ----- START THE CUT ---------

But this part 
we want to skip

\fi % ---------- END THE CUT -----------

Here it begins again

This we want to keep

Here it begins again

The \iffalse-conditional is always false.

Quoting text

LaTeX provides several environments for quoting text; they have small differences and they are aimed for different types of quotations. All of them are indented on either margin, and you will need to add your own quotation marks if you want them. The provided environments are:

for a short quotation, or a series of small quotes, separated by blank lines.
for use with longer quotations, of more than one paragraph, because it indents the first line of each paragraph.
is for quotations where line breaks are important, such as poetry. Once in, new stanzas are created with a blank line, and new lines within a stanza are indicated using the newline command, \\. If a line takes up more than one line on the page, then all subsequent lines are indented until explicitly separated with \\.


In scientific publications it is customary to start with an abstract which gives the reader a quick overview of what to expect. See Document Structure.

Notes and References

This page uses material from Andy Roberts' Getting to grips with LaTeX with permission from the author.


Adding colors to your text is supported by the xcolor package (supersedes package color). Using this package, you can set the font color, text background, or page background. You can choose from predefined colors or define your own colors using RGB, Hex, or CMYK. Mathematical formulas can also be colored.

Adding the xcolor package

To make use of these features, the xcolor package must be imported. xcolor starts from the basic facilities of the color package and extends it.


The package allows you to use the names of 19 base colors (black, white, blue, green, yellow, red etc.); these names are always available. Besides, the package has some options to get more predefined colors, which should be added globally. dvipsnames allows you to access more than 60 colors, and svgnames allows access to about 150 colors. If you need more color names, then you may also want to look at the x11names option that offers more than 300 colors.

The table option allows colors to be added to tables.

Entering colored text

The simplest way to type colored text is by:


where declared-color is a color that, if necessary, was previously defined by \definecolor.

Another possible way by

{\color{declared-color}some text}

that will switch the standard text color to the color you want. It will work until the end of the current TeX group. For example:

\emph{some black text, {\color{red}followed by a red fragment}}, going black again.


The difference between \textcolor and \color is the same as that between \texttt and \ttfamily, you can use the one you prefer. The \color environment allows the text to run over multiple lines and other text environments whereas the text in \textcolor must all be one paragraph and not contain other environments.

You can change the background color of the whole page by:


Entering colored background for the text


If the background color and the text color is changed, then:


There is also \fcolorbox to make framed background color in yet another color:


Predefined colors

The predefined color names are

black, blue, brown, cyan, darkgray, gray, green, lightgray, lime, magenta, olive, orange, pink, purple, red, teal, violet, white, yellow.

There may be other pre-defined colors on your system, but these should be available on all systems.

If you would like a color not pre-defined, you can use one of the 68 dvips colors, or define your own. These options are discussed in the following sections

The 68 standard colors known to dvips

Invoke the package with the usenames and dvipsnames option. If you are using tikz or pstricks package you must declare the xcolor package before that, otherwise it will not work.


This above syntax may result in an error if you are using beamer with tikz. To go around it, include usenames and dvipsnames options when defining the document class.


Be wary that the below color names are case-sensitive. For example, \color{olivegreen} raises an "undefined color" error, but \color{OliveGreen} works fine. Table can be sorted by color name, by hue, by saturation, or by lightness.

Name Color Hex Hue Saturation Lightness
Apricot   FBB982 27.3 239.2 190.5
Aquamarine   00B5BE 182.8 255.0 95.0
Bittersweet   C04F17 19.9 200.4 107.5
Black   221E1F 0.3 15.9 32.0
Blue   2D2F92 238.2 134.8 95.5
BlueGreen   00B3B8 181.6 255.0 92.0
BlueViolet   473992 249.4 111.8 101.5
BrickRed   B6321C 8.6 187.0 105.0
Brown   792500 18.3 255.0 60.5
BurntOrange   F7921D 32.2 237.6 138.0
CadetBlue   74729A 243.0 42.1 134.0
CarnationPink   F282B4 333.2 207.0 186.0
Cerulean   00A2E3 197.2 255.0 113.5
CornflowerBlue   41B0E4 199.1 191.5 146.5
Cyan   00AEEF 196.3 255.0 119.5
Dandelion   FDBC42 39.1 249.7 159.5
DarkOrchid   A4538A 319.3 83.6 123.5
Emerald   00A99D 175.7 255.0 84.5
ForestGreen   009B55 152.9 255.0 77.5
Fuchsia   8C368C 300.0 113.0 97.0
Goldenrod   FFDF42 49.8 255.0 160.5
Gray   949698 0.2 4.9 150.0
Green   00A64F 148.6 255.0 83.0
GreenYellow   DFE674 63.7 177.3 173.0
JungleGreen   00A99A 174.7 255.0 84.5
Lavender   F49EC4 333.5 203.1 201.0
LimeGreen   8DC73E 85.4 140.3 130.5
Magenta   EC008C 324.4 255.0 118.0
Mahogany   A9341F 9.1 176.0 100.0
Maroon   AF3235 358.6 141.7 112.5
Melon   F89E7B 16.8 229.3 185.5
MidnightBlue   006795 198.5 255.0 74.5
Mulberry   A93C93 312.1 121.4 114.5
NavyBlue   006EB8 204.1 255.0 92.0
OliveGreen   3C8031 111.6 113.8 88.5
Orange   F58137 23.4 230.7 150.0
OrangeRed   ED135A 340.5 218.9 128.0
Orchid   AF72B0 299.0 71.9 145.0
Peach   F7965A 22.9 231.4 168.5
Periwinkle   7977B8 241.8 80.1 151.5
PineGreen   008B72 169.2 255.0 69.5
Plum   92268F 301.7 149.7 92.0
ProcessBlue   00B0F0 196.0 255.0 120
Purple   99479B 298.6 94.8 113.0
RawSienna   974006 24.0 235.5 78.5
Red   ED1B23 357.7 217.7 132.0
RedOrange   F26035 13.7 224.2 147.5
RedViolet   A1246B 325.9 161.8 98.5
Rhodamine   EF559F 331.2 211.1 162.0
RoyalBlue   0071BC 203.9 255.0 94.0
RoyalPurple   613F99 262.7 106.3 108.0
RubineRed   ED017D 328.5 252.9 119.0
Salmon   F69289 5.0 218.9 191.5
SeaGreen   3FBC9D 165.1 127.0 125.5
Sepia   671800 14.0 255.0 51.5
SkyBlue   46C5DD 189.5 175.8 145.5
SpringGreen   C6DC67 71.3 159.5 161.5
Tan   DA9D76 23.4 146.6 168.0
TealBlue   00AEB3 181.7 255.0 89.5
Thistle   D883B7 323.3 133.0 173.5
Turquoise   00B4CE 187.6 255.0 103.0
Violet   58429B 254.8 102.7 110.5
VioletRed   EF58A0 331.4 210.4 163.5
White   FFFFFF 0.1 0.0 255.0
WildStrawberry   EE2967 341.1 217.5 139.5
Yellow   FFF200 56.9 255.0 127.5
YellowGreen   98CC70 93.9 120.9 158.0
YellowOrange   FAA21A 36.4 244.1 138.0

Defining new colors

If the predefined colors are not adequate, you may wish to define your own.


Define the colors in the preamble of your document. (Reason: do so in the preamble, so that you can already refer to them in the preamble, which is useful, for instance, in an argument of another package that supports colors as arguments, such as the listings package.)


You need to include the xcolor package in your preamble to define new colors. In the abstract, the colors are defined following this scheme:



  • name is the name of the color; you can call it as you like
  • model is the way you describe the color, and is one of gray, rgb, RGB, HTML, and cmyk.
  • color-spec is the description of the color

Color Models

Among the models you can use to describe the color are the following (several more are described in the xcolor manual):

Color Models
Model Description Color Specification Example
gray Shades of gray
Just one number between 0 (black) and 1 (white), so 0.95 will be very light gray, 0.30 will be dark gray. \definecolor{light-gray}{gray}{0.95}
rgb Red, Green, Blue
Three numbers given in the form red,green,blue; the quantity of each color is represented with a number between 0 and 1. \definecolor{orange}{rgb}{1,0.5,0}
RGB Red, Green, Blue
Three numbers given in the form red,green,blue; the quantity of each color is represented with a number between 0 and 255. \definecolor{orange}{RGB}{255,127,0}
HTML Red, Green, Blue
Six hexadecimal numbers given in the form RRGGBB; similar to what is used in HTML. \definecolor{orange}{HTML}{FF7F00}
cmyk Cyan, Magenta, Yellow, Black
Four numbers given in the form cyan,magenta,yellow,black; the quantity of each color is represented with a number between 0 and 1. \definecolor{orange}{cmyk}{0,0.5,1,0}


To define a new color, follow the following example, which defines orange for you, by setting the red to the maximum, the green to one half (0.5), and the blue to the minimum:


The following code should give a similar results to the last code chunk.


If you loaded the xcolor package, you can define colors upon previously defined ones.

The first specifies 20 percent blue and 80 percent white; the second is a mixture of 20 percent blue and 80 percent black; and the last one is a mixture of (20*0.3) percent blue, ((100-20)*0.3) percent black and (100-30) percent green.


xcolor also features a handy command to define colors from color mixes:


Using color specifications directly

Normally one would predeclare all the colors as above, but sometimes it is convenient to directly use a color without naming it first. To achieve this, \color and \textcolor have an alternative syntax specifying the model in square brackets, and the color specification in curly braces. For example:

{\color[rgb]{1,0,0} This text will appear red-colored}
\textcolor[rgb]{0,1,0}{This text will appear green-colored}

Creating / Capturing colors

You may want to use colors that appear on another document, web pages, pictures, etc. Alternatively, you may want to play around with rgb values to create your own custom colors.

Image processing suites like the free GIMP suite for Linux/Windows/Mac offer color picker facilities to capture any color on your screen or synthesize colors directly from their respective rgb / hsv / hexadecimal values.

Smaller, free utilities also exist:

Spot colors

Spot colors are customary in printing. They usually refer to pre-mixed inks based on a swatchbook (like Pantone, TruMatch or Toyo). The package colorspace extends xcolor to provide real spot colors (CMYK and CIELAB). They are defined with, say:

\definespotcolor{mygreen}{PANTONE 7716 C}{.83, 0, .40, .11}



Font families

There are hundreds—if not thousands—of typefaces, or font families. Common examples include Times, Courier, and Helvetica. These families can generally be grouped into three main categories: serif, sans serif, and monospaced. LaTeX commands generally refer to these with the shorthand rm, sf, and tt respectively.

By default, LaTeX uses Computer Modern, a family of typefaces designed by Donald Knuth for use with TeX. It contains serif, sans serif, and monospaced fonts, each available in several weights and optical sizes.

The bodies of LaTeX documents are set in Roman (serif) type by default, but this can be changed by setting the family default:


where <family> is any of the following:

  • \rmdefault
  • \sfdefault
  • \ttdefault

Emphasizing text

Tex recognizes two types of markup commands:

  • Semantic - \emph{text} -- by default italise font. Can be overrriten by \renewcommand\emph{\textbf}. Replace textbf per requirements/preferences.
  • Visual - which actually applies required formatting:
    • Family - \textrm{} \textsf{} \texttt
    • Weight - \textbf{} - bold, \textmd{} - medium
    • Shape - \textup{}, \textit{}, \textsl{}

There are variation of the visual markup for a specific font. The above are commonly used elements. For reference on the items see section Font styles

To add some emphasis to a word or a phrase, use the \emph{text} command, which usually italicizes the text.

I want to \emph{emphasize} a word.

I want to emphasize a word.

The command is dynamic: if you emphasize a word which is already in an emphasized sentence, it will be reverted to the upright font.

\emph{In this emphasized sentence, there is an emphasized \emph{word} which looks upright.}

In this emphasized sentence, there is an emphasized word which looks upright.

Text may be emphasized more heavily through the use of boldface, particularly for keywords the reader may be trying to find when reading the text. As bold text is generally read before any other text in a paragraph or even on a page, it should be used sparingly. It may also be used in place of italics when using sans-serif typefaces to provide a greater contrast with unemphasized text. Bold text can be generated with the \textbf{text} command.

\textbf{Bold text} may be used to heavily emphasize very important words or phrases.

Bold text may be used to heavily emphasize very important words or phrases.

Font styles

Typefaces usually come in various styles and weights, such as italic and bold. The following table lists the commands you will need to access typical font shapes.

Note: Paragraph breaks are not allowed inside the command forms.

LaTeX command Equivalent switch Output style Remarks
\textnormal{...} {\normalfont ...} document font family This is the default or normal font.
\emph{...} {\em ...} emphasis Typically italics. Using emph{} inside of italic text removes the italics on the emphasized text.
\textrm{...} {\rmfamily ...} roman font family
\textsf{...} {\sffamily ...} sans serif font family
\texttt{...} {\ttfamily ...} teletypefont family This is a fixed-width or monospace font. Depending on the font zero have a dash in it to differenciate from letter O
\textup{...} {\upshape ...} upright shape The same as the normal typeface.
\textit{...} {\itshape ...} italic shape
\textsl{...} {\slshape ...} slanted shape A skewed version of the normal typeface (similar to, but slightly different from, italics).
\textsc{...} {\scshape ...} Small Capitals
\uppercase{...} uppercase (all caps) Also \lowercase. There are some caveats, though; see here.
\textbf{...} {\bfseries ...} bold
\textmd{...} {\mdseries ...} medium weight The normal font weight.
\textlf{...} {\lfseries ...} light A font weight lighter than normal. Not supported by all typefaces.

Generally, one should prefer the commands over their equivalent switches because the former automatically corrects spacing immediately following the end of the selected style.

You may have noticed the absence of underline—this is because underlining is a byproduct of the typewriter era, and is not recommended when bold and italic type is available instead.[2] However, underlining can be useful in some cases, such as to draw attention to changes during editing. Although underlining is available via the \underline{...} command, text underlined in this way will not break properly. Instead, use the \ul{...} command from the soul package or \uline{...} command from the ulem (underline emphasis) package. By default, the latter package also overrides \emph to underline instead of italicize the text. In the likely case that this is not your intent, use the normalem option, i.e. \usepackage[normalem]{ulem}. Both packages also provide strikethrough text with \st{...} or \sout{...}, respectively.

Sizing text

Built-in sizes

To scale text relative to the default body text size, use the following commands:

Command Output
\tiny sample text
\scriptsize sample text
\footnotesize sample text
\small sample text
\normalsize sample text
\large sample text
\Large sample text
\LARGE sample text
\huge sample text
\Huge sample text

These commands change the size within a given scope. For instance {\Large some words} will change the size of only some words, and does not affect the font in the rest of the document. It will work for most parts of the text.


These commands cannot be used in math mode. However, part of a formula may be set in a different size by using an \mbox command containing the size command. The new size takes effect immediately after the size command; if an entire paragraph or unit is set in a certain size, the size command should include the blank line or the \end{...} which delimits the unit.

By default, \normalsize is 10 points, but this can be changed in the \documentclass declaration, e.g. \documentclass[12pt]{article}. Note that not every document class has unique sizes for all of the above size commands.

Absolute Point Sizes
size standard classes (except slides), beamer AMS classes, memoir slides
[10pt] [11pt] [12pt] [10pt] [11pt] [12pt]
\tiny 5 6 6 6 7 8 13.82
\scriptsize 7 8 8 7 8 9 16.59
\footnotesize 8 9 10 8 9 10 16.59
\small 9 10 10.95 9 10 10.95 16.59
\normalsize 10 10.95 12 10 10.95 12 19.907
\large 12 12 14.4 10.95 12 14.4 23.89
\Large 14.4 14.4 17.28 12 14.4 17.28 28.66
\LARGE 17.28 17.28 20.74 14.4 17.28 20.74 34.4
\huge 20.74 20.74 24.88 17.28 20.74 24.88 41.28
\Huge 24.88 24.88 24.88 20.74 24.88 24.88 41.28

Points in TeX follow the standard American point system in which 1 pt is approximately 0.35136 mm. The standard point size used in most modern computer programs (known as the desktop publishing point or PostScript point) has 1 pt equal to approximately 0.3527 mm while the standard European point size (known as the Didot point) had 1 pt equal to approximately 0.37597151 mm (see: point (typography)).

Arbitrary sizes

The \tiny...\Huge commands are often enough for your needs, but you may occasionally want an arbitrary font size. This is done with \fontsize{<size>}{<line space>}\selectfont. For example:


sets the current font size to 5cm with 5.5 centimeter leading.

If you are using the latex or pdflatex engines, you may get a warning similar to the following:

LaTeX Font Warning: Font shape `OT1/cmr/m/n' in size <142.26378> not available
(Font)              size <24.88> substituted on input line 103.

This is because these older engines only support a fixed set of sizes—between 5 and 17 point. When he designed Computer Modern, Knuth created individual font files for these sizes, each with stroke widths and spacing optimized for that particular size. To avoid distorting them, scaling these fonts is disabled by default.

This issue is avoided when using lualatex or xelatex, which use Latin Modern - a vectorized version of Computer Modern - as the default font family. This still provides individual files at each of the original optical sizes, but will automatically scale the closest one when asked for an arbitrary size.

Using alternative fonts

When TeX was originally designed in the late 1970s, vector-based fonts didn't exist in any common format - PostScript wouldn't be released until 1982. Consequently, TeX was designed to use its own font system, METAFONT. Over time, TeX (and LaTeX) were extended to support PostScript fonts, and modern LaTeX engines also support the TrueType (TTF) and OpenType (OTF) fonts found on modern systems.

Using TTF and OTF fonts

If you are using lualatex or xelatex, you can use TTF and OTF fonts with the fontspec package:



Lorem ipsum...

The [Ligatures=TeX] option allows you to use the standard TeX ligatures mentioned in the Text Formatting chapter instead of Unicode characters that are unlikely to be on your keyboard. For example, --- can be used to create em dashes (—), quotes can be typed ``like this'' instead of “like this”, and so on.

The fontspec package is extremely configurable. See the manual[3] for details, but some basics are covered below.

Selecting font files

Different weights and styles of a given typeface are usually stored as separate font files. A typical typeface might offer four files to represent its normal weight, italics, bold, and bold italics. Given a typefaces's name, fontspec can generally deduce the names of the individual files. However, many typefaces come in more than two weights—some versions of Futura, for example, comes in light, book, medium, demi, bold, and extra bold weights. Sometimes small caps are stored as separate files as well.

We might want to hand-pick weights to achieve a certain look or better match the weights of other fonts in our document. Continuing to use Futura as an example, say we want to use the "book" weight for our default weight, "demi" for bold, and the font files are named:

  • Futura-Boo for upright book weight
  • Futura-BooObl for oblique book weight
  • FuturaSC-Boo for small caps, book weight
  • Futura-Dem for upright demi(bold)
  • Futura-DemObl for oblique demibold

Our font setup might resemble:

    UprightFont = *-Boo,
    ItalicFont = *-BooObl,
    SmallCapsFont = *SC-Boo,
    BoldFont = *-Dem,
    BoldItalicFont = *-DemObl

Note that instead of typing out Futura-Boo, Futura-BooObl, and so on, we can use * to insert the base name.

Controlling font features

The OpenType (OTF) format allows type designers to embed font features that can be turned on and off, such as:

  • Alternate versions of glyphs
  • Lining and "oldstyle" figures, each with tabular and proportional spacing[4]
  • Up to three sets of ligatures: standard, contextual, and historical
  • Superscript and subscript glyphs
  • Small caps (in the same file as the standard upper and lowercase characters)

All of these features can be turned on and off using different fontspec options. If we wanted to set our body text in Linux Libertine with oldstyle, proportionally-spaced figures, for example, we might set up our fonts as follows:

    Numbers={OldStyle, Proportional}
]{Linux Libertine}

Features can be turned on and off using \addfontfeatures{...}. Say you wanted to set a table in lining, tabular figures:

{\addfontfeatures{Numbers={Lining, Tabular}}
    \begin{tabular}{l r}
        Widgets: & 25 \\
        Gadgets: & 6 \\
        Whatsits & 24 \\
} % Return to previous figure style

Changing fonts in latex and pdflatex

If you are not using one of the Unicode-aware engines, font selection is more complicated. (See the discussion of encoding below.) Useful resources for latex and pdflatex font configuration include:

Font encoding

Digitising human language is a complicated topic that has evolved significantly since TeX's inception.


Today, text is usually represented in computer systems using Unicode. Briefly,

  • A Unicode text file is made of a series of code points, each of which can represent a character to be drawn, an accent or other diacritical mark to combine with an adjacent character, or some non-printing character, such as instruction to print subsequent text right-to-left.
  • One or more of these code points combines to represent a grapheme cluster or glyph, the shapes within a font that we informally call "characters".
  • Modern font formats such as TrueType and OpenType contain encoding tables which map code points to the glyphs the font file contains.

LuaLaTeX and XeLaTeX use these tools to render Unicode-encoded input files (LuaLaTeX accepts UTF-8 files, while XeLaTeX is a bit more flexible and also accepts UTF-16 and UTF-32) into PDF documents.

TeX encodings

The original TeX and LaTeX, designed long before the advent of Unicode, use a very different scheme. When using latex or pdflatex, you must choose an input encoding, which the engine uses to interpret your file, and an output encoding, which the engine uses to map your inputs to glyphs. The default font encoding is OT1, the encoding of the original Computer Modern fonts. It contains only 128 characters, many from ASCII, but leaving out some others and including a number that are not in ASCII. When accented characters are required, TeX creates them by combining a normal character with an accent. While the resulting output looks correct, this approach has some caveats compared to Unicode-based approaches:

  • It prevents automatic hyphenation from working inside words containing accented characters.
  • Searches for words with accents in PDFs will fail.
  • Extracting (e.g., via copy-paste) the umlaut 'Ä' via a PDF viewer actually extracts the two characters '"A'.
  • Some Latin letters cannot be created with this approach, to say nothing about letters of non-Latin alphabets such as Greek or Cyrillic.

To overcome these shortcomings, several other 8-bit output encodings were created. Extended Cork (EC) fonts in T1 encoding contains letters and punctuation characters for most European languages that use Latin alphabets. The LH font set contains letters necessary to typeset documents in languages using Cyrillic script. Because of the large number of Cyrillic glyphs, they are arranged into four font encodings—T2A, T2B, T2C, and X2. The CB bundle contains fonts in LGR encoding for the composition of Greek text. By using these fonts you can improve/enable hyphenation in non-English documents. Another advantage of using new CM-like fonts is that they provide fonts of CM families in all weights, shapes, and optically scaled font sizes.

All this is not possible with OT1; that's why you may want to change the font encoding of your document.

Note that different fonts support different output encodings. The default Computer Modern font does not support T1, for example. You will need Computer Modern Super (cm-super) or Latin Modern (lmodern), which are Computer Modern-like fonts with T1 support. If you have none of these, it is quite frequent (depends on your TeX installation) that tex chooses a Type3 font such as the Type3 EC, which is a bitmap font. Bitmap fonts look rather ugly when zoomed or printed.

The fontenc package tells LaTeX what font encoding to use. Font encoding is set with:


where encoding is the font encoding. It is possible to load several encodings simultaneously.

There is nothing to change in your document to use CM Super fonts (assuming they are installed), they will get loaded automatically if you use T1 encoding. For lmodern, you will need to load the package after the T1 encoding has been set:


The package ae (almost European) is obsolete. It provided some workarounds for hyphenation of words with special characters. These are not necessary any more with fonts like lmodern. Using the ae package leads to text encoding problems in PDF files generated via pdflatex (e.g. text extraction and searching), besides typographic issues.

PDF fonts and properties

PDF documents have the capability to embed font files. It makes them portable, hence the name Portable Document Format.

Many PDF viewers have a Properties feature to list embedded fonts and document metadata.

Many Unix systems make use of the poppler tool set which features pdfinfo to list PDF metadata, and pdffonts to list embedded fonts.


  1. Matthew Butterick. "Bold or italic". Practical Typography.
  2. Matthew Butterick. "Underlining". Practical Typography.
  4. Matthew Butterick. "Alternate figures". Practical Typography.

List Structures

Convenient and predictable list formatting is one of the many advantages of using LaTeX. Users of WYSIWYG word processors can sometimes be frustrated by the software's attempts to determine when they intend lists to begin and end. As a mark-up language, LaTeX gives more control over the structure and content of lists.

List structures

Lists often appear in documents, especially academic, as their purpose is often to present information in a clear and concise fashion. List structures in LaTeX are simply environments which essentially come in three types:

  • itemize for a bullet list
  • enumerate for an enumerated list and
  • description for a descriptive list.

All lists follow the basic format:

   \item {The first item}
   \item The second item 
   \item The third etc \ldots

All three of these types of lists can have multiple paragraphs per item: just type the additional paragraphs in the normal way, with a blank line between each. So long as they are still contained within the enclosing environment, they will automatically be indented to follow underneath their item. Item content could be in the curly brackets for reading convenience of long items.

Try out the examples below, to see what the lists look like in a real document.


  \item \blindtext
  \item \blindtext

  \item \blindtext
  \item \blindtext

  \item [Ant] \blindtext
  \item [Elephant] \blindtext

Sample output of lists in LaTeX. Itemize, enumerate, and description.

LaTeX will happily allow you to insert a list environment into an existing one (up to a depth of four, more levels are available using packages). Simply begin the appropriate environment at the desired point within the current list. Latex will sort out the layout and any numbering for you.

\item The first item
\item Nested item 1
\item Nested item 2
\item The second item
\item The third etc \ldots


Some special lists

Sometimes you feel the need to better align the different list items. If you are using a KOMA-script class (or package scrextend), the labeling environment is handy. It takes a mandatory argument that contains the longest of your labels.

  \item [ant] really busy all the time
  \item [chimp] likes bananas
  \item [alligator] very dangerous animal, sharp teeth, long
muscular tail and a bit of text that is longer than one
line and shows the alignment of text quite nicely


If you are on tight space limitations and only have short item descriptions, you may want to have the list inline. Please note that the example also shows how to change the font.

\blindtext Coco likes fruit. Her favorites are:
\item bananas
\item apples
\item oranges and
\item lemons.


If you want a horizontal list, package tasks can be handy. In combination with a package like exsheets, you can prepare exam papers for students.

	Which one of the entries does not fit with the others?
		\task mercury
		\task iron
		\task lead
		\task zinc
	What is a funkyton?
		\task A dancing electron
		\task A dancing proton
		\task A dancing neutron
		\task A Dixie Dancing Duck


Customizing lists

When dealing with lists containing just a few words per item, the standard lists often take up too much space. Package enumitem provides you a simple interface to customize the appearance of lists.

You can change the appearance of lists globally in the preamble, or just for single lists using the optional argument of the environment. Have a look at the following example where the list on the right is more compact using noitemsep.

  \item more work
  \item more responsibility
  \item more satisfaction
  \item more work
  \item more responsibility
  \item more satisfaction


An example for alignment and the width of the label.

\blindtext Coco likes fruit. Her favourites are:
  \item [Kate] some detail
  \item [Christina]some detail
  \item [Laura]some detail

  \item [Kate] some detail
  \item [Christina]some detail
  \item [Laura]some detail

  \item [Kate] some detail
  \item [Christina]some detail
  \item [Laura]some detail



The documentation of package enumitem goes into more detail with respect to what can be changed and how. You can even define your own lists. Environments like labeling and tasks can be changed differently, details can be found in the package documentation respectively.

Easylist package

The easylist package allows you to create list using a more convenient syntax and with infinite nested levels. It is also very customizable.

Load the package with the control character as optional argument:


The easylist environment will default to enumerations.

& Main item~:
&& Sub item.
&& Another sub item.

It features predefined styles which you can set as optional argument.

% ...

Available styles:

  • tractatus
  • checklist - All items have empty check boxes next to them
  • booktoc - Approximately the format used by the table of contents of the book class
  • articletoc - Approximately the format used by the table of contents of the article class
  • enumerate - The default
  • itemize

You can customize lists with the \ListProperties(...) command and revert back the customization with \NewList. Yes, that's parentheses for \ListProperties parameters.

The Style parameter sets the style of counters and text, the Style* parameter sets the style of counters, and the Style** parameter sets the style of text. The parameter Numbers determines the way that the numbers are displayed and the possible values are r or R (for lower and upper case Roman numerals), l or L (for lower and upper case letters), a (for Arabic numbers, the default), and z (for Zapf's Dingbats).

The FinalMark parameter sets the punctuation of the final counter (Ex: FinalMark3={)}) while FinalSpace sets the amount of space between the item and the item's text. The Margin parameter sets the distance from the left margin (Ex: FinalSpace2=1cm). The Progressive parameter sets the distance from the left margin of all items in proportion to their level.

The Hide = n parameter prevents the first n counters from appearing in all levels. If there is a number after a parameter (Ex: Style3*) then this numbers indicates the level that it will affect (Ex: Style3=\color{red}).

Example of custom enumerate:

& Main item~:
&& Sub item.
&& Another sub item.

Note that we put the FinalMark argument between {} to avoid LaTeX understanding it as the end of the properties list. Now we change the default properties to print a custom itemize:

\ListProperties(Hide=100, Hang=true, Progressive=3ex, Style*=-- ,
Style2*=$\bullet$ ,Style3*=$\circ$ ,Style4*=\tiny$\blacksquare$ )
% ...

& Blah
& Blah
&& Blah
&&& Blah
&&&& Blah
&&&&& Blah

– Blah
     – Blah

Spaces in Style parameters are important. The Style* parameter acts as a default value and easylist will use a medium dash for level 1, 5 and onward.

You can also define custom styles using LaTeX macros:

\newcommand\myitemize{\ListProperties(Hide=100, Hang=true, Progressive=3ex, Style*=$\star$ )}

% ...
\begin{easylist} \myitemize
& Blah

Important note: easylist has some drawbacks. First if you need to put an easylist inside an environment using the same control character as the one specified for easylist, you may get an error. To circumvent it, use the following commands provided by easylist:

& ...

Besides using easylist along with figures may cause some trouble to the layout and the indentation. LaTeX lists do not have this problem.

To use easylist with Beamer, each frame that uses easylist must be marked as fragile:


Special Characters

In this chapter we will tackle matters related to input encoding, typesetting diacritics and special characters.

In the following document, we will refer to special characters for all symbols other than the lowercase letters a–z, uppercase letters A-Z, figures 0–9, and English punctuation marks.

Some languages usually need a dedicated input system to ease document writing. This is the case for Arabic, Chinese, Japanese, Korean and others. This specific matter will be tackled in Internationalization.

The rules for producing characters with diacritical marks, such as accents, differ somewhat depending whether you are in text mode, math mode, or the tabbing environment.

Input encoding

TeX uses ASCII by default. But 128 characters is not enough to support non-English languages. TeX has its own way of doing that with commands for every diacritical marking (see Escaped codes). But if we want accents and other special characters to appear directly in the source file, we have to tell TeX that we want to use a different encoding.

There are several encodings available to LaTeX:

  • ASCII: the default. Only bare English characters are supported in the source file.
  • ISO-8859-1 (a.k.a., Latin 1): 8-bits encoding. It supports most characters for Latin languages, but that's it.
  • UTF-8: a Unicode multi-byte encoding. Supports the complete Unicode specification.
  • Others...

In the following we will assume that you want to use UTF-8.

There are some important steps to specify encoding.

  • Make sure your text editor decodes the file in UTF-8.
  • Make sure it saves your file in UTF-8. Most text editors do not make the distinction, but some do, such as Notepad++.
  • If you are working in a terminal, make sure it is set to support UTF-8 input and output. Some old Unix terminals may not support UTF-8. PuTTY is not set to use UTF-8 by default, you have to configure it.
  • Tell LaTeX that the source file is UTF-8 encoded.

inputenc [1] package tells LaTeX what the text encoding format of your .tex files is.

The inputenc package allows the user to change the encoding within the document as well — by means of the command \inputencoding{'encoding name'}.

% ...
% In this area
% The UTF-8 encoding is specified.
% ...
% ...
% Here the text encoding is specified as ISO Latin-1.
% ...
% Back to the UTF-8 encoding.
% ...

Extending the support

The LaTeX support of UTF-8 is fairly specific: it includes only a limited range of Unicode input characters. It only defines those symbols that are known to be available with the current font encoding. You might encounter a situation where using UTF-8 might result in error:

! Package inputenc Error: Unicode char \u8:ũ not set up for use with LaTeX.

This is due to the utf8 definition not necessarily having a mapping of all the character glyphs you are able to enter on your keyboard. Such characters include, for example:

ŷ Ŷ ũ Ũ ẽ Ẽ ĩ Ĩ

In such case, you may try to use the utf8x option to define more character combinations. utf8x is not officially supported, but can be viable in some cases. However, it might break up compatibility with some packages like csquotes.

Another possiblity is to stick with utf8 and to define the characters yourself. This is easy:

\DeclareUnicodeCharacter{'codepoint'}{'TeX sequence'}

where codepoint is the unicode codepoint of the desired character. TeX sequence is what to print when the character matching the codepoint is met. You may find codepoints on this site. Codepoints are easy to find on the web. Example:


Now inputting ŷ will effectively print ŷ.

Escaped codes

In addition to direct UTF-8 input, LaTeX supports the composition of special characters as well. This is convenient if your keyboard lacks some desired accents and other diacritics.

The following accents may be placed on letters. Although "o" letter is used in most of the examples, the accents may be placed on any letter. Accents may even be placed above a "missing" letter; for example, \~{} produces a tilde over a blank space.

The following commands may be used only in paragraph (default) or LR (left-right) mode.

LaTeX command Sample Description
\`{o} ò grave accent
\'{o} ó acute accent
\^{o} ô circumflex
\"{o} ö umlaut, trema or dieresis
\H{o} ő long Hungarian umlaut (double acute)
\~{o} õ tilde
\c{c} ç cedilla
\k{a} ą ogonek
\l{} ł barred l (l with stroke)
\={o} ō macron accent (a bar over the letter)
\b{o} o bar under the letter
\.{o} ȯ dot over the letter
\d{u} dot under the letter
\r{a} å ring over the letter (for å there is also the special command \aa)
\u{o} ŏ breve over the letter
\v{s} š caron/háček ("v") over the letter
\t{oo} o͡o "tie" (inverted u) over the two letters
\o{} ø slashed o (o with stroke)
{\i} ı dotless i (i without tittle)

Older versions of LaTeX would not remove the dot on top of the i and j letters when adding a diacritic. To correct this, one had to use the dotless version of these letters, by typing \i and \j. For example:

  • \^{\i} should be used for i-circumflex î;
  • \"{\i} should be used for i-umlaut ï.

However, current versions of LaTeX do not need this anymore (and may, in fact, crash with an error).

If a document is to be written completely in a language that requires particular diacritics several times, then using the right configuration allows those characters to be written directly in the document. For example, to achieve easier coding of umlauts, the babel package can be configured as \usepackage[german]{babel}. This provides the short hand "o for \"o. This is very useful if one needs to use some text accents in a label, since no backslash will be accepted otherwise.

More information regarding language configuration can be found in the Internationalization section.

Less than < and greater than >

The two symbols '<' and '>' are actually ASCII characters, but you may have noticed that they will print '¡' and '¿' respectively. This is a font encoding issue. If you want them to print their real symbol, you will have to use another font encoding such as T1, loaded with the fontenc package. See Fonts for more details on font encoding.

Alternatively, they can be printed with dedicated commands:


Euro currency symbol

When writing about money these days, you need the euro sign. The textcomp package features a \texteuro command which gives you the euro symbol as supplied by your current text font. Depending on your chosen font this may be quite far from the official symbol.

An official version of the euro symbol is provided by eurosym. Load it in the preamble (optionally with the official option):


then you can insert it with the \euro{} command. Finally, if you want a euro symbol that matches with the current font style (e.g., bold, italics, etc.) you can use a different option:


again you can insert the euro symbol with \euro{}.

Alternatively, you can use the marvosym package which also provides the official euro symbol.

% ...


Now that you have succeeded in printing a euro sign, you may want the '€' on your keyboard to actually print the euro sign as above. There is a simple method to do that. You must make sure you are using UTF-8 encoding along with a working \euro{} or \EUR{}command.

% or

Complete example:


Degree symbol for temperature and math

The easiest way to print temperature and angle values is to use the \SI{value}{unit} command from the siunitx package, which works both in text and math mode:


A $\SI{45}{\degree}$ angle.

It is \SI{17}{\degreeCelsius} outside.

For more information, see the documentation of the siunitx package.

A common mistake is to use the \circ command. It will not print the correct character (though $^\circ$ will). Use the textcomp package instead, which provides a \textdegree command.


A $45$\textdegree angle.

For temperature, you can use the same command or opt for the gensymb package and write


17\,\celsius % best (with textcomp)

Some keyboard layouts feature the degree symbol, you can use it directly if you are using UTF-8 and textcomp. For better results in terms of font quality, we recommend the use of an appropriate font, like lmodern:


% ...


17\,% best

Other symbols

LaTeX has many symbols at its disposal. The majority of them are within the mathematical domain, and later chapters will cover how to get access to them. For the more common text symbols, use the following commands:

Command Sample Character
\%   %
\$   $
\{   {
\_   _
\ddag n/a
\textbar n/a |
\textgreater   >
\textendash n/a
\texttrademark n/a
\textexclamdown n/a ¡
\textsuperscript{a}   a
\pounds n/a £
\#   #
\&   &
\}   }
\S   §
\dag n/a
\textbackslash n/a \
\textless   <
\textemdash n/a
\textregistered n/a ®
\textquestiondown n/a ¿
\textcircled{a} n/a
\copyright n/a ©

Not mentioned in above table, tilde (~) is used in LaTeX code to produce non-breakable space. To get printed tilde sign, either write \~{} or \textasciitilde{}. And a visible space can be created with \textvisiblespace.

For some more interesting symbols, the Postscript ZapfDingbats font is available thanks to the pifont package. Add the declaration to your preamble: \usepackage{pifont}. Next, the command \ding{number}, will print the specified symbol. Here is a table of the available symbols:


In special environments

Math mode

Several of the above and some similar accents can also be produced in math mode. The following commands may be used only in math mode.

LaTeX command Sample Description Text-mode equivalence
\hat{o}   circumflex \^
\widehat{oo}   wide version of \hat over several letters
\check{o}   vee or check \v
\tilde{o}   tilde \~
\widetilde{oo}   wide version of \tilde over several letters
\acute{o}   acute accent \'
\grave{o}   grave accent \`
\dot{o}   dot over the letter \.
\ddot{o}   two dots over the letter (umlaut in text-mode) \"
\breve{o}   breve \u
\bar{o}   macron \=
\vec{o}   vector (arrow) over the letter

When applying accents to letters i and j, you can use \imath and \jmath to keep the dots from interfering with the accents:

LaTeX command Sample Description Sample with upper dot
\hat{\imath}   circumflex on letter i without upper dot  
\vec{\jmath}   vector (arrow) on letter j without upper dot  

Tabbing environment

Some of the accent marks used in running text have other uses in the tabbing environment. In that case they can be created with the following command:

  • \a' for an acute accent
  • \a` for a grave accent
  • \a= for a macron accent

Unicode keyboard input

Some operating systems provide a keyboard combination to input any Unicode code point, the so-called unicode compose key.

Many X applications (*BSD and GNU/Linux) support the Ctrl+Shift+u combination. A "u" symbol should appear. Type the code point and press enter or space to actually print the character. Example:

<Ctrl+Shift+u> 20AC <space>

will print the euro character.

Desktop environments like GNOME and KDE may feature a customizable compose key for more memorizable sequences.

Xorg features advanced keyboard layouts with variants that let you enter a lot of characters easily with combination using the appropriate modifier, like Alt Gr. It highly depends on the selected layout+variant, so we suggest you to play a bit with your keyboard, preceding every key and dead key with the Alt Gr modifier.

In Windows, you can hold Alt and type a <codepoint> to get a desired character. For example,

<Alt> + 0252

will print the German letter ü.

External links

Notes and References

  1. For a detailed information on the package, see complete specifications written by the package's authors.


LaTeX requires some additional configuration to typeset documents in languages other than English. There are currently two packages providing international language support, namely, Babel and Polyglossia:

  • Babel[1] works with the three main engines, namely, pdfTeX, LuaTeX and XeTeX. Depending on the engine the number of supported languages (with various levels of coverage) goes from about 170 to 300, and new ones can be declared easily from scratch. It also provides partial support for Plain TeX.
  • Polyglossia was devised as an alternative to Babel for XeTeX (although currently also provides partial support for LuaTeX, but not for pdfTeX). It supports about 90 languages.

Both packages cover the major languages around the World (French, Spanish, Arabic, Chinese, Japanese, Thai, Hindi, Marathi, etc.) and handle the following tasks:

Setting the script and language tags of the current font, if possible, and switching between fonts for each language, as specified by the user (mainly XeTeX and LuaTeX). With Babel + LuaTeX the font can be switched automatically based on script. See also the discussion of fontspec in the Fonts chapter.
Linebreaking, justification and hyphenation
Activating for each script and language the corresponding line breaking algorithm. In the case of hyphenated languages, loading the language-specific hyphenation patterns. Babel provides basic line breaking for CJK scripts, as well as non-standard hyphenation, like “ff” → “ff-f”, repeated hyphens, and ranked rules. There is also some tentative support for Arabic and Tibetan justificacion.
Cultural elements
Translating document labels (like “chapter”, “figure”, “bibliography”), as well as formatting dates according to language-specific conventions and formatting numbers for languages that have their own numbering system. Polyglossia can generate the current date in the Hebrew, Islamic (Civil) and Persian calendars; Babel supports in addition Islamic Umm al-Qura, Coptic, Ethiopic, Chinese, and Buddhist.
Bidirectional typesetting
Supporting documents that contain right to left scripts. Babel + LuaTeX uses an algorithm based on the Unicode one, which changes the direction automatically. Layout elements such as tables, margins and so on must be reversed too, and this is done by Babel with LuaTeX to a great extent. With XeTeX, both Babel and Polyglossia rely on the bidi package, which requires explicit markup to change the direction.
Typographical rules and transliterations
Performing miscellaneous transformation both at the character level (like transliterations) as well as at the typographical level (like inserting spaces or penalties at appropriate places). Babel with LuaTeX can do this automatically by means of “transforms”; with XeTeX this can be done to some extent (both Babel and Polyglossia), while in 8-bit engines many of them must be done by hand.

With Babel, LaTeX ≥ 2018-04-01, and a monolingual document in UTF-8 encoding (which is the recommended encoding), all you need in many European languages is something like, for example:

\usepackage[T1]{fontenc}  % <- With XeTeX or LuaTeX, delete this line

Plus ça change, plus c'est la même chose!

In addition, there are some specialized frameworks for languages like Japanese, Korean or Chinese, described below.


Unicode engines

When using the xelatex or lualatex engines, many of the problems described below are solved for you. Input files are assumed to be UTF-8 (XeLaTeX also accepts UTF-16 and UTF-32), and the engine automatically maps Unicode characters to their glyphs in the TrueType or OpenType fonts you selected for your document. (This is, of course, assuming those fonts contain the glyphs you need, so you must ensure that your fonts support the languages you are using.)

8-bit engines

With engines not supporting Unicode internally (latex or pdflatex), LaTeX must handle two fundamental problems:

  1. Mapping the bytes of your input file into the characters of the language(s) you want to use.
  2. Mapping those characters to their glyphs in the fonts your document uses.

With them, you must tell LaTeX which encoding to use for your input files, and what "output" encoding it should use to map characters to their glyphs in the fonts. In most cases (especially for multilingual documents), UTF-8 is an optimal input encoding, which is currently the default encoding.

For most Latin languages, T1 is the desired output encoding, and can be set with:


Other output encodings for specific languages are shown below.

For additional information, see the discussion of encoding in the Fonts chapter, as well as the Special Characters chapter.


The core package babel supports the 3 main engines (PDFLaTeX, LuaLaTeX and XeLaTeX). There are two ways to specify the document languages. One of them is as arguments to the package when it is loaded:


Another approach is making the language a global option in order to let other packages detect and use it:


Finally, babel provides total or partial support for about 250 languages with a set of ini files, which are accessed with \babelprovide. This command can be used to define easily your own language from scratch, too.

Babel will automatically activate the appropriate hyphenation rules for the language you choose. If your LaTeX format does not support hyphenation in the language of your choice, babel will still work but will disable hyphenation, which has quite a negative effect on the appearance of the typeset document (with LuaLaTeX, however, hyphenation rules can be loaded when the document is being typeset). Babel also specifies new commands for some languages, which simplify the input of special characters. See the sections about languages below for more information.

If you call babel with multiple languages:


Short texts in a secondary language does not require an explicit declaration when loading babel. Just select it as explained in what follows and the basic declarations will be loaded on the fly.

The last language in the option list will be active (i.e. languageB), and you can use the command


to change the active language (when the document begins, with \begin{document}, the main language is automatically selected). You can also add short pieces of text in another language using the command

\foreignlanguage{languageB}{Text in another language}

Babel also offers various environments for entering larger pieces of text in another language:

Text in language B. This environment switches all language-related definitions, like the language 
specific names for figures, tables etc. to the other language.

The starred version of this environment typesets the main text according to the rules of the other language, but keeps the language specific string for ancillary things like figures in the main language of the document. The environment hyphenrules switches only the hyphenation patterns used; it can also be used to disallow hyphenation by using the language name 'nohyphenation' (but note otherlanguage* is preferred).

The babel manual provides much more information on these and many other options.

Font management

If you are using XeTeX or LuaTeX, Babel supports OpenType fonts with fontspec. To ease font handling, it provides the macro \babelfont, which switches the font across languages and sets the OpenType ‘language system’ (ie, language and script). Let us assume you are setting up a document in Swedish, with some words in Hebrew, with a font suited for both languages:


If, on the other hand, you have to resort to different fonts, you would say:


Also, with version >=3.38 the locale identifiers (\language and \localeid) and the fonts can be switched without explicit markup, depending on the script (only LuaTeX). In the following example, bidi=basic switches the direction, and onchar=ids fonts switches the identifiers and the font:

\usepackage[swedish, bidi=basic]{babel} 
\babelprovide[import, onchar=ids fonts]{hebrew} 
Svenska עִבְרִית svenska.

Bidirectional texts

Babel provides basic support fo bidi texts, mainly in LuaTeX. The package option may take three values, namely, default, basic-r, and basic. With bidi=basic RTL and LTR text can be mixed without explicit markup (only LuaTeX).

Multilingual versions

It is possible in LaTeX to typeset the content of one document in several languages and to choose upon compilation which language to output in predefined strings (chapter name, date, etc.). Using the commands above in multilingual documents can be cumbersome, and therefore babel provides a way to define shorter names. With

\babeltags{de = german}

You can write:

text \textde{German text} text
German text

There is a clear drawback to this feature, namely, the ‘prefix’ \text... is heavily overloaded in LaTeX and conflicts with existing macros may arise. The babel manual recommends to to stick to the default selectors or to define your own alternatives.

The current language can also be tested by using the iflang package by Heiko Oberdiek (the built-in feature from the babel package is not reliable). Here comes a simple example:


This allows to easily distinguish between two languages without the need of defining own commands. Another approach for localized strings is translator.


When using XeLaTeX or LuaLaTeX, polyglossia provides an alternative to the core babel package for international language support, as described in its manual.

The original aim was to be compatible with babel, but there is a number of differences. For example, the standard mechanism in LaTeX to declare languages, via package or class options, is not recognized, and the user must rely on a set of new commands, as shown in the example. Unlike babel, secondary languages must be always explicitly declared. It also adds the concept of ‘language variant’, while in babel all locales are treated on an equal footing. Not only languages are declared in a non standard way, but also a new way to switch languages has been devised, with commands like \textenglish or \textlang.

To use polyglossia, load it in your preamble and specify the languages you will be using, along with any language-specific options you wish.

If, for example, we have a document with American English as the main language, and some short texts in French, Bulgarian and Serbian, you might use:

\setotherlanguages{french, bulgarian, serbian}
English. \textlang{french}{French}. \textlang{bulgarian}{Български}.

As a comparison, here is the code with `babel`:

English. \foreignlanguage{french}{French}.
\foreignlanguage{bulgarian}{Български}. \foreignlanguage{serbian}{Српски}.

Specific languages

Here is a collection of language-specific suggestions. If you have experience in a language not listed below, please add some notes about it. Some of the methods described in this chapter may be useful when dealing with non-English author names in bibliographies.

Arabic script

Documents in the Arabic script, including Arabic, Persian, Urdu, Pashto, Kurdish, Uyghur, etc., are best typeset with either XeTeX or LuaTeX. An example with babel and LuaTeX follows (rendering by the browser may be different from an editor):

\babelprovide[import, main]{arabic}
وﻗﺪ ﻋﺮﻓﺖ ﺷﺒﻪ ﺟﺰﻳﺮة اﻟﻌﺮب ﻃﻴﻠﺔ اﻟﻌﺼﺮ اﻟﻬﻴﻠﻴﻨﻲ )اﻻﻏﺮﻳﻘﻲ( ﺑـ
Arabia أو Aravia )ﺑﺎﻻﻏﺮﻳﻘﻴﺔ Αραβία (، اﺳﺘﺨﺪم اﻟﺮوﻣﺎن ﺛﻼث
ﺑﺎدﺋﺎت ﺑـ “Arabia” ﻋﻠﻰ ﺛﻼث ﻣﻨﺎﻃﻖ ﻣﻦ ﺷﺒﻪ اﻟﺠﺰﻳﺮة اﻟﻌﺮﺑﻴﺔ، إﻻ أﻧﻬﺎ
ﺣﻘﻴﻘﺔً ﻛﺎﻧﺖ أﻛﺒﺮ ﻣﻤﺎ ﺗﻌﺮف ﻋﻠﻴﻪ اﻟﻴﻮم.

With XeTeX, you may set bidi=bidi-r, but mixed LR and RL text must be marked up explicitly. The same applies to polyglossia.

babel with LuaTeX provides partial and tentative support for Arabic justification based on kashida (with the ARABIC TATWEEL Unicode character) or on the ‘justification alternatives’ OpenType table (jalt).

An alternative package for LuaTeX is arabluatex, which is an extension for LuaTeX of arabtex, described below. For XeTeX there is arabxetex.

In 8-bit engines, they can be typeset in a number of ways, one of the oldest being arabtex. Add the following code to your preamble:


You can input text in either romanized characters or native Arabic script encodings. Use any of the following commands and environments to enter in text:

\< ... >
\RL{ ... }
\begin{arabtext} ... \end{arabtext}.

See the ArabTeX Wikipedia article for further details.

You may also use the Arabi package within Babel to typeset Arabic and Persian


You may also copy and paste from PDF files produced with Arabi thanks to the support of the cmap package. You may use Arabi with LyX, or with tex4ht to produce HTML.

See Arabi page on CTAN


The Armenian script uses its own characters, which will require you to install a text editor that supports Unicode and will allow you to enter UTF-8 text, such as Texmaker or WinEdt. These text editors should then be configured to compile using XeLaTeX or LuaLaTeX.

Once the text editor is set up to compile with XeLaTeX or LuaLaTeX, the fontspec package can be used to write in Armenian:

\setmainfont{DejaVu Serif}



The Sylfaen font lacks italic and bold, but DejaVu Serif supports them.

See Armenian Wikibooks for further details, especially on how to configure the Unicode supporting text editors to compile with Unicode engines.

Cyrillic script

Currently the most convenient way to typeset Cyrillic texts is with XeTeX or LuaTeX in the UTF-8 encoding. An example for Russian with these engines, which do not require encoding transformations because everything is done directly in that encoding, is:


\babelfont{rm}{DejaVu Serif}


Россия, находящаяся на пересечении множества культур, а также
с учётом многонационального характера её населения, — отличается
высокой степенью этнокультурного многообразия и способностью к
межкультурному диалогу.


Support for Cyrillic in non-Unicode engines is based on standard LaTeX mechanisms plus the fontenc and inputenc packages. babel includes support for the T2* encodings and for typesetting Bulgarian, Russian and Ukrainian texts using Cyrillic letters[2] with non-Unicode engines. AMS-LaTeX packages should be loaded before fontenc and babel(Why?). If you are going to use Cyrillics in mathmode, you also need to load mathtext package before fontenc:



Generally, babel will automatically choose the default font encoding, for the above three languages this is T2A. However, documents are not restricted to a single font encoding. For multilingual documents using Cyrillic and Latin-based languages it makes sense to include Latin font encoding explicitly. Babel will take care of switching to the appropriate font encoding when a different language is selected within the document.

On modern operating systems it is beneficial to use Unicode (utf8 or utf8x) instead of KOI8-RU (koi8-ru) as an input encoding for Cyrillic text.

In addition to enabling hyphenations, translating automatically generated text strings, and activating some language specific typographic rules (like \frenchspacing), babel provides some commands allowing typesetting according to the standards of Bulgarian, Russian, or Ukrainian languages.

For all three languages, language specific punctuation is provided: the Cyrillic dash for the text (it is little narrower than Latin dash and surrounded by tiny spaces), a dash for direct speech, quotes, and commands to facilitate hyphenation:

Key combination Action
"| No ligature at this position.
"- Explicit hyphen sign, allowing hyphenation in the rest of the word.
"--- Cyrillic emdash in plain text.
"--~ Cyrillic emdash in compound names (surnames).
"--* Cyrillic emdash for denoting direct speech.
"" Similar to "-, but it produces no hyphen sign (used for compound words with hyphen, e.g. x-""y or some other signs as “disable/enable”).
"~ Compound word mark without a breakpoint.
"= Compound word mark with a breakpoint, allowing hyphenation in the composing words.
", Thinspace for initials with a breakpoint in a following surname.
"‘ German opening double quote (,,).
"’ German closing double quote (“).
"< French opening double quote (<<).
"> French closing double quote (>>).

The Russian and Ukrainian options of babel define the commands


which act like \Alph and \alph (commands for turning counters into letters, e.g. a, b, c...), but produce capital and small letters of Russian or Ukrainian alphabets (whichever is the active language of the document).

The Bulgarian option of babel provides the commands


which make \Alph and \alph produce letters of either Bulgarian or Latin (English) alphabets. The default behaviour of \Alph and \alph for the Bulgarian language option is to produce letters from the Bulgarian alphabet.

See the Bulgarian translation of "The Not So Short Introduction to LaTeX" [3] for a method to type Cyrillic letters directly from the keyboard using a different distribution.


Typesetting Chinese texts (and, in general, CJK script ones) is best done with a complete framework, like CJK o xeCJK, although for short texts or a few words in horizontal typesetting babel with XeTeX and LuaTeX could be enough, with basic line breaking.

CJK Package

One possible Chinese support is made available thanks to the CJK package collection. If you are using a package manager or a portage tree, the CJK collection is usually in a separate package because of its size (mainly due to fonts).

Make sure your document is saved using the UTF-8 character encoding. See Special Characters for more details. Put the parts where you want to write chinese characters in a CJK environment.



You can mix Latin letters and Chinese.


The last argument specifies the font. It must fit the desired language, since fonts are different for Chinese, Japanese and Korean. Possible choices for Chinese include:

  • gbsn (简体宋体, simplified Chinese)
  • gkai (简体楷体, simplified Chinese)
  • bsmi (繁體細上海宋體, traditional Chinese)
  • bkai (繁體標楷體, traditional Chinese)

In CTeX distribution (which has been outdated), six more fonts for simplified Chinese are included, corresponding to default Windows fonts:

  • song (宋体, Simsun)
  • hei (黑体, Simhei)
  • fang (仿宋, STFangSong)
  • kai (楷体, STKaiti)
  • li (隶书, SimLi)
  • you (幼圆, SimYou)

xeCJK Package

When using the XeTeX engine, there is another package called xeCJK, which is based on fontspec and offers similar interface to CJK package.

When using the package, one can define CJK fonts like this:

\setCJKmainfont{FZSSK.ttf} % use Foundertype's Chinese font, which has a free license
You can still mix Latin letters and Chinese!


Czech is fine using


UTF-8 allows you to have „czech quotation marks“ directly in your text. Otherwise, there are macros \clqq and \crqq to produce left and right quote. You can place quotated text inside \uv.

Copying and searching in PDF

Although czech letters with diacritical sign are displayed correctly, they are not copy-able or search-able in PDF files generated with pdfLaTeX with just command above. Using package cmap solves this for some fonts, for others is also neccessary to use command glyphtounicode.

Combinations of commands with different fonts
Font (no additional command)
ešcržýáíédtnúuŠCRŽÁÚ ěščřžýáíéďťňúůŠČŘŽÁÚ ěščřžýáíéďťňúůŠČŘŽÁÚ ěščřžýáíéďťňúůŠČŘŽÁÚ
ešcržýáíédtnúuŠCRŽÁÚ ešcržýáíédtnúuŠCRŽÁÚ ešcržýáíédtnúuŠCRŽÁÚ ěščřžýáíéďťňúůŠČŘŽÁÚ

Devanagari and other Indic scripts

The Devanagari script is used by many languages, including Marathi, Pāḷi, Sanskrit, Hindi, Nepali, Bodo, Konkani, Prakrit. Here is an example for Hindi with babel, for both XeTeX and LuaTeX:

\usepackage[hindi, provide=*]{babel}
देवनागरी एक भारतीय लिपि है जिसमें अनेक भारतीय भाषाएँ तथा कई विदेशी
भाषाएँ लिखी जाती हैं।

Other Indic scripts have a similar setup (Malayalam, Bengali, Sinhala, Telugu, Tamil, Kannada, Assamese, Punjabi, etc.).

If any additional features are required, you need an alternative approach, as illustrated in the following example for Bangla, which sets the option mapdigits for the Arabic digits to be converted to the local ones (only LuaTeX).

\babelprovide[import, main, mapdigits]{bengali}
গাইতে গাইতে গায়েন।

Mapping the digits is accomplished in XeTeX at the font level, with the option Mapping=, like:


This is actually a XeTeX feature and doesn't require babel. It can be used directly with fontspec.

Support in pdfTeX is based mainly on the velthuis package. An alternative for XeTeX is latexbangla, which relies on polyglossia.


Finnish language hyphenation is enabled with:


This will also automatically change document language (section names, etc.) to Finnish.


As of version 3.0 of babel-french, it is advised to choose the language as a global option with the following command[4]:


Formerly, you could load French language support with the following command:




There are multiple options for typesetting French documents, depending on the flavor of French: french for Parisian French, and acadian and canadien for new-world French. If you do not know or do not really care, we would recommend using french.

All enable French hyphenation, if you have configured your LaTeX system accordingly. All of these also change all automatic text into French: \chapter prints Chapitre, \today prints the current date in French and so on. A set of new commands also becomes available, which allows you to write French input files more easily. Check out the following table for inspiration:

input code rendered output
\og guillemets \fg{} « guillemets »
M\up{me}, D\up{r} Mme, Dr
1\ier{}, 1\iere{}, 1\ieres{} 1er, 1re, 1res
2\ieme{} 4\iemes{} 2e 4es
\No 1, \no 2 N° 1, n° 2
20~\degres C, 45\degres 20 °C, 45°
M. \bsc{Durand} M. Durand
\nombre{1234,56789} 1 234,567 89

You may want to typeset guillemets and other French characters directly if your keyboard has them. Running Xorg (*BSD and GNU/Linux), you may want to use the oss variant which features some nice shortcuts, like

Key combination Character
Alt Gr + w «
Alt Gr + x »
Alt Gr + Shift + é É
Alt Gr + Shift + è È
Alt Gr + Shift + ç Ç

You will need the T1 font encoding for guillemets to print properly.

For the degree character you will get an error like

! Package inputenc Error: Unicode char \u8:° not set up for use with LaTeX.

The textcomp package will fix it for you.

The great advantage of Babel for French is that it will handle some elements of French typography for you, especially non-breaking spaces before all two-parts punctuation marks. So now you can write:

Il répondit: «Ce pain coûte-t-il 2~€?»

The non-breaking space before the euro symbol is still necessary because currency symbols and other units or not supported in general (that's not specific to French).

You can use the numprint package along Babel. It will let you print numbers the French way.

\usepackage[autolanguage]{numprint} % Must be loaded *after* babel.

% ...

\nombre{123456.123456 e-17}


You will also notice that the layout of lists changes when switching to the French language. This is customizable using the \frenchsetup command. For more information on what the french option of babel does and how you can customize its behavior, run LaTeX on file frenchb.dtx and read the produced file frenchb.pdf or frenchb.dvi. You can get the PDF version on CTAN.


You can load German language support using either one of the two following commands (pdfTeX, XeTeX and LuaTeX are supported).

For traditional ("old") German orthography use


or for reform ("new") German orthography use


This enables German hyphenation, if you have configured your LaTeX system accordingly. It also changes all automatic text into German, e.g. “Chapter” becomes “Kapitel”. A set of new commands also becomes available, which allows you to write German input files more quickly even when you don't use the inputenc package. Check out the table below for inspiration. With inputenc, all this becomes moot, but your text also is locked in a particular encoding world.

German Special Characters.
"A "O "U Ä Ö Ü
"a "o "u "s ä ö ü ß
"` or \glqq
"' or \grqq
\glq \grq
"< or \flqq «
"> or \frqq »
\flq \frq ‹ ›
\dq "

In German books you sometimes find French quotation marks («guillemets»). German typesetters, however, use them differently. A quote in a German book would look like »this«. In the German speaking part of Switzerland, typesetters use «guillemets» the same way the French do. A major problem arises from the use of commands like \flq: If you use the OT1 font encoding (which is the default) the guillemets will look like the math symbol " ", which turns a typesetter's stomach. T1 encoded fonts, on the other hand, do contain the required symbols. So if you are using this type of quote, make sure you use the T1 encoding.

Decimal numbers usually have to be written like 0{,}5 (not just 0,5). Packages like ziffer enable input like 0,5. Alternatively, one can use the \num command from the babel and (globally) set the decimal marker using

% ...



This is the preamble you need to write in the Greek language.


This preamble enables hyphenation and changes all automatic text to Greek. A set of new commands also becomes available, which allows you to write Greek input files more easily.

Modern Monotonic Greek, as well as Polytonic and Ancient Greek are supported.

If you need a language in the Latin script and you are using LuaTeX, you can switch automatically the font in the following way, with no explicit markup:


\usepackage[portuguese, greek]{babel}
\babelprovide[onchar=ids fonts]{portuguese}
\babelfont[portuguese]{rm}{DejaVu Sans}

   abelha -- μελισσα

There is a dedicated package for XeTeX named xgreek.


Use the following lines:


More information in hungarian.

Icelandic and Faroese

The following lines can be added to write Icelandic text:


This changes text like Part into Hluti. It makes additional commands available:

Icelandic special commands
"` or \glqq
\th þ
\dh ð

To make special characters such as Þ and Æ become available just add:


The default LATEX font encoding is OT1, but it contains only the 128 characters. The T1 encoding contains letters and punctuation characters for most of the European languages using Latin script.


Italian is well supported by LaTeX. Just add


at the beginning of your document and the output of all the commands will be translated properly.


Norwegian is well supported by LaTeX. Just add


at the beginning of your document and the output of all the commands will be translated properly.



The package provides the class file and JFM (Japanese font metric) files for LuaTeX-ja / pLaTeX / upLaTeX. This aims to implement Requirements for Japanese Text Layout.

upTeX, pTeX

There is a variant of TeX intended for Japanese named upTeX, which supports vertical typesetting.


Another possible way to write in japanese is to use Lualatex and the luatex-ja package. Adapted example from the Luatexja documentation :

\usepackage{luatexja} % This line is unnecessary when using ltjclasses or ltjsclasses.

You can also use capabilities provided by the fontspec package and those provided by luatexja-fontspec to declare the font you want to use in your paper. Let us take an example:

% **********************************
% Basic setup
\setmainfont[Numbers={OldStyle,Proportional}]{Arno Pro} %setup of western font
\usepackage{luatexja-fontspec}%needed to call \setmainjfont bellow
\setmainjfont[BoldFont=KozGoPr6N-Bold]{KozGoPr6N-Regular} %setup of japanese font
It is a test to show japanese and english mix. テスト中です。どうですか皆さん。

Use UTF-8 as your encoding. In case you don't know how to do this, take a look at Texmaker, a LaTeX editor which uses UTF-8 by default.

luatex-ja can collaborate with babel. For example:


For short Japanese texts (a few words or a few paragraphs) in a document in another language, babel (≥3.31) with luatex could be enough; eg:


For hyperref package to show the Table of Contents correctly, the encoding has to be explicitly specified.


CJK, XeCJK, bxcjkjatype

Another (but old) possible Japanese support is made available thanks to the CJK or XeCJK package collection. If you are using a package manager or a portage tree, the CJK collection is usually in a separate package because of its size (mainly due to fonts).

Make sure your document is saved using the UTF-8 character encoding. See Special Characters for more details. Put the parts where you want to write japanese characters in a CJK environment.



You can mix latin letters as well as hiragana, katakana and kanji.


The last argument specifies the font. It must fit the desired language, since fonts are different for Chinese, Japanese and Korean. min is an example for Japanese.

The bxcjkjatype pack­age pro­vides a work­ing con­fig­u­ra­tion of the CJK pack­age, suit­able for Ja­panese type­set­ting of mod­er­ate qual­ity. More­over, it fa­cil­i­tates use of the CJK pack­age for pLATEX users, by pro­vid­ing com­mands that are sim­i­lar to those used by the pLATEX ker­nel and some other pack­ages used with it.



The two most widely used encodings for Korean text files are EUC-KR and its upward compatible extension used in Korean MS-Windows, CP949/Windows-949/UHC. In these encodings each US-ASCII character represents its normal ASCII character similar to other ASCII compatible encodings such as ISO-8859-x, EUC-JP, Big5, or Shift_JIS. On the other hand, Hangul syllables, Hanjas (Chinese characters as used in Korea), Hangul Jamos, Hiraganas, Katakanas, Greek and Cyrillic characters and other symbols and letters drawn from KS X 1001 are represented by two consecutive octets. The first has its MSB set. Until the mid-1990's, it took a considerable amount of time and effort to set up a Korean-capable environment under a non-localized (non-Korean) operating system. You can skim through the now much-outdated to get a glimpse of what it was like to use Korean under non-Korean OS in mid-1990's.

TeX and LaTeX were originally written for scripts with no more than 256 characters in their alphabet. To make them work for languages with considerably more characters such as Korean or Chinese, a subfont mechanism was developed. It divides a single CJK font with thousands or tens of thousands of glyphs into a set of subfonts with 256 glyphs each.

For Korean, there are three widely used packages.

  • HLATEX by UN Koaunghi
  • hLATEXp by CHA Jaechoon
  • the CJK package by Werner Lemberg

HLATEX and hLATEXp are specific to Korean and provide Korean localization on top of the font support. They both can process Korean input text files encoded in EUC-KR. HLATEX can even process input files encoded in CP949/Windows-949/UHC and UTF-8 when used along with Λ, Ω.

The CJK package is not specific to Korean. It can process input files in UTF-8 as well as in various CJK encodings including EUC-KR and CP949/Windows-949/UHC, it can be used to typeset documents with multilingual content (especially Chinese, Japanese and Korean). The CJK package has no Korean localization such as the one offered by HLATEX and it does not come with as many special Korean fonts as HLATEX.

The ultimate purpose of using typesetting programs like TeX and LaTeX is to get documents typeset in an aesthetically satisfying way. Arguably the most important element in typesetting is a set of welldesigned fonts. The HLATEX distribution includes UHC PostScript fonts of 10 different families and Munhwabu fonts (TrueType) of 5 different families. The CJK package works with a set of fonts used by earlier versions of HLATEX and it can use Bitstream's cyberbit True-Type font.

To use the HLATEX package for typesetting your Korean text, put the following declaration into the preamble of your document:


This command turns the Korean localization on. The headings of chapters, sections, subsections, table of content and table of figures are all translated into Korean and the formatting of the document is changed to follow Korean conventions. The package also provides automatic particle selection. In Korean, there are pairs of post-fix particles grammatically equivalent but different in form. Which of any given pair is correct depends on whether the preceding syllable ends with a vowel or a consonant. (It is a bit more complex than this, but this should give you a good picture.) Native Korean speakers have no problem picking the right particle, but it cannot be determined which particle to use for references and other automatic text that will change while you edit the document. It takes a painstaking effort to place appropriate particles manually every time you add/remove references or simply shuffle parts of your document around. HLATEX relieves its users from this boring and error-prone process.

In case you don't need Korean localization features but just want to typeset Korean text, you can put the following line in the preamble, instead.


For more details on typesetting Korean with HLATEX, refer to the HLATEX Guide. Check out the web site of the Korean TeX User Group (KTUG).

In the FAQ section of KTUG it is recommended to use the kotex package


Persian script

For Persian language, there is a dedicated package called XePersian which uses XeLaTeX as the typesetting engine. Just add the following code to your preamble:


See XePersian page on CTAN

Moreover, Arabic script can be used to type Persian as illustrated in the corresponding section.


If you plan to use Polish in your encoded document, use the following code:


The above code merely allows to use Polish letters and translates the automatic text to Polish, so that "chapter" becomes "rozdział". There are a few additional things one must remember about.


Polish has many single letter connectives: "a", "o", "w", "i", "u", "z", etc., grammar and typography rules don't allow for them to end a printed line. To ensure that LaTeX won't set them as last letter in the line, you have to use non breakable space:

Noc była sierpniowa, ciepła i~słodka, Księżyc oświecał srebrnem światłem wgłębienie, tak,
że twarze małego rycerza i~Basi były skąpane w blasku.
Poniżej, na podwórzu zamkowem, widać było uśpione kupy żołnierzy, a~także i~ciała zabitych
podczas dziennej strzelaniny, bo nie znaleziono dotąd czasu na ich pogrzebanie.

Babel (>=3.58) with LuaTeX provides a transform for this purpose, without explicit markup, which is activated with:

\babelprovide[transforms = oneletter.nobreak]{polish}


According to Polish grammar rules, you have to put dots after numerals in chapter, section, subsection, etc. headers.

This is achieved by redefining few LaTeX macros.

For books:


For articles:


Alternatively you can use dedicated document classes:

  • the mwart class instead of article,
  • mwbk instead of book
  • and mwrep instead of report.

Those classes have much more European typography settings but do not require the use of Polish babel settings or character encoding.

Simple usage:

Pójdź kińże tę chmurność w głąb flaszy.

Full documentation for those classes is available at (Polish).


It may be customary (depending on publisher) to indent the first paragraph in sections and chapters:


Hyphenation and typography

It's much more frowned upon to set pages with hyphenation between pages than it is customary in American typesetting.

To adjust penalties for hyphenation spanning pages, use this command:


To adjust penalties for leaving widows and orphans (clubs in TeX nomenclature) use those commands:


Commas in math

According to some typography rules, fractional parts of numbers should be delimited by a comma, not a dot. To make LaTeX not insert additional space in math mode after a comma (unless there is a space after the comma), use the icomma package.


Unfortunately, it is partially incompatible with the dcolumn package. One needs to either use dots in columns with numerical data in the source file and make dcolumn switch them to commas for display or define the column as follows:

\begin{tabular}{... D{,}{\mathord\mathcomma}{2} ...}

The alternative is to use the numprint package, but it is much less convenient.

Another alternative is using package siunitx that lets you typeset numbers and their according units consistently. Number alignment in tables and different output modes re supported.

Further information

Refer the Słownik Ortograficzny (in Polish) for additional information on Polish grammar and typography rules.

Good extract is available at Zasady Typograficzne Składania Tekstu (in Polish).


Add the following code to your preamble:


You can substitute the language for brazilian portuguese by choosing brazilian or brazil.


Basic settings are fine when left the same as Czech, but Slovak needs special signs for 'ď', 'ť', 'ľ'. To be able to type them from keyboard use the following settings:



Include the appropriate Babel option:


The trick is that Spanish has several options and commands to control the layout. The options may be loaded either at the call to Babel, or before, by defining the command \spanishoptions. Therefore, the following commands are roughly equivalent:


On average, the former syntax should be preferred, as the latter is not recognized by some programs (LyX, latex2rtf) interacting with LaTeX.

Spanish also defines shorthands for the dot and << >> so that they are used as logical markup: the former is used as decimal marker in math mode, and the output is typically either a comma or a dot; the latter is used for quoted text, and the output is typically either «» or “”. This allows different typographical conventions with the same input, as preferences may be quite different from, say, Spain and Mexico.

Two particularly useful options are es-noquoting,es-nolists: some packages and classes are known to collide with Spanish in the way they handle active characters, and these options disable the internal workings of Spanish to allow you to overcome these common pitfalls. Moreover, these options may simplify the way LyX customizes some features of the Spanish layout from inside the GUI.

The options mexico,mexico-com provide support for local custom in Mexico: the former using decimal dot, as customary, and the latter allowing decimal comma, as formerly required by the Mexican Official Norm (NOM) of the Department of Economy for labels in foods and goods. More localizations are in the making.

The other commands modify the Spanish layout after loading Babel. Two particularly useful commands are \spanishoperators and \spanishdeactivate.

The macro \spanishoperators{<list of operators>}{ contains a list of spanish mathematical operators, and may be redefined at will. For instance, the command


only defines sen, overriding all other definitions; the command \let\spanishoperators\relax disables them all. This command supports accented or spaced operators: the \acute{<letter>} command puts an accent, and the \, command adds a small space. For instance, the following operators are defined by default.

l\acute{i}m l\acute{i}m\,sup l\acute{i}m\,inf m\acute{a}x 
\acute{i}nf m\acute{i}n sen tg arc\,sen arc\,cos arc\,tg 
cotg cosec senh tgh

Finally, the macro \spanishdeactivate{<list of characters>} disables some active characters, to keep you out of trouble if they are redefined by other packages. The candidates for deactivation are the set {<>."'}. Please, beware that some option preempt the availability of some active characters. In particular, you should not combine the es-noquoting option with \spanishdeactivate{<>}, or the es-noshorthands with \spanishdeactivate{<>."}.

Please check the documentation for Babel or spanish.dtx for further details.


Both babel (luatex and xetex) and polyglossia (only xetex) support Thai. Word division in luatex is based on the standard hyphenation mechanism, so that patterns can be modified with \babelpatterns, while xetex relies on its own built-in mechanism. In pdftex you need an external tool for word segmentation (like swath). An example with babel (luatex and xetex) is:

\babelprovide[main, import]{thai}


One option to use Tibetan script in LaTeX is to add


to your preamble and use a slightly modified Wylie transliteration for input. Refer to the excellent package documentation for details. More information can be found on [3]

`babel` for `luatex` provides tentative support for justification with trailing tshegs.[4]


The following preamble could be used to directly type Vietnamese (xetex or luatex).

\setmainfont[Ligatures=TeX]{Linux Libertine O}

For a document written in this language:



  1. Babel: The multilingual framework to localize LaTeX, LuaLaTeX, XeLaTeX
  2. The Not So Short Introduction to LaTeX, 2.5.6 Support for Cyrillic, Maksym Polyakov
  3. The Not So Short Introduction to LaTeX, Bulgarian translation
  4. babel-french documentation: "the French language should now be loaded as french, not as frenchb or francais and preferably as a global option of \documentclass. Some tolerance still exists in v3.0, but do not rely on it."


The rotating package

The package rotating gives you the possibility to rotate any object of an arbitrary angle. Once you have loaded it with the standard command in the preamble:


you can use three new environments:


it will rotate the whole argument by 90 degrees counterclockwise. Moreover:


it will turn the argument of 30 degrees. You can give any angle as an argument, whether it is positive or negative. It will leave the necessary space to avoid any overlapping of text.


like turn, but it will not add any extra space.

If you want to make a float sideways so that the caption is also rotated, you can use




Note, though, they will be placed on a separate page.

If you would like to rotate a TikZ picture you could use sideways together with minipage.


You can also use the \rotatebox command. Let's rotate a tabular inside a table for example:



Default is sidewaysfigures/sidewaystables are oriented depending on page number in two-sided documents (takes two passes).

The rotating package takes the following options.

In single-sided documents turn sidewaysfigures/sidewaystables counterclockwise.
In single-sided documents turn sidewaysfigures/sidewaystables clockwise (default).
In two-sided documents all sidewaysfigures/sidewaystables are same orientation (left of figure, table now bottom of page). This is the style preferred by the Chicago Manual of Style (broadside).
In two-sided documents all sidewaysfigures/sidewaystables are same orientation (left of figure, table now at top of page).

The rotfloat package

When it is desirable to place the rotated table at the exact location where it appears in the source (.tex) file, rotfloat package may be used. Then one can use


just like for normal tables. The H option can not be used without this package.


Tables are a common feature in academic writing, often used to summarize research results. Mastering the art of table construction in LaTeX is therefore necessary to produce quality papers and with sufficient practice one can print beautiful tables of any kind.

Keeping in mind that LaTeX is not a spreadsheet, it makes sense to use a dedicated tool to build tables and then to export these tables into the document. Basic tables are not too taxing, but anything more advanced can take a fair bit of construction; in these cases, more advanced packages can be very useful. However, first it is important to know the basics. Once you are comfortable with basic LaTeX tables, you might have a look at more advanced packages or the export options of your favorite spreadsheet. Thanks to the modular nature of LaTeX, the whole process can be automated in a fairly comfortable way.


LaTeX has built-in support to typeset tables and provides two environments: tabular and table. To typeset material in rows and columns, the tabular environment is needed; the optional table environment is a container for floating material similar to figure, into which a tabular environment may be included.

The table environment contains the caption and defines the float for the table, i.e., where in the document the table should be positioned and whether we want it to be displayed centered. The \caption and \label commands can be used in the same way as for pictures. For more information about the table environment, see the Floating with table section.

In any case, the actual content of the table is contained within the tabular environment.

The tabular environment

The tabular environment can be used to typeset tables with optional horizontal and vertical lines. LaTeX determines the width of the columns automatically.

The first line of the environment has the form:

\begin{tabular}[pos]{table spec}

The table spec argument tells LaTeX the alignment to be used in each column and the vertical lines to insert.

The number of columns does not need to be specified as it is inferred by looking at the number of arguments provided. It is also possible to add vertical lines between the columns here. The following symbols are available to describe the table columns (some of them require that the package array has been loaded):

l left-justified column
c centered column
r right-justified column
p{'width'} paragraph column with text vertically aligned at the top
m{'width'} paragraph column with text vertically aligned in the middle (requires array package)
b{'width'} paragraph column with text vertically aligned at the bottom (requires array package)
| vertical line
|| double vertical line

By default, if the text in a column is too wide for the page, LaTeX won’t automatically wrap it. Using p{'width'} you can define a special type of column which will wrap-around the text as in a normal paragraph. You can pass the width using any unit supported by LaTeX, such as 'pt' and 'cm', or command lengths, such as \textwidth. You can find a list in chapter Lengths.

The optional parameter pos can be used to specify the vertical position of the table relative to the baseline of the surrounding text. In most cases, you will not need this option. It becomes relevant only if your table is not in a paragraph of its own. You can use the following letters:

b bottom
c center (default)
t top

To specify a font format (such as bold, italic, etc.) for an entire column, you can add >{\format} before you declare the alignment. For example \begin{tabular}{ >{\bfseries}l c >{\itshape}r } will indicate a three column table with the first one aligned to the left and in bold font, the second one aligned in the center and with normal font, and the third aligned to the right and in italic. The "array" package needs to be activated in the preamble for this to work.

In the first line you have pointed out how many columns you want, their alignment and the vertical lines to separate them. Once in the environment, you have to introduce the text you want, separating between cells and introducing new lines. The commands you have to use are the following:

& column separator
\\ start new row (additional space may be specified after \\ using square brackets, such as \\[6pt])
\hline horizontal line
\newline start a new line within a cell (in a paragraph column)
\cline{i-j} partial horizontal line beginning in column i and ending in column j

Note, any white space inserted between these commands is purely down to one's preferences. I personally add spaces between to make it easier to read.

Basic examples

This example shows how to create a simple table in LaTeX. It is a three-by-three table, but without any lines.

\begin{tabular}{ l c r }
  1 & 2 & 3 \\
  4 & 5 & 6 \\
  7 & 8 & 9 \\


Expanding upon that by including some vertical lines:

\begin{tabular}{ l | c | r }
  1 & 2 & 3 \\
  4 & 5 & 6 \\
  7 & 8 & 9 \\


To add horizontal lines to the very top and bottom edges of the table:

\begin{tabular}{ l | c | r }
  1 & 2 & 3 \\
  4 & 5 & 6 \\
  7 & 8 & 9 \\


And finally, to add lines between all rows, as well as centering (notice the use of the center environment - of course, the result of this is not obvious from the preview on this web page):

  \begin{tabular}{ l | c | r }
    1 & 2 & 3 \\ \hline
    4 & 5 & 6 \\ \hline
    7 & 8 & 9 \\


  \begin{tabular}{ | l | c | r }
    1 & 2 & 3 \\ \hline
    4 & 5 & 6 \\ \hline
    7 & 8 & 9 \\


  7C0 & hexadecimal \\
  3700 & octal \\ \cline{2-2}
  11111000000 & binary \\
  \hline \hline
  1984 & decimal \\


Text wrapping in tables

LaTeX's algorithms for formatting tables have a few shortcomings. One is that it will not automatically wrap text in cells, even if it overruns the width of the page. For columns that will contain text whose length exceeds the column's desired width, it is recommended that you use the p attribute and specify the desired width of the column (although it may take some trial-and-error to get the result you want). For a more convenient method, have a look at The tabularx package, or The tabulary package.

Instead of p, use the m attribute to have the lines aligned toward the middle of the box or the b attribute to align along the bottom of the box.

Here is a simple example. The following code creates two tables with the same code; the only difference is that the last column of the second one has a defined width of 5 centimeters, while in the first one we didn't specify any width. Compiling this code:



Without specifying width for last column:
    \begin{tabular}{| l | l | l | l |}
    Day & Min Temp & Max Temp & Summary \\ \hline
    Monday & 11C & 22C & A clear day with lots of sunshine.
    However, the strong breeze will bring down the temperatures. \\ \hline
    Tuesday & 9C & 19C & Cloudy with rain, across many northern regions. Clear spells 
    across most of Scotland and Northern Ireland, 
    but rain reaching the far northwest. \\ \hline
    Wednesday & 10C & 21C & Rain will still linger for the morning. 
    Conditions will improve by early afternoon and continue 
    throughout the evening. \\

With width specified:
    \begin{tabular}{ | l | l | l | p{5cm} |}
    Day & Min Temp & Max Temp & Summary \\ \hline
    Monday & 11C & 22C & A clear day with lots of sunshine.  
    However, the strong breeze will bring down the temperatures. \\ \hline
    Tuesday & 9C & 19C & Cloudy with rain, across many northern regions. Clear spells 
    across most of Scotland and Northern Ireland, 
    but rain reaching the far northwest. \\ \hline
    Wednesday & 10C & 21C & Rain will still linger for the morning. 
    Conditions will improve by early afternoon and continue 
    throughout the evening. \\


You get the following output:


Note that the first table has been cropped, since the output is wider than the page width.

Manually broken paragraphs in table cells

Sometimes it is necessary to not rely on the breaking algorithm when using the p specifier, but rather specify the line breaks by hand. In this case it is easiest to use a \parbox:

  boring cell content & \parbox[t]{5cm}{rather long par\\new par}

Here the t argument controls the placement of the text inside the box. Other allowed values are c for center and b for bottom.

Space between columns

To tweak the space between columns (LaTeX will by default choose very tight columns), one can alter the column separation: \setlength{\tabcolsep}{5pt}. The default value is 6pt.

One can also introduce a horizontal space with hspace like this: \begin{tabular}{r@{\hspace{1in}}rr} There are different options for the hspace length.

Space between rows

Re-define the \arraystretch command to set the space between rows:


Default value is 1.0.

An alternative way to adjust the rule spacing is to add \noalign{\smallskip} before or after the \hline and \cline{i-j} commands:

\begin{tabular}{ | l | l | r | }
  \multicolumn{2}{c}{Item} \\
  Animal & Description & Price (\$) \\
  Gnat  & per gram & 13.65 \\
        & each     &  0.01 \\
  Gnu   & stuffed  & 92.50 \\
  Emu   & stuffed  & 33.33 \\
  Armadillo & frozen & 8.99 \\

You may also specify the skip after a line explicitly using glue after the line terminator

Mineral & Color \\[1cm]
Ruby & red \\
Sapphire & blue \\

Other environments inside tables

If you use some LaTeX environments inside table cells, like verbatim or enumerate:

\begin{tabular}{c c}
	& description
 	\\ \hline

you might encounter errors similar to

! LaTeX Error: Something's wrong--perhaps a missing \item.

To solve this problem, change column specifier to "paragraph" (p, m or b).

\begin{tabular}{m{5cm} c}

Defining multiple columns

It is possible to define many identically aligned columns at once using the *{num}{str} syntax. This is particularly useful when your table has many columns.

Here is a table with six centered columns flanked by a single column on each side:

Team              & P & W & D & L & F  & A & Pts \\
Manchester United & 6 & 4 & 0 & 2 & 10 & 5 & 12  \\
Celtic            & 6 & 3 & 0 & 3 &  8 & 9 &  9  \\
Benfica           & 6 & 2 & 1 & 3 &  7 & 8 &  7  \\
FC Copenhagen     & 6 & 2 & 1 & 3 &  5 & 8 &  7  \\


Column specification using >{\cmd} and <{\cmd}

The column specification can be altered using the array package. This is done in the argument of the tabular environment using >{\command} for commands executed right before each column element and <{\command} for commands to be executed right after each column element. As an example: to get a column in math mode enter: \begin{tabular}{>{$}c<{$}}. Another example is changing the font: \begin{tabular}{>{\small}c} to print the column in a small font.

The argument of the > and < specifications must be correctly balanced when it comes to { and } characters. This means that >{\bfseries} is valid, while >{\textbf} will not work and >{\textbf{} is not valid. If there is the need to use the text of the table as an argument (for instance, using the \textbf to produce bold text), one should use the \bgroup and \egroup commands: >{\textbf\bgroup}c<{\egroup} produces the intended effect. This works only for some basic LaTeX commands. For other commands, such as \underline to underline text, it is necessary to temporarily store the column text in a box using lrbox. First, you must define such a box with \newsavebox{\boxname} and then you can define:

>{\begin{lrbox}{\boxname} }%
  \underline{\unhbox\boxname} }%

This stores the text in a box and afterwards, takes the text out of the box with \unhbox (this destroys the box, if the box is needed again one should use \unhcopy instead) and passing it to \underline. (For LaTeX2e, you may want to use \usebox{\boxname} instead of \unhbox\boxname.)

This same trick done with \raisebox instead of \underline can force all lines in a table to have equal height, instead of the natural varying height that can occur when e.g. math terms or superscripts occur in the text.

Here is an example showing the use of both p{...} and >{\centering} :

\begin{tabular}{>{\centering}p{3.5cm}>{\centering}p{3.5cm} }
Geometry  & Algebra
 Points & Addition 
 Spheres & Multiplication 

Note the use of \tabularnewline instead of \\ to avoid a Misplaced \noalign error.

@ and ! expressions

The column separator can be specified with the @{...} or !{...} constructs.

It typically takes some text as its argument, and when appended to a column, it will automatically insert that text into each cell in that column before the actual data for that cell. The @{...} command kills the inter-column space and replaces it with whatever is between the curly braces. To keep the initial space, use !{...}. To add space, use @{\hspace{''width''}}.

Admittedly, this is not that clear, and so will require a few examples to clarify. Sometimes, it is desirable in scientific tables to have the numbers aligned on the decimal point. This can be achieved by doing the following:

  3   & 14159 \\
  16  & 2     \\
  123 & 456   \\


The space-suppressing qualities of the @-expression actually make it quite useful for manipulating the horizontal spacing between columns. Given a basic table, and varying the column descriptions:

\begin{tabular}{ |l|l| }
  stuff & stuff \\ \hline
  stuff & stuff \\








Aligning columns at decimal points using dcolumn

Instead of using @-expressions to build columns of decimals aligned to the decimal point (or equivalent symbol), it is possible to center a column on the decimal separator using the dcolumn package, which provides a new column specifier for floating point data. See the dcolumn package documentation for more information, but a simple way to use dcolumn is as follows.

\newcolumntype{d}[1]{D{.}{\cdot}{#1} }
%the argument for d specifies the maximum number of decimal places
\begin{tabular}{l r c d{1} }


A negative argument provided for the number of decimal places in the new column type allows unlimited decimal places, but may result in rather wide columns. Rounding is not applied, so the data to be tabulated should be adjusted to the number of decimal places specified. Note that a decimal aligned column is typeset in math mode, hence the use of \mathrm for the column heading in the example above. Also, text in a decimal aligned column (for example the header) will be right-aligned before the decimal separator (assuming there's no decimal separator in the text). While this may be fine for very short text, or numeric column headings, it looks cumbersome in the example above. A solution to this is to use the \multicolumn command described below, specifying a single column and its alignment. For example to center the header Decimal over its column in the above example, the first line of the table itself would be Left&Right&Center&\multicolumn{1}{c}{Decimal}\\

Bold text and dcolumn

To draw attention to particular entries in a table, it may be nice to use bold text. Ordinarily this is easy, but as dcolumn needs to see the decimal point it is rather harder to do. In addition, the usual bold characters are wider than their normal counterparts, meaning that although the decimals may align nicely, the figures (for more than 2--3 digits on one side of the decimal point) will be visibly misaligned. It is however possible to use normal width bold characters and define a new bold column type, as shown below.[1]

%here we're setting up a version of the math fonts with normal x-width
\SetSymbolFont{operators}{nxbold}{OT1}{cmr} {b}{n}
\SetSymbolFont{letters}  {nxbold}{OML}{cmm} {b}{it}
\SetSymbolFont{symbols}  {nxbold}{OMS}{cmsy}{b}{n}

\newcolumntype{d}{D{.}{.}{-1} } %decimal column as before
%wide bold decimal column
\newcolumntype{B}[3]{>{\boldmath\DC@{#1}{#2}{#3} }c<{\DC@end} } 
%normal width bold decimal column
\newcolumntype{Z}[3]{>{\mathversion{nxbold}\DC@{#1}{#2}{#3} }c<{\DC@end} } 
\begin{tabular}{l l d}
    Type &M & \multicolumn{1}{c}{N} \\
    Normal & 1 & 22222.222 \\
    Bold (standard)&10 & \multicolumn{1}{B{.}{.}{-1} }{22222.222}\\
    Bold (nxbold)&100 & \multicolumn{1}{Z{.}{.}{-1} }{22222.222}\\


Row specification

It might be convenient to apply the same command over every cell of a row, just as for column. Unfortunately the tabular environment cannot do that by default. We will need tabu instead, which provides the \rowfont option.

\rowfont{\bfseries\itshape\large} Header1 & Header2 \\
Cell2 & Cell2


To complete this tutorial, we take a quick look at how to generate slightly more complex tables. Unsurprisingly, the commands necessary have to be embedded within the table data itself.

Rows spanning multiple columns

The command for this looks like this: \multicolumn{num_cols}{alignment}{contents}. num_cols is the number of subsequent columns to merge; alignment is either l, c, r, or to have text wrapping specify a width p{5.0cm} . And contents is simply the actual data you want to be contained within that cell. A simple example:

\begin{tabular}{ |l|l| }
  \multicolumn{2}{|c|}{Team sheet} \\
  GK & Paul Robinson \\
  LB & Lucas Radebe \\
  DC & Michael Duberry \\
  DC & Dominic Matteo \\
  RB & Dider Domi \\
  MC & David Batty \\
  MC & Eirik Bakke \\
  MC & Jody Morris \\
  FW & Jamie McMaster \\
  ST & Alan Smith \\
  ST & Mark Viduka \\


Columns spanning multiple rows

The first thing you need to do is add \usepackage{multirow} to the preamble[2]. This then provides the command needed for spanning rows: \multirow{''num_rows''}{''width''}{''contents''}. The arguments are pretty simple to deduce (* for the width means the content's natural width).


\begin{tabular}{ |l|l|l| }
\multicolumn{3}{ |c| }{Team sheet} \\
Goalkeeper & GK & Paul Robinson \\ \hline
\multirow{4}{*}{Defenders} & LB & Lucas Radebe \\
 & DC & Michael Duburry \\
 & DC & Dominic Matteo \\
 & RB & Didier Domi \\ \hline
\multirow{3}{*}{Midfielders} & MC & David Batty \\
 & MC & Eirik Bakke \\
 & MC & Jody Morris \\ \hline
Forward & FW & Jamie McMaster \\ \hline
\multirow{2}{*}{Strikers} & ST & Alan Smith \\
 & ST & Mark Viduka \\


The main thing to note when using \multirow is that a blank entry must be inserted for each appropriate cell in each subsequent row to be spanned.

If there is no data for a cell, just don't type anything, but you still need the "&" separating it from the next column's data. The astute reader will already have deduced that for a table of   columns, there must always be   ampersands in each row (unless \multicolumn is also used).

Spanning in both directions simultaneously

Here is a nontrivial example of how to use spanning in both directions simultaneously and have the borders of the cells drawn correctly:


& & \multicolumn{4}{ c| }{Primes} \\ \cline{3-6}
& & 2 & 3 & 5 & 7 \\ \cline{1-6}
\multicolumn{1}{ |c  }{\multirow{2}{*}{Powers} } &
\multicolumn{1}{ |c| }{504} & 3 & 2 & 0 & 1 &     \\ \cline{2-6}
\multicolumn{1}{ |c  }{}                        &
\multicolumn{1}{ |c| }{540} & 2 & 3 & 1 & 0 &     \\ \cline{1-6}
\multicolumn{1}{ |c  }{\multirow{2}{*}{Powers} } &
\multicolumn{1}{ |c| }{gcd} & 2 & 2 & 0 & 0 & min \\ \cline{2-6}
\multicolumn{1}{ |c  }{}                        &
\multicolumn{1}{ |c| }{lcm} & 3 & 3 & 1 & 1 & max \\ \cline{1-6}


The command \multicolumn{1}{ is just used to draw vertical borders both on the left and on the right of the cell. Even when combined with \multirow{2}{*}{...}, it still draws vertical borders that only span the first row. To compensate for that, we add \multicolumn{1}{ in the following rows spanned by the multirow. Note that we cannot just use \hline to draw horizontal lines, since we do not want the line to be drawn over the text that spans several rows. Instead we use the command \cline{2-6} and opt out the first column that contains the text "Powers".

Here is another example exploiting the same ideas to make the familiar and popular "2x2" or double dichotomy:

\begin{tabular}{ r|c|c| }
 &  \multicolumn{1}{c}{noninteractive}
 & \multicolumn{1}{c}{interactive} \\
massively multiple & Library & University \\
one-to-one & Book & Tutor \\


Controlling table size

Resize tables

The graphicx packages features the command \resizebox{width}{height}{object} which can be used with tabular to specify the height and width of a table. The following example shows how to resize a table to 8cm width while maintaining the original width/height ratio.

% ...

\resizebox{8cm}{!} {

Resizing table including the caption

& \multicolumn{1}{c}{noninteractive}
& \multicolumn{1}{c}{interactive} \\
massively multiple & Library & University \\
one-to-one & Book & Tutor \\
\caption[Table caption text]{Table taken from \cite[p.10]{refid} }
\end{minipage} }

Alternatively you can use \scalebox{ratio}{object} in the same way but with ratios rather than fixed sizes:

% ...


Changing font size

A table can be globally switched to a different font size by simply adding the desired size command (here: \footnotesize) in the table scope, which may be after the \begin{table} statement if you use floats, otherwise you need to add a group delimiter.

  \begin{tabular}{| r | r || c | c | c |}
      % ...
  \caption{Performance at peak F-measure}
  \begin{tabular}{| r | r || c | c | c |}
      % ...

Alternatively, you can change the default font for all the tables in your document by placing the following code in the preamble:


See Fonts for named font sizes. The table caption font size is not affected. To control the caption font size, see Caption Styles.

It is also possible to change the vertical space between rows using \renewcommand{\arraystretch}{0.8} before \begin{tabular}.


Alternate row colors in tables

The xcolor package provides the necessary commands to produce tables with alternate row colors, when loaded with the table option. The command \rowcolors{<''starting row''>}{<''odd color''>}{<''even color''>} has to be specified right before the tabular environment starts.





odd 	& odd 	& odd \\
even 	& even 	& even\\
odd 	& odd 	& odd \\
even 	& even 	& even\\


The command \hiderowcolors is available to deactivate highlighting from a specified row until the end of the table. Highlighting can be reactivated within the table via the \showrowcolors command. If while using these commands you experience "misplaced \noalign errors" then use the commands at the very beginning or end of a row in your tabular.

\hiderowcolors odd & odd & odd \\


odd & odd & odd \\ \showrowcolors

Colors of individual cells

As above this uses the xcolor package.

% Include this somewhere in your document

% Enter this in the cell you wish to color a light grey.
% NB: the word 'gray' here denotes the grayscale color scheme, not the color grey. '0.9' denotes how dark the grey is.
% The following will color the cell red.

Width and stretching

We keep providing documentation for tabular* and tabularx.

The tabular* environment

This is basically a slight extension on the original tabular version, although it requires an extra argument (before the column descriptions) to specify the preferred width of the table.

\begin{tabular*}{0.75\textwidth}{ | c | c | c | r | }
  label 1 & label 2 & label 3 & label 4 \\
  item 1  & item 2  & item 3  & item 4  \\

However, that may not look quite as intended. The columns are still at their natural width (just wide enough to fit their contents) while the rows are as wide as the table width specified. If you do not like this default, you must also explicitly insert extra column space. LaTeX has rubber lengths, which, unlike others, are not fixed. LaTeX can dynamically decide how long the lengths should be. So, an example of this is the following.

\begin{tabular*}{0.75\textwidth}{@{\extracolsep{\fill} } | c | c | c | r | }
  label 1 & label 2 & label 3 & label 4 \\
  item 1  & item 2  & item 3  & item 4  \\

You will notice the @{...} construct added at the beginning of the column description. Within it is the \extracolsep command, which requires a width. A fixed width could have been used. However, by using a rubber length, such as \fill, the columns are automatically spaced evenly.

The tabularx package

This package provides a table environment called tabularx, which is similar to the tabular* environment except that it has a new column specifier X (in uppercase). The column(s) specified with this specifier will be stretched to make the table as wide as specified, greatly simplifying the creation of tables.

% ...

\begin{tabularx}{\textwidth}{ |X|X|X|X| }
  label 1 & label 2 & label 3 & label 4 \\
  item 1  & item 2  & item 3  & item 4  \\

The content provided for the boxes is treated as for a p column, except that the width is calculated automatically. If you use the package array, you may also apply any >{\cmd} or <{\cmd} command to achieve specific behavior (like \centering, or \raggedright\arraybackslash) as described previously.

Another option is to use \newcolumntype to format selected columns in a different way. It defines a new column specifier, e.g. R (in uppercase). In this example, the second and fourth column is adjusted in a different way (\raggedleft):

% ...

\begin{tabularx}{\textwidth}{ |l|R|l|R| }
  label 1 & label 2 & label 3 & label 4 \\
  item 1  & item 2  & item 3  & item 4  \\

Tabularx with rows spanning multiple columns using \multicolumn. The two central columns are posing as one by using the X@{} option. Note that the \multicolumn width (which in this example is 2) should equal the (in this example 1+1) width of the spanned columns:

% ...

\begin{tabularx}{1\textwidth}{ |>{\setlength\hsize{1\hsize}\centering}X|>{\setlength\hsize{1\hsize}\raggedleft}X@{} >{\setlength\hsize{1\hsize}\raggedright}X|>{\setlength\hsize{1\hsize}\centering}X| } 
Label 1 & \multicolumn{2}{>{\centering\setlength\hsize{2\hsize} }X|}{Label 2} & Label 3\tabularnewline
  123  & 123  & 456  & 123  \tabularnewline
  123  & 123  & 456  & 123  \tabularnewline

In a way analogous to how new commands with arguments can be created with \newcommand, new column types with arguments can be created with \newcolumntype as follows:

\usepackage[table]{xcolor} %Used to color the last column
% ...


\begin{tabularx}{\textwidth}{ | L{1} | R{0.5} | R{0.5} | C{2}{gray} | }
  label 1 & label 2 & label 3 & label 4 \\
  item 1  & item 2  & item 3  & item 4  \\

where since there are 4 columns, the sum of the \hsize's (1 + 0.5 + 0.5 + 2) must be equal to 4. The default value used by tabularx for \hsize is 1.

The tabulary package

tabulary is a modified tabular* allowing width of columns set for equal heights. tabulary allows easy and convenient writing of well balanced tables.

The problem with tabularx is that it leaves much blank if your cells are almost empty. Besides, it is not easy to have different column sizes.

tabulary tries to balance the column widths so that each column has at least its natural width, without exceeding the maximum length.


    Short sentences      & \#  & Long sentences                                                 \\
    This is short.       & 173 & This is much loooooooonger, because there are many more words.  \\
    This is not shorter. & 317 & This is still loooooooonger, because there are many more words. \\

The first parameter is the maximum width. tabulary will try not to exceed it, but it will not stretch to it if there is not enough content, contrary to tabularx.

The second parameter is the column disposition. Possible values are those from the tabular environment, plus

L left-justified balanced column
C centered balanced column
R right-justified balanced column
J left-right-justified balanced column

These are all capitals.

The tabu environment

It works pretty much like tabularx.

Note: tabu is currently broken and unmaintained

\begin{tabu} to \linewidth {llX[2]lllXl}
% ...

to \linewidth specifies the target width. The X parameter can have an optional span factor.

Table across several pages

Long tables are natively supported by LaTeX thanks to the longtable environment, which replaces both the tabular and table environments or rather combines them into a single environment. Unfortunately, this environment does not support stretching (X columns).

The tabu packages provides the longtabu environment. It has most of the features of tabu, with the additional capability to span multiple pages.

LaTeX can do well with long tables: you can specify a header that will repeat on every page, a header for the first page only, and the same for the footer.

\begin{longtabu} to \linewidth {lX[2]lXl}

\rowfont\bfseries H1 & H2 & H3 & H4 & H5 \\ \hline 

\\ \hline
\multicolumn{5}{r}{There is more to come} \\

\\ \hline

% Content ...

It uses syntax similar to longtable, so you should have a look at its documentation if you want to know more.

Alternatively you can try one of the following packages supertabular or xtab, an extended and somewhat improved version of supertabular.

Partial vertical lines

Adding a partial vertical line to an individual cell:

\begin{tabular}{ l c r }
  1 & 2 & 3 \\ \hline
  4 & 5 & \multicolumn{1}{r|}{6}  \\ \hline
  7 & 8 & 9 \\ \hline


Removing part of a vertical line in a particular cell:

\begin{tabular}{ | l | c | r | }
  1 & 2 & 3 \\ \hline
  4 & 5 & \multicolumn{1}{r}{6} \\ \hline
  7 & 8 & 9 \\ \hline


Vertically centered images

Inserting images into a table row will align it at the top of the cell. By using the array package this problem can be solved. Defining a new columntype will keep the image vertically centered.

\newcolumntype{V}{>{\centering\arraybackslash} m{.4\linewidth} }

Or use a parbox to center the image.

\parbox[c]{1em}{\includegraphics{image.png} }

A raisebox works as well, also allowing to manually fine-tune the alignment with its first parameter.

\raisebox{-.5\height}{\includegraphics{image.png} }

Footnotes in tables

The tabular environment does not handle footnotes properly. The longtable fixes that.

Professional tables

Many professionally typeset books and journals feature simple tables, which have appropriate spacing above and below lines, and almost never use vertical rules. Many examples of LaTeX tables (including this Wikibook) showcase the use of vertical rules (using "|"), and double-rules (using \hline\hline or "||"), which are regarded as unnecessary and distracting in a professionally published form. The booktabs package is useful for easily providing this professionalism in LaTeX tables, and the documentation also provides guidelines on what constitutes a "good" table.

In brief, the package uses \toprule for the uppermost rule (or line), \midrule for the rules appearing in the middle of the table (such as under the header), and \bottomrule for the lowermost rule. This ensures that the rule weight and spacing are acceptable. In addition, \cmidrule can be used for mid-rules that span specified columns. The following example contrasts the use of booktabs and two equivalent normal LaTeX implementations (the second example requires \usepackage{array} or \usepackage{dcolumn}, and the third example requires \usepackage{booktabs} in the preamble).

Normal LaTeX

\multicolumn{5}{c}{Item} \\
Animal    & Description & Price (\$) \\
Gnat      & per gram    & 13.65      \\
          & each        & 0.01       \\
Gnu       & stuffed     & 92.50      \\
Emu       & stuffed     & 33.33      \\
Armadillo & frozen      & 8.99       \\


Using array

%or \usepackage{dcolumn}
\multicolumn{2}{c}{Item} \\
Animal    & Description & Price (\$) \\
Gnat      & per gram    & 13.65      \\
          & each        & 0.01       \\
Gnu       & stuffed     & 92.50      \\
Emu       & stuffed     & 33.33      \\
Armadillo & frozen      & 8.99       \\


Using booktabs

\multicolumn{2}{c}{Item} \\
Animal    & Description & Price (\$) \\
Gnat      & per gram    & 13.65      \\
          & each        & 0.01       \\
Gnu       & stuffed     & 92.50      \\
Emu       & stuffed     & 33.33      \\
Armadillo & frozen      & 8.99       \\


Usually the need arises for footnotes under a table (and not at the bottom of the page), with a caption properly spaced above the table. These are addressed by the ctable package. It provides the option of a short caption given to be inserted in the list of tables, instead of the actual caption (which may be quite long and inappropriate for the list of tables). The ctable uses the booktabs package.

Sideways tables

Tables can also be put on their side within a document using the rotating or the rotfloat package. See the Rotations chapter.

Table with legend

To add a legend to a table the caption package can be used. With the caption package a \caption*{...} statement can be added besides the normal \caption{...}. Example:

  \begin{tabular}{| r | r || c | c | c |}


  \caption{A normal caption}
    A legend, even a table can be used
    \begin{tabular}{l l}
      item 1 & explanation 1 \\

The normal caption is needed for labels and references.

The eqparbox package

On rare occasions, it might be necessary to stretch every row in a table to the natural width of its longest line, for instance when one has the same text in two languages and wishes to present these next to each other with lines synching up. A tabular environment helps control where lines should break, but cannot justify the text, which leads to ragged right edges. The eqparbox package provides the command \eqmakebox which is like \makebox but instead of a width argument, it takes a tag. During compilation it bookkeeps which \eqmakebox with a certain tag contains the widest text and can stretch all \eqmakeboxes with the same tag to that width. Combined with the array package, one can define a column specifier that justifies the text in all lines:comand

  >{\begin{lrbox}{\tstretchbox} }%
  \eqmakebox[#1][s]{\unhcopy\tstretchbox} }%

See the documentation of the eqparbox package for more details.

The paracol package

The various tabular environments available for LaTeX are feature rich; however, they lack the ability to automatically page break large rows. The paracol package provides automatic page breaks in between rows and in certain cases can replace the tabular environment. Such situations could be common in documents that require translations and definitions, which may also includes lists.

For further detail see the documentation of the paracol package.

Floating with table

In WYSIWYG document processors, it is common to put tables in the middle of the text. This is what we have been doing until now. Professional documents, however, often make it a point to print tables on a dedicated page so that they do not disrupt the flow. From the point of view of the source code, one has no idea on which page the current text is going to lie, so it is hardly possible to guess which page may be appropriate for our table. LaTeX can automate this task by abstracting objects such as tables, pictures, etc., and deciding for us where they might fit best. This abstraction is called a float. Generally, an object that is floated will appear in the vicinity of its introduction in the source file, but one can choose to control its position also.

To tell LaTeX we want to use our table as a float, we need to put a table environment around the tabular environment, which is able to float and add a label and caption.

The table environment initiates a type of float just as the environment figure. In fact, the two bear a lot of similarities (positioning, captions, etc.). More information about floating environments, captions etc. can be found in Floats, Figures and Captions.

The environment names may now seem quite confusing. Let's sum it up:

  • tabular is for the content itself (columns, lines, etc.).
  • table is for the location of the table on the document, plus caption and label support.
\begin{table}[position specifier]
    ... your table ...
  \caption{This table shows some data}

In the table, we used a label, so now we can refer to it just like any other reference:


The table environment is also useful when you want to have a list of tables at the beginning or end of your document with the command


The captions now show up in the list of tables, if displayed.

You can set the optional parameter position specifier to define the position of the table, where it should be placed. The following characters are all possible placements. Using sequences of it define your "wishlist" to LaTeX.

h where the table is declared (here)
t at the top of the page
b at the bottom of the page
p on a dedicated page of floats
! override the default float restrictions. E.g., the maximum size allowed of a b float is normally quite small; if you want a large one, you need this ! parameter as well.

Default is tbp, which means that it is by default placed on the top of the page. If that's not possible, it's placed at the bottom if possible, or finally with other floating environments on an extra page.

You can force LaTeX to use one given position. E.g. [!h] forces LaTeX to place it exactly where you place it (Except when it's really impossible, e.g you place a table here and this place would be the last line on a page). Again, understand it correctly: it urges LaTeX to put the table at a specific place, but it will not be placed there if LaTeX thinks it will not look great. If you really want to place your table manually, do not use the table environment.

Centering the table horizontally works like everything else, using the \centering command just after opening the table environment, or by enclosing it with a center environment.

Using spreadsheets and data analysis tools

For complex or dynamic tables, you may want to use a spreadsheet. You might save lots of time by building tables using specialized software and exporting them in LaTeX format. The following plugins and libraries are available for some popular software:

However, copying the generated source code to your document is not convenient at all. For maximum flexibility, generate the source code to a separate file which you can input from your main document file with the \input command. If your spreadsheet supports command-line, you can generate your complete document (table included) in one command, using a Makefile for example.

See Modular Documents for more details.

Need more features?

Have a look at one of the following packages:

  • hhline: do whatever you want with horizontal lines
  • array: gives you more freedom on how to define columns
  • colortbl: make your table more colorful
  • threeparttable makes it possible to put footnotes both within the table and its caption
  • arydshln: creates dashed horizontal and vertical lines
  • ctable: allows for footnotes under table and properly spaced caption above (incorporates booktabs package)
  • slashbox: create 2D tables with the first cell containing a description for both axes. Not available in Tex Live 2011 or later.
  • diagbox: compatible to slashbox, come with Tex Live 2011 or later
  • dcolumn: decimal point alignment of numeric cells
  • rccol: advanced decimal point alignment of numeric cells with rounding
  • numprint: print numbers, in the current mode (text or math) in order to use the correct font, with separators, exponent and/or rounded to a given number of digits. tabular(*), array, tabularx, and longtable environments are supported using all features of numprint
  • spreadtab: spread sheets allowing the use of formulae
  • siunitx: alignment of tabular entries
  • pgfplotstable: loads, rounds, formats and postprocesses numerical tables, e.g. by importing the data directly from .csv (comma-separated value) files instead of manually writing the whole tables in LaTeX code. Programs such as Excel, LibreOffice Calc etc. can export data sheets as .csv files.


  1. D Carlisle. "Decimals in table don't align with dcolumn when bolded". Stackexchange.
  2. Package multirow on CTAN

Title creation

LaTeX/Title creation

Page Layout

LaTeX and the document class will normally take care of page layout issues for you. For submission to an academic publication, this entire topic will be out of your hands, as the publishers want to control the presentation. However, for your own documents, there are some obvious settings that you may wish to change: margins, page orientation and columns, to name but three. The purpose of this tutorial is to show you how to configure your pages.

We will often have to deal with TeX lengths in this chapter. You should have a look at Lengths for comprehensive details on the topic.

Two-sided documents

Documents can be either one- or two-sided. Articles are by default one-sided, books are two-sided. Two-sided documents differentiate the left (even) and right (odd) pages, whereas one-sided do not. The most notable effect can be seen in page margins. If you want to make the article class two-sided, use \documentclass[twoside]{article}.

Many commands and variables in LaTeX take this concept into account. They are referred to as even and odd. For one-sided document, only the odd commands and variables will be in effect.

Page dimensions

A page in LaTeX is defined by many internal parameters. Each parameter corresponds to the length of an element of the page, for example, \paperheight is the physical height of the page. Here you can see a diagram showing all the variables defining the page. All sizes are given in TeX points (pt), there are 72.27pt in an inch or 1pt ≈ 0.3515mm.

  1. one inch + \hoffset
  2. one inch + \voffset
  3. \oddsidemargin = 31pt
  4. \topmargin = 20pt
  5. \headheight = 12pt
  6. \headsep = 25pt
  7. \textheight = 592pt
  8. \textwidth = 390pt
  9. \marginparsep = 10pt
  10. \marginparwidth = 35pt
  11. \footskip = 30pt
  • \marginparpush = 7pt (not shown)
  • \hoffset = 0pt
  • \voffset = 0pt
  • \paperwidth = 597pt
  • \paperheight = 845pt

The current details plus the layout shape can be printed from a LaTeX document itself. Use the layout package and the command of the same name: \usepackage{layout} ... \layout{}

To render a frame marking the margins of a document you are currently working on, add


to the document.

Page size

It will not have been immediately obvious - because it doesn't really cause any serious problems - that the default page size for all standard document classes is US letter. This is shorter by 18 mm (about 3/4 inch), and slightly wider by 8 mm (about 1/4 inch), compared to A4 (which is the standard in almost all the rest of the world). While this is not a serious issue (most printers will print the document without any problems), it is possible to specify alternative sizes as class option. For A4 format:


More size options with geometry

One of the most versatile packages for page layout is the geometry package. The immediate advantage of this package is that it lets you customize the page size even with classes that do not support the options. For instance, to set the page size, add the following to your preamble:


The geometry package has many pre-defined page sizes, like a4paper, built in. Others include:

  • a0paper, a1paper, ..., a6paper,
  • b0paper, b1paper, ..., b6paper,
  • letterpaper,
  • legalpaper,
  • executivepaper.

To explicitly change the paper dimensions using the geometry package, the paperwidth and paperheight options can be used. For example:

\usepackage[paperwidth=5.5in, paperheight=8.5in]{geometry}

Package provides many flexibility on setting the page layout, including specifying specific layout of each page using the:

    key=val // package options 

While very flexible, this package also comes with limitations. For instance, page size cannot be provided to individual pages, which requires different approaches for the workaround.

Page size issues

If you intend to get a PDF in the end, there are basically three ways:

  • TeX → PDF
pdflatex myfile               # TeX → PDF
  • TeX → DVI → PDF
latex myfile                  # TeX → DVI
dvipdf myfile                 # DVI → PDF
  • TeX → DVI → PS → PDF
latex myfile                  # TeX → DVI
dvips myfile -o     # DVI → PS
ps2pdf myfile.pdf   # PS  → PDF

Sadly the PDF output page size may not be completely respectful of your settings. Some of these tools do not have the same interpretation of the DVI, PS and PDF specifications, and you may end up with a PDF which has not exactly the right size. Thankfully there is a solution to that: the \special command lets the user pass PostScript or PDF parameters, which can be used here to set the page size appropriately.

  • For pdflatex to work fine, using the package geometry usually works.
  • For the DVI and PS ways, the safest way to always get the right paper size in the end is to add

to the tex file, and to append the appropriate parameters to the processors used during output generation:

dvips -t a4 ...
ps2pdf -sPAPERSIZE=a4 ... # On Windows: ps2pdf -sPAPERSIZE#a4 ... [1]

If you want US Letter instead, replace 210mm,297mm by 8.5in,11in and a4paper by letter. Also replace a4 by letter in command-line parameters.

Page size for tablets

Those who want to read on tablets or other handheld digital devices need to create documents without the extra whitespace. In order to create PDF documents with optimal handheld viewing, not only must the text field and margins be adjusted, so must the page size. If you are looking for a sensible dimension, consider following the paper size used by the Supreme Court of the United States, 441pt by 666pt (or 6.125 inches by 9.25 inches), which looks great on tablets. You could also use the Supreme Court's text field size of 297 pt by 513 pt, but this is too wide for fonts other than Century Schoolbook, the font required by the Supreme Court.


Readers used to perusing typical physical literature are probably wondering why there is so much white space surrounding the text. For example, on A4 paper a document will typically have 44 mm margin widths on the left and right of the page, leaving about 60% of the page width for text. The reason is improved readability. Studies have shown[2][3] that it's easier to read text when there are 60–70 characters per line—and it would seem that 66 is the optimal number. Therefore, the page margins are set to ensure optimal readability, and excessive margin white space is tolerated as a consequence. Sometimes, this white space is left in the inner margin with the assumption that the document will be bound.

If you wish to avoid excessive white space, rather than changing the margins, consider instead using a two-column (or more) layout. This approach is the one usually taken by print magazines because it provides both readable line lengths and good use of the page. Another option for reducing the amount of whitespace on the page without changing the margins is to increase the font size using the 12pt option to the document class.

If you wish to change the margins of your document, there are many ways to do so:

  • One older approach is to use the fullpage package for somewhat standardized smaller margins (around an inch), but it creates lines of more than 100 characters per line with the 10pt default font size (and about 90 if the 12pt documentclass option is used):

For even narrower margins, the fullpage package has a cm option (around 1.5cm), which results in about 120 characters per line at the 10pt default font size, about double what is considered readable:

  • A more modern and flexible approach is to use the geometry package. This package allows you to specify the 4 margins without needing to remember the particular page dimensions commands. You can enter the measures in centimeters and inches as well. Use cm for centimeters and in for inches after each value (e.g. 1.0in or 2.54cm). Note that by default (i.e. without any options) this package already reduces the margins, so for a 'standard layout' you may not need to specify anything. These values are relative to the edge of paper (0in) and go inward. For example, this command provides more conventional margins, better using the vertical space of the page, without creating the dramatically long lines of the fullpage package (if the 11pt documentclass option is used, the line lengths are about 88 characters for letter-sized paper and slightly less when using a4paper).
\usepackage[top=1in, bottom=1.25in, left=1.25in, right=1.25in]{geometry}

It can also recreate the behavior of the fullpage package using


You can combine the margin options with the page size options seen in this paragraph.

  • If the page size is A4, you can use the layaureo package. The big option further narrows the margins.
  • You should not use the a4wide package for a page with A4 document size with smaller margins. It is obsolete and buggy. Use geometry package instead like this:
  • Edit individual page dimension variables described above, using the \addtolength and \setlength commands. See the Lengths chapter. For instance,

Odd and even margins

Using the geometry package, the options left and right are used for the inside and outside margins respectively. They also have aliases inner and outer. Thus, the easiest way to handle different margins for odd and even pages is to give the twoside option in the document class command and specify the margins as usually.

\usepackage[inner=4cm,outer=2cm]{geometry} %left=4cm,right=2cm would be equivalent

This will result in a value of 4cm on all inner margins (left margin for odd number pages and right margin for even pages) and 2cm margin on outer margins.

Setting the same value for the inner and outer for geometry will remove the difference between the margins. Another quick way to eliminate the difference in position between even and odd numbered pages would be setting the values to evensidemargin and oddsidemargin to the half of odd's default:


By default, the value of evensidemargin is larger than oddsidemargin in the two-sided layout, as one could wish to write notes on the side of the page. The side for the large margin is chosen opposite to the side where pages are joined together.

See the Lengths.

Top margin above Chapter

The top margin above a chapter can be changed using the titlesec package. Example: [6]

\titleformat{\chapter}[display]{\normalfont\huge\bfseries}{\chaptertitlename\ \thechapter}{20pt}{\Huge}

The command \titleformat must be used when the spacing of a chapter is changed. In case of a section this command can be omitted.

Page orientation

When you talk about changing page orientation, it usually means changing to landscape mode, since portrait is the default. We shall introduce two slightly different styles of changing orientation.

Change orientation of the whole document

The first is for when you want all of your document to be in landscape from the very beginning. There are various packages available to achieve this, but the one we prefer is the geometry package. All you need to do is call the package, with landscape as an option:


Although, if you intend to use geometry to set your paper size, don't add the \usepackage commands twice, simply string all the options together, separating with a comma:


Using standard LaTeX classes, you can use the same class options:


Change orientation of specific part

The second method is for when you are writing a document in portrait, but you have some contents, like a large diagram or table that would be displayed better on a landscape page. However, you still want the consistency of your headers and footers appearing the same place as the other pages.

The lscape package is for this very purpose. It supplies a landscape environment, and anything inside is basically rotated. No actual page dimensions are changed. This approach is more applicable to books or reports than to typical academic publications. Using pdflscape instead of lscape when generating a PDF document will make the page appear right side up when viewed: the single page that is in landscape format will be rotated, while the rest will be left in portrait orientation.

Also, to get a table to appear correctly centered on a landscaped page, one must place the tabular environment inside a table environment, which is itself inside the landscape environment. For instance it should look like this:

% ...

\centering     % optional, probably makes it look better to have it centered on the page
% ...

For books (and in general documents using the twoside option), the landscape-environment unfortunately does not pay attention to the different layout of even and odd pages. The macro can be fixed using a few lines of extra code in the preamble[4].

Change orientation of floating environment

If you use the above code, you will see that the table is inserted where it is in the code. It will not be floated! To fix this you need the package rotating. See the Rotations chapter.

Margins, page size and rotation of a specific page

If you need to rotate the page so that the figure fits, the chances are good that you need to scale the margins and the font size too. Again, the geometry package comes in handy for specifying new margins for a single page only.

% ...

\thispagestyle{empty} %% Remove header and footer.

\footnotesize %% Smaller font size.

% ...



Note that order matters!

Page background

There are many ways on implementing background for a page. Below is most common packages used:

  • eso-pic package will let you print content in the background of every page or individual pages.
\usepackage{tikz} % for \gradientbox below.

    \node[left color=#1,right color=#2] {#3};

          \hspace*{ \stretch{1} }\textcopyright~2013 \makeatletter\@author\makeatother.\hspace*{ \stretch{1} }

The starred-version of the \AddToShipoutPicture command applies to the current page only.

  • background package let users include watermarks and pictures in the background.

  opacity = 0.5,
  angle = 0,
  contents = {\includegraphics{example.pdf} }

Multi-column pages

Using the twocolumn optional class argument

Using a standard Latex document class, like article, you can simply pass the optional argument twocolumn to the document class: \documentclass[twocolumn]{article} which will give the desired effect.

While this approach is useful, it has limitations. The multicol package provides the following advantages:

  • Can support up to ten columns.
  • Implements a multicols environment, therefore, it is possible to mix the number of columns within a document.
  • Additionally, the environment can be nested inside other environments, such as figure.
  • multicol outputs balanced columns, whereby the columns on the final page will be of roughly equal length.
  • Use multicols* environment for unbalanced columns, whereby each column is completely filled before starting with the next column.
  • Vertical rules between columns can be customised.
  • Column environments can be easily customised locally or globally.

Using multicol package

The multicol package overcomes some of the shortcomings of twocolumn and provides the multicol environment. To create a typical two-column layout:

  lots of text

Floats are not fully supported by this environment. It can only cope if you use the starred forms of the float commands (e.g., \begin{figure*} ) which makes the float span all columns. This is not hugely problematic, since floats of the same width as a column may be too small, and you would probably want to span them anyway. See this section for a more detailed discussion.

The multicol package has two important parameters which can be set using \setlength:

  • \columnseprule, sets the width of the vertical rule between columns and defaults to 0pt
  • \columnsep, sets the horizontal space between columns and the defaults to 10pt, which is quite narrow

To force a break in a column, the command \columnbreak is used.

Manual page formatting

There may be instances, especially in very long documents, such as books, that LaTeX will not get all page breaks looking as good as it could. It may, therefore, be necessary to manually tweak the page formatting. Of course, you should only do this at the very final stage of producing your document, once all the content is complete. LaTeX offers the following:

\newpage Ends the current page and starts a new one.
\pagebreak[number] Breaks the current page at the point of the command. The optional number argument sets the priority in a scale from 0 to 4.
\nopagebreak[number] Stops the page being broken at the point of the command. The optional number argument sets the priority in a scale from 0 to 4.
\clearpage Ends the current page and causes any floats encountered in the input, but yet to appear, to be printed.

Widows and orphans

In professional books, it's not desirable to have single lines at the beginning or end of a page. In typesetting such situations are called 'widows' and 'orphans'. Normally it is possible that widows and orphans appear in LaTeX documents. You can try to deal with them using manual page formatting, but there's also an automatic solution.

LaTeX has a parameter for 'penalty' for widows and orphans ('club lines' in LaTeX terminology). With the greater penalty LaTeX will try more to avoid widows and orphans. You can try to increase these penalties by putting following commands in your document preamble:


If this does not help, you can try increasing these values even more, to a maximum of 10000. However, it is not recommended to set this value too high, as setting it to 10000 forbids LaTeX from doing this altogether, which might result in strange behavior.

It also helps to have rubber band values for the space between paragraphs:

\setlength{\parskip}{3ex plus 2ex minus 2ex}

Alternatively, you can use the needspace package to reserve some lines and thus to prevent page breaking for those lines.


Troubleshooting / Debugging

A very useful troubleshooting and designing technique is to turn on the showframe option in the geometry package (which has the same effect as the showframe package described above). It draws bounding boxes around the major page elements, which can be helpful because the boundaries of various regions are usually invisible, and complicated by padding whitespace.


Examining log of the tex could be helpful as well - it should provide many useful information. In addition, following packages can help on visual debugging layout:

% \layout % - place it inside "document" section to see the layout where you need it more conveniently

Which will print layout and frames. The statements could be on top of the source file and uncommented when required.

Notes and References

This page uses material from Andy Roberts' Getting to grips with LaTeX with permission from the author.

Importing Graphics

Importing external graphics

You can import external graphics using package graphicx. The most important command is \includegraphics. LaTeX itself treats the image like normal text, i.e. as a box of certain height and width.

The following image does not show any wombats


\includegraphics[height=3cm]{example-image-a} \includegraphics[width=5cm]{example-image-b}

The package documentation list the options width and height, as well as others.

Using pdflatex several graphics formats are supported: pdf, png and jpg. Modern installations of LaTeX can use eps files as well, but indirectly.

LaTeX in dvi-mode supports only eps-files.


You can align the images in a matrix. You just have to think of a proper width for the images.

\\[\baselineskip]% adds vertical line spacing

Three images in one line, each image has a width of just 30 % of the available line width, centered with respect to the text below.


If you want to add a caption and let LaTeX keep track of the numbering, have a look at the floats section.

Converting graphics


You should also take a look at Export To Other Formats for other possibilities.


You can convert EPS to PDF with the epstopdf utility, included in package of the same name. This tool is actually called by pdflatex to convert EPS files to PDF in the background when the graphicx package is loaded. This process is completely invisible to the user.

You can batch convert files using the command-line. In Bourne Shell (Unix) this can be done by:

$ for i in *.eps; do epstopdf "$i"; done

In Windows, multiple files can be converted by placing the following line in a batch file (a text file with a .bat extension) in the same directory as the images:

for %%f in (*.eps) do epstopdf %%f

which can then be run from the command line.

If epstopdf produces a whole page with your small graphics somewhere on it, use

$ epstopdf --gsopt=-dEPSCrop foo.eps

or try using the ps2pdf utility which should be installed with Ghostscript (required for any TeX distribution).

$ ps2pdf -dEPSCrop foo.eps

to crop the final PDF.


When all of the above fails, one can simplify the EPS file before attempting other conversions, by using the eps2eps tool (also see next section):

$ eps2eps input.eps input-e2.eps

This will convert all the fonts to pre-drawn images, which is sometimes desirable when submitting manuscripts for publication. However, on the downside, the fonts are NOT converted to lines, but instead to bitmaps, which reduces the quality of the fonts.


imgtops is a lightweight graphics utility for conversions between raster graphics (JPG, PNG, ...) and EPS/PS files.


Inkscape can also convert files from and to several formats, either from the GUI or from the command-line. For instance, to obtain a PDF from a SVG image you can do:

$ inkscape -z -D --file=input.svg --export-pdf=output.pdf

It is possible to run this from within a LaTeX file, the Template:LaTeX/package package (when running (pdf)latex with the --shell-escape option) can do this using Inkscape's pdf+tex export option, or a simple macro can be used. See How to include SVG diagrams in LaTeX? -- Stackexchange See Export To Other Formats for more details.


To properly edit an EPS file, you can convert it to an editable format using pstoedit. For instance, to get an Xfig-editable file, do:

$ pstoedit -f fig input.eps output.fig

And to get an SVG file (editable with any vector graphics tool like Inkscape) you can do:

$ pstoedit -f plot-svg input.eps output.svg

Sometimes pstoedit fails to create the target format (for example when the EPS file contains clipping information).


Under Windows, PDFCreator is an open source software that can create PDF as well as EPS files. It installs a virtual printer that can be accessed from other software having a "print..." entry in their menu (virtually any program).

Raster graphics converters

These three programs operate much the same way, and can convert between most graphics formats. Sam2p however is the most recent of the three and seems to offer both the best quality and to result in the smallest files.

PNG alpha channel

Acrobat Reader sometimes has problems with displaying colors correctly if you include graphics in PNG format with alpha channel. You can solve this problem by dropping the alpha channel. On Linux it can be achieved with convert from the ImageMagick program:

convert -alpha off input.png output.png

Converting a color EPS to grayscale

Sometimes color EPS figures need to be converted to black-and-white or grayscale to meet publication requirements. This can be achieved with the eps2eps of the Ghostscript package and [7] programs:

$ eps2eps input.eps input-e2.eps
$ pscol -0gray input-e2.eps input-gray.eps

Third-party graphics tools

We will not tackle the topic of procedural graphics created from within LaTeX code here (TikZ, PSTricks, MetaPost and friends). See Introducing Procedural Graphics for that.

You should prefer vector graphics over raster graphics for their quality. Raster graphics should only be used in case of photos. Diagrams of any sort should be vectors.

As we have seen before, LaTeX handles

  • EPS and PDF for vector graphics;
  • PNG and JPG for raster graphics.

If some tools cannot save in those formats, you may want to convert them before importing them.

Vector graphics


Dia is a cross platform diagramming utility which can export eps images, or generate tex drawn using the tikz package.


Another program for creating vector graphics is Inkscape. It can run natively under Windows, Linux or Mac OS X (with X11). It works with Scalable Vector Graphics (SVG) files, although it can export to many formats that can be included in LaTeX files, such as EPS and PDF. From version 0.48, there is a combined PDF/EPS/PS+LaTeX output option, similar to that offered by Xfig. There are instructions on how to save your vector images in a PDF format understood by LaTeX and have LaTeX manage the text styles and sizes in the image automatically.[1]. Today there is the svg package[2] which provides an \includesvg command to convert and include svg-graphics directly in your LaTeX document using Inkscape. You may have a look at this extended example too.

An extremely useful plug-in is textext, which can import LaTeX objects. This can be used for inserting mathematical notation or LaTeX fonts into graphics (which may then be imported into LaTeX documents).


The Ipe extensible drawing editor is a free vector graphics editor for creating figures in PDF or EPS format. Unlike Xfig, Ipe represents LaTeX fonts in their correct size on the screen which makes it easier to place text labels at the right spot. Ipe also has various snapping modes (for example, snapping to points, lines, or intersections) that can be used for geometric constructions.


Yet another solution is provided by the lpic packages [8], which allows TeX annotations to imported graphics. See Labels in the figures.


It is also possible to export vector graphics to EPS format using LibreOffice Draw, which is an open source office suite available for Windows, Linux and Mac.


Vector editor TpX separates geometric objects from text objects. Geometric objects are saved into .PDF file, the rest is saved in .TpX file to be processed by LaTeX. User just create the graphics in TpX editor and calls the .TpX file from latex file by command \input{...TpX}.


Xfig is a basic program that can produce vector graphics, which can be exported to LaTeX. It can be installed on Unix platforms.

On Microsoft Windows systems, Xfig can only be installed using Cygwin-X; however, this will require a fast internet connection and about 2 gigabytes of space on your computer. With Cygwin, to run Xfig, you need to first start the "Start X - Server", then launch "xterm" to bring up a terminal. In this terminal type "xfig" (without the quotation marks) and press return.

Alternatively, WinFIG. WinFIG is an attempt to achieve the functionality of xfig on Windows computers.

There are many ways to use xfig to create graphics for LaTeX documents. One method is to export the drawing as a LaTeX document. This method, however, suffers from various drawbacks: lines can be drawn only at angles that are multiples of 30 and 45 degrees, lines with arrows can only be drawn at angles that are multiples of 45 degrees, several curves are not supported, etc.

Exporting a file as PDF/LaTeX or PS/LaTeX, on the other hand, offers a good deal more flexibility in drawing. Here's how it's done:

  1. Create the drawing in xfig. Wherever you need LaTeX text, such as a mathematical formula, enter a LaTeX string in a textbox.
  2. Use the Edit tool to open the properties of each of those textboxes, and change the option on the "Special Flag" field to Special. This tells LaTeX to interpret these textboxes when it opens the figure.
  3. Go to File -> Export and export the file as PDF/LaTeX (both parts) or PS/LaTeX (both parts), depending on whether you are using pdflatex or pslatex to compile your file.
  4. In your LaTeX document, where the picture should be, use the following, where "test" is replaced by the name of the image:
     \caption{Your figure}

    Observe that this is just like including a picture, except that rather than using \includegraphics, we use \input. If the export was into PS/LaTeX, the file extension to include would be .pstex_t instead of .pdf_t.

  5. Make sure to include packages graphicx and color in the file, with the \usepackage command right below the \documentclass command, like this:

And you're done!

For more details on using xfig with LaTeX, this chapter of the xfig User Manual may prove helpful.

Other tools

Commercial vector graphics software, such as Adobe Illustrator, CorelDRAW, and FreeHand are commonly used and can read and write EPS figures. However, these products are limited to Windows and Mac OS X platforms.

Raster graphics

Adobe Photoshop

Photoshop can save to EPS.


GIMP, has a graphical user interface, and it is multi-platform. It can save to EPS and PDF.

Plots and Charts

Generic Mapping Tools (GMT)

Generic Mapping Tools (GMT), maps and a wide range of highly customisable plots.


Gnumeric, spreadsheets has SVG, EPS, PDF export


Gnuplot, producing scientific graphics since 1986. If you want to make mathematical plots, then Gnuplot can save in any format. You can get best results when used along PGF/TikZ.


matplotlib, plotting library written in python, with PDF and EPS export. On the other hand there is a PGF export also. There are some tricks to be able to import formats other than EPS into your DVI document, but they're very complicated. On the other hand, converting any image to EPS is very simple, so it's not worth considering them.


R, statistical and scientific figures.

Editing EPS graphics

As described above, graphics content can be imported into LaTeX from outside programs as EPS files. But sometimes you want to edit or retouch these graphics files. An EPS file can be edited with any text editor since it is formatted as ASCII. In a text editor, you can achieve simple operations like replacing strings, changing the bounding box, or moving items slightly, but anything further becomes cumbersome. Vector graphics editors, like Inkscape, may also be able to import EPS files for subsequent editing. This approach also for easier editing. However, the importing process may occasionally modify the original EPS image.

Notes and References

  1. Johan B. C. Engelen. "How to include an SVG image in LATEX" (PDF).
  2. Philip Ilten. "The svg package on CTAN".

Floats, Figures and Captions

The previous chapter introduced importing graphics. However, just having a picture stuck in between paragraphs does not look professional. To start with, we want a way of adding captions, and to be able to cross-reference. What we need is a way of defining figures. It would also be good if LaTeX could apply principles similar to when it arranges text to look its best to arrange pictures as well. This is where floats come into play.


Floats are containers for things in a document that cannot be broken over a page. LaTeX by default recognizes "table" and "figure" floats, but you can define new ones of your own (see Custom floats below). Floats are there to deal with the problem of the object that won't fit on the present page and to help when you really don't want the object here just now.

Floats are not part of the normal stream of text, but separate entities, positioned in a part of the page to themselves (top, middle, bottom, left, right, or wherever the designer specifies). They always have a caption describing them and they are always numbered so they can be referred to from elsewhere in the text. LaTeX automatically floats Tables and Figures, depending on how much space is left on the page at the point that they are processed. If there is not enough room on the current page, the float is moved to the top of the next page. This can be changed by moving the Table or Figure definition to an earlier or later point in the text, or by adjusting some of the parameters which control automatic floating.

Authors sometimes have many floats occurring in rapid succession, which raises the problem of how they are supposed to fit on the page and still leave room for text. In this case, LaTeX stacks them all up and prints them together if possible, or leaves them to the end of the chapter in protest. The skill is to space them out within your text so that they intrude neither on the thread of your argument or discussion nor on the visual balance of the typeset pages.

As with various other entities, there exist limitations on the number of unprocessed (placed) floats in line. LaTeX by default can cope with maximum 18 floats and a symptomatic error is:

! LaTeX Error: Too many unprocessed floats.

The morefloats package lifts this limit.


To create a figure that floats, use the figure environment.

\begin{figure}[placement specifier]
... figure contents ...

The previous section mentioned how floats are used to allow LaTeX to handle figures while maintaining the best possible presentation. However, there may be times when you disagree, and a typical example is with its positioning of figures. The placement specifier parameter exists as a compromise, and its purpose is to give the author a greater degree of control over where certain floats are placed.

Specifier Permission
h Place the float here, i.e., approximately at the same point it occurs in the source text (however, not exactly at the spot)
t Position at the top of the page.
b Position at the bottom of the page.
p Put on a special page for floats only.
! Override internal parameters LaTeX uses for determining "good" float positions.
H Places the float at precisely the location in the LaTeX code. Requires the float package,[1] i.e., \usepackage{float}.

What you do with these placement permissions is to list which of the options you wish to make available to LaTeX. These are simply possibilities, and LaTeX will decide when typesetting your document which of your supplied specifiers it thinks is best. Frank Mittelbach describes the algorithm[2]:

  • If a float is encountered, LaTeX attempts to place it immediately according to its rules (detailed later)
    • If this succeeds, the float is placed and that decision is never changed;
    • If this does not succeed, then LaTeX places the float into a holding queue to be reconsidered when the next page is started (but not earlier).
  • Once a page has finished, LaTeX examines this holding queue and tries to empty it as best as possible. For this, it will first try to generate as many float pages as possible (in the hope of getting floats off the queue). Once this possibility is exhausted, it will next try to place the remaining floats into top and bottom areas. It looks at all the remaining floats and either places them or defers them to a later page (i.e., re-adding them to the holding queue once more).
  • After that, it starts processing document material for this page. In the process, it may encounter further floats.
  • If the end of the document has been reached or if a \clearpage is encountered, LaTeX starts a new page, relaxes all restrictive float conditions, and outputs all floats in the holding queue by placing them on float page(s).

In some special cases, LaTeX won't follow these positioning parameters and additional commands will be necessary, for example, if one needs to specify an alignment other than centered for a float that sits alone in one page[3].

Use \listoffigures to add a list of the figures in the beginning of the document. To change the name used in the caption from Figure to Example, use \renewcommand{\figurename}{Example} in the figure contents. See the New commands section for more information on this.

Figures with borders

It's possible to get a thin border around all figures. You have to write the following once at the beginning of the document:


The border will not include the caption.


Floating tables are covered in a separate chapter. Let's give a quick reminder here. The tabular environment that was used to construct the tables is not a float by default. Therefore, for tables you wish to float, wrap the tabular environment within a table environment, like this:

  ... table data ...

You may feel that it is a bit long-winded, but such distinctions are necessary, because you may not want all tables to be treated as a float.

Use \listoftables to add a list of the tables in the beginning of the document.

Keeping floats in their place

The placeins[9] package provides the command \FloatBarrier, which can be used to prevent floats from being moved over it. This can, e.g., be useful at the beginning of each section. The package even provides an option to change the definition of \section to automatically include a \FloatBarrier. This can be set by loading the package with the option [section] (\usepackage[section]{placeins}). \FloatBarrier may also be useful to prevent floats intruding on lists created using itemize or enumerate.

The flafter package can be used to force floats to appear after they are defined, and the endfloat[10] package can be used to place all floats at the end of a document.

The float[11] package provides the H option to floating environments, which completely stops them from floating.

Package caption[12] provides the command \captionof{<type>}{<caption text>} that lets you typeset a caption without a floating environment. You have the full and absolute control about the placement of your figures and captions.


It is always good practice to add a caption to any figure or table. Fortunately, this is very simple in LaTeX. All you need to do is use the \caption{''text''} command within the float environment. LaTeX will automatically keep track of the numbering of figures, so you do not need to include this within the caption text.

The location of the caption is traditionally underneath the float. However, it is up to you to, therefore, insert the caption command after the actual contents of the float (but still within the environment). If you place it before, then the caption will appear above the float. Try out the following example to demonstrate this effect:




  \caption{A picture of a gull.}

  \caption{A picture of the same gull
           looking the other way!}

    \begin{tabular}{| l c r |}
    1 & 2 & 3 \\
    4 & 5 & 6 \\
    7 & 8 & 9 \\
  \caption{A simple table}

Notice how the tables and figures
have independent counters.


Note that the command \reflectbox{...} flips its content horizontally.

Side captions

It is sometimes desirable to have a caption appear on the side of a float, rather than above or below. The sidecap package can be used to place a caption beside a figure or table. The following example demonstrates this for a figure by using a SCfigure environment in place of the figure environment. The floatrow package is newer and has more capabilities.




  \caption{ ... caption text ... }
    {Giraffe_picture}% picture filename



Unnumbered captions

In some types of documents (such as presentations), it may not be desirable for figure captions to start with Figure:. This is easy to suppress by just placing the caption text in the figure environment, without enclosing it in a caption. This however means that there is no caption available for inclusion in a list of figures.

Caption Setup Rename

In case you want to rename your table caption from "Table" to something else, you can use the \captionsetup command. For example,

\captionsetup[table]{name=New Table Name}

\usepackage{caption} \captionsetup[table] {name=New Table Name}

Lists of figures and tables

Captions can be listed at the beginning of a paper or report in a "List of Tables" or a "List of Figures" section by using the \listoftables or \listoffigures commands, respectively. The caption used for each figure will appear in these lists, along with the figure numbers, and page numbers that they appear on.

The \caption command also has an optional parameter, \caption[''short'']{''long''} which is used for the List of Tables or List of Figures. Typically the short description is for the caption listing, and the long description will be placed beside the figure or table. This is particularly useful if the caption is long, and only a "one-liner" is desired in the figure/table listing. Here is an example of this usage:

\newcommand{\species}[1]{\textit{#1} sp.}


	\caption[Close up of \species{Hemidactylus}]
	{Close up of \species{Hemidactylus}, which is part the genus of the gecko family. It is the second most speciose genus in the family.}



Labels and cross-referencing

Labels and cross-references work fairly similarly to the general case - see the Labels and Cross-referencing section for more information.

If the label picks up the section or list number instead of the figure number, put the label inside the caption to ensure correct numbering. If you get an error when the label is inside the caption, use \protect in front of the \label command.

Wrapping text around figures

An author may prefer that some floats do not break the flow of text, but instead allow text to wrap around it. (Obviously, this effect only looks decent when the figure in question is significantly narrower than the text width.)

A word of warning: Wrapping figures in LaTeX will require a lot of manual adjustment of your document. There are several packages available for the task, but none of them work perfectly. Before you make the choice of including figures with text wrapping in your document, make sure you have considered all the options. For example, you could use a layout with two columns for your documents and have no text-wrapping at all.

Anyway, we will look at the package wrapfig. Note that wrapfig may not come with the default installation of LaTeX; you might need to install additional packages. Noted also, wrapfig is incompatible with the enumerate and itemize environments

To use wrapfig, you must first add this to the preamble:


This then gives you access to:


The lineheight is expressed as the number of lines of text the figure spans. LaTeX will automatically calculate the value if this option is left blank but this can result in figures that look ugly (with too much spacing). The LaTeX calculation is manually overridden by entering the number of lines you would like the figure to span. This option can't be entered in pt, mm, etc...

There are overall eight possible positioning targets:

r R right side of the text
l L left side of the text
i I inside edge–near the binding (in a twoside document)
o O outside edge–far from the binding

The uppercase-character allows the figure to float, while the lowercase version means "exactly here". [4]

The overhang of the figure can be manually set using the overhang option in pt, cm, etc...

The width is, of course, the width of the figure. An example:

  \caption{A gull}

You can also allow LaTeX to assign a width to the wrap by setting the width to 0pt: \begin{wrapfigure}{l}{0pt}

Note that we have specified a size for both the wrapfigure environment and the image we have included. We did it in terms of the text width: it is always better to use relative sizes in LaTeX, let LaTeX do the work for you! The "wrap" is slightly bigger than the picture, so the compiler will not return any strange warning and you will have a small white frame between the image and the surrounding text. You can change it to get a better result, but if you don't keep the image smaller than the "wrap", you will see the image over the text.

The wrapfig package can also be used with user-defined floats with float package. See below in the section on custom floats.

Tip for figures with too much white space

You can use intextsep parameter to control additional space above and below the figure: \setlength\intextsep{0pt}

To control the horizontal space between the image and the text, use the columnsep parameter.

It happens that you'll generate figures with too much (or too little) white space on the top or bottom. In such a case, you can simply make use of the optional argument [lineheight]. It specifies the height of the figure in the number of lines of text. Also remember that the environment center adds some extra white space at its top and bottom; consider using the command \centering instead.

Another possibility is adding space within the float using the \vspace{...} command. The argument is the size of the space you want to add, you can use any unit you want, including pt, mm, in, etc. If you provide a negative argument, it will add a negative space, thus removing some white space. Using \vspace tends to move the caption relative to the float while the [lineheight] argument does not. Here is an example using the \vspace command, the code is exactly the one of the previous cases, we just added some negative vertical spaces to shrink everything up:

  \caption{A gull}

In this case, it may look too shrunk, but you can manage spaces the way you like. In general, it is best not to add any space at all: let LaTeX do the formatting work!

(In this case, the problem is the use of \begin{center} to center the image. The center environment adds extra space that can be avoided if \centering is used instead.)

Alternatively, you might use the picins package instead of the wrapfig package which produces a correct version without the excess white space out of the box without any hand-tuning.

There is also an alternative to wrapfig: the package floatflt [13].

To remove the white space from a figure once for all, one should refer to the program pdfcrop, included in most TeX installations.


A useful extension is the subcaption[14] package, which uses subfloats within a single float. The subfig package (subfigure package is deprecated[5]) is a useful alternative when used in conjunction with LaTeX templates (i.e. templates for journals from Springer and IOP, IEEETran and ACM SIG) that are not compatible with subcaption. These packages give the author the ability to have subfigures within figures, or subtables within table floats. Subfloats have their own caption, and an optional global caption. An example will best illustrate the usage of the subcaption package:


        \caption{A gull}
    ~ %add desired spacing between images, e. g. ~, \quad, \qquad, \hfill etc. 
      %(or a blank line to force the subfigure onto a new line)
        \caption{A tiger}
    ~ %add desired spacing between images, e. g. ~, \quad, \qquad, \hfill etc. 
    %(or a blank line to force the subfigure onto a new line)
        \caption{A mouse}
    \caption{Pictures of animals}\label{fig:animals}

You will notice that the figure environment is set up as usual. You may also use a table environment for subtables. For each subfloat, you need to use:

\begin{table}[<placement specifier>]
    \begin{subtable}[<placement specifier>]{<width>}
        ... table 1 ...
    \caption{<sub caption>}
    \begin{subtable}[<placement specifier>]{<width>}
        ... table 2 ...
        \caption{<sub caption>}

If you intend to cross-reference any of the subfloats, see where the label is inserted; \caption outside the subfigure-environment will provide the global caption.

subcaption will arrange the figures or tables side-by-side providing they can fit, otherwise, it will automatically shift subfloats below. This effect can be added manually, by putting the newline command (\\) before the figure you wish to move to a newline.

Horizontal spaces between figures are controlled by one of several commands, which are placed in between \begin{subfigure} and \end{subfigure}:

  • A non-breaking space (specified by ~ as in the example above) can be used to insert a space in between the subfigs.
  • Math spaces: \qquad, \quad, \;, and \,
  • Generic space: \hspace{''length''}
  • Automatically expanding/contracting space: \hfill

Figures in multiple parts

Sometimes you need to divide up a figure over multiple floats, for instance because the figure is too big to fit on one page. In this case you can use continued figures using the caption package.

Put this in your preamble:


Then you can use continued floats as follows.

	\[ e^{i \pi} + 1 = 0 \]
	\caption{Euler's identity, first form.}
	\[ e^{i \pi} = -1 \]
	\caption{Euler's identity, second form.}

Wide figures in two-column documents

If you are writing a document using two columns (i.e. you started your document with something like \documentclass[twocolumn]{article}), you might have noticed that you can't use floating elements that are wider than the width of a column (using a LaTeX notation, wider than 0.5\textwidth), otherwise you will see the image overlapping with text. If you really have to use such wide elements, the only solution is to use the "starred" variants of the floating environments, that are {figure*} and {table*}. Those "starred" versions work like the standard ones, but they will be as wide as the page, so you will get no overlapping.

A bad point of those environments is that they can be placed only at the top of the page or on their own page. If you try to specify their position using modifiers like b or h, they will be ignored. Add \usepackage{dblfloatfix} to the preamble in order to alleviate this problem with regard to placing these floats at the bottom of a page, using the optional specifier [b]. Default is [tbp]. However, h still does not work.

Custom floats

If tables and figures are not adequate for your needs, then you always have the option to create your own! Examples of such instances could be source code examples, or maps. For a program float example, one might therefore wish to create a float named program. The package float is your friend for this task. All commands to set up the new float must be placed in the preamble, and not within the document.

  1. Add \usepackage{float} to the preamble of your document
  2. Declare your new float using: \newfloat{type}{placement}{ext}[outer counter], where:
    • type - the new name you wish to call your float, in this instance, 'program'.
    • placement - t, b, p, or h (as previously described in Placement), where letters enumerate permitted placements.
    • ext - the file name extension of an auxiliary file for the list of figures (or whatever). LaTeX writes the captions to this file.
    • outer counter - the presence of this parameter indicates that the counter associated with this new float should depend on outer counter, for example 'chapter'.
  3. The default name that appears at the start of the caption is the type. If you wish to alter this, use \floatname{type}{floatname}
  4. Changing float style can be issued with \floatstyle{style} (Works on all subsequent \newfloat commands, therefore, must be inserted before \newfloat to be effective).
    • plain - the normal style for LaTeX floats, but the caption is always below the content.
    • plaintop - the normal style for LaTeX floats, but the caption is always above the content.
    • boxed - a box is drawn that surrounds the float, and the caption is printed below.
    • ruled - the caption appears above the float, with rules immediately above and below. Then the float contents, followed by a final horizontal rule.

Float styles can also be customized as the second example below illustrates.

An example document using a new program float type:






class HelloWorldApp {
  public static void main(String[] args) {
    //Display the string
    System.out.println("Hello World!");
  \caption{The Hello World! program in Java.}


The verbatim is an environment already discussed in Special paragraphs. It is good for source code, but if you want to introduce a lot of code you might consider using the listings package described in Source Code Listings, that was made just for it.

While this is useful, one should be careful when embedding the float within another float. In particular, the error

not in outer par mode

may occur. One solution might be to use the [H] option (not any other) on the inner float, as this option "pins" the inner float to the outer one.

Newly created floats with \newfloat can also be used in combination with the wrapfig package from above. E.g. the following code creates a floating text box, which floats in the text on the right side of the page and is complete with caption, numbering, an index file with the extension .lob and a customization of the float's visual layout:


% have hyperref package before float in order to get strange errors with .\theHfloatbox


% allows use of "@" in control sequence names

% this creates a custom and simpler ruled box style
\newcommand\floatc@simplerule[2]{{\@fs@cfont #1 #2}\par}
  \def\@fs@pre{\hrule height.8pt depth0pt \kern4pt}%
  \def\@fs@post{\kern4pt\hrule height.8pt depth0pt \kern4pt \relax}%

% this code block defines the new and custom floatbox float environment
\floatname{floatbox}{Text Box}


  \textit{Bootstrapping} is a resampling technique used 
  for robustly estimating statistical quantities, such as 
  the model fit $R^2$. It offers some protection against 
  the sampling bias.


Caption styles

To change the appearance of captions, use the caption [15] package. For example, to make all caption labels small and bold:


The KOMA script packages [16] have their own caption customizing features with e.g. \captionabove, \captionformat and \setcapwidth. However these definitions have limited effect on newly created float environments with the wrapfig package.

Alternatively, you can redefine the \thefigure command:


See this page for more information on counters. Finally, note that the caption2 package has long been deprecated.

Labels in the figures

There is a LaTeX package lpic [17] to put LaTeX on top of included graphics, thus allowing to add TeX annotations to imported graphics. It defines a convenient interface to put TeX over included graphics, and allows for drawing a white background under the typeset material to overshadow the graphics. It is a better alternative for labels inside of graphics; you do not have to change text size when rescaling pictures, and all LaTeX power is available for labels.

A very similar package, with somewhat different syntax, is pinlabel [18]. The link given also points to the packages psfrag and overpic.

A much more complicated package which can be used in the same way is TikZ. TikZ is a front-end to a drawing library called pgf (which is used to make beamer for instance). It can be used to label figures by adding text nodes on top of an image node.


That concludes all the fundamentals of floats. You will hopefully see how much easier it is to let LaTeX do all the hard work and tweak the page layouts in order to get your figures in the best place. As always, the fact that LaTeX takes care of all caption and reference numbering is a great time saver.

Notes and references

This page uses material from Andy Roberts' Getting to grips with LaTeX with permission from the author.

Footnotes and Margin Notes


Footnotes are a very useful way of providing extra information to the reader. Usually, it is non-essential information which can be placed at the bottom of the page. This keeps the main body of text concise.

The footnote facility is easy to use. The command you need is: \footnote{text}. Do not leave a space between the command and the word where you wish the footnote marker to appear, otherwise LaTeX will process that space and will leave the output not looking as intended.

Creating a footnote is easy.\footnote{An example footnote.}


LaTeX will obviously take care of typesetting the footnote at the bottom of the page. Each footnote is numbered sequentially - a process that, as you should have guessed by now, is automatically done for you.

If you want your footnote to be at the bottom of the page (instead of the default position of `glued` under the text), consider using:


You can also choose to place the footnote text manually. In this case we use the \footnotemark-\footnotetext duo:

% ...
Somewhere else\footnotetext{This is my footnote!}

The footnote number can also be explicitly specified.

% ...
Somewhere else\footnotetext[17]{This is my footnote!}


  • Enumeration/counter style

It is possible to customize the footnote marking. By default, they are numbered sequentially (Arabic). However, without going too much into the mechanics of LaTeX at this point, it is possible to change this using the following command (which needs to be placed at the beginning of the document, or at least before the first footnote command is issued).

\renewcommand{\thefootnote}{\arabic{footnote}} Arabic numerals, e.g., 1, 2, 3...
\renewcommand{\thefootnote}{\roman{footnote}} Roman numerals (lowercase), e.g., i, ii, iii...
\renewcommand{\thefootnote}{\Roman{footnote}} Roman numerals (uppercase), e.g., I, II, III...
\renewcommand{\thefootnote}{\alph{footnote}} Alphabetic (lowercase), e.g., a, b, c...
\renewcommand{\thefootnote}{\Alph{footnote}} Alphabetic (uppercase), e.g., A, B, C...
\renewcommand{\thefootnote}{\fnsymbol{footnote}} A sequence of nine symbols, try it and see!

To make a footnote without number mark use this declaration:

\let\thefootnote\relax\footnote{There is no number in this footnote}

In this way, the numbering is switched off globally. To have only one footnote without number mark, the above command has to be placed inside { }. Nevertheless, in that case, the current footnote counter is still incremented, so for instance you'd get footnote 1, unnumbered, and 3. A better solution[1] consists in defining the following macro in the preamble, and to use it:


The package footmisc offers many possibilities for customizing the appearance of footnotes. It can be used, for example, to use a different font within footnotes.

Reset counter

  • every section
  • every page

(This may require running LaTeX twice)

\usepackage{perpage} %the perpage package
\MakePerPage{footnote} %the perpage package command

Referencing a footnote

Can be done by placing a \label{labelName} in the end of the footnote and use \ref{labelName} to refer to the footnote.

Common problems and workarounds

  • Footnotes don't work with tables, as it is considered a bad practice. You can overcome this limitation with several techniques: you can use \footnotemark[123] in the table, and \footnotetext[123]{HelloWorld!} somewhere on the page. The same with references: use \footnote{HelloWorld!\label{fnote}} somewhere on the page and \textsuperscript{\ref{fnote}} in the table. Or, you can add \usepackage{footnotehyper} and \makesavenoteenv{tabular} to the preamble, and put your table environment in a \begin{savenotes} environment. The latter does not work with the packages color or colortbl. See this FAQ page for other approaches (such as the use of tablenotes with threeparttable).
  • Footnotes also don't work inside minipage environment. (In fact, several environments break footnote support. The \makesavenoteenv{environmentname} command of the footnote package might fix most.) The minipage includes its own footnotes, independent of the document's. The package mpfnmark allows greater flexibility in managing these two sets of footnotes.
  • If the text within the footnote is a URL (using \url or \href commands) with special characters, it will not compile. You must either escape the characters with a leading backslash, or use another command.
  • If the text within the footnote is very long, LaTeX may split the footnote over several pages. You can prevent LaTeX from doing so by increasing the penalty for such an operation. To do this, insert the following line into the preamble of your document:

or setting \samepage after footnote command:

\footnote{ \samepage text of the footnote follows }

  • To make multiple references to the same footnote, you can use the following syntax:
Text that has a footnote\footnote{This is the footnote} looks like this. Later text referring to the same footnote\footnotemark[\value{footnote}] uses the other command.

If you need hyperref support, use instead:

Text that has a footnote\footnote{This is the footnote}\addtocounter{footnote}{-1}\addtocounter{Hfootnote}{-1} looks like this. Later text referring to the same footnote\footnotemark uses the other command.

These approaches will not work if there are other footnotes between the first reference and the subsequent "duplicate" references. For more general solutions, see here and here.

  • If the footnote is intended to be added to the title of a chapter, a section, or similar, two methods can be used:
  1. Write \section[title] {title\footnote{I'm a footnote referred to the section} } where title is the title of the section.
  2. Use the footmisc package, with package option stable, and simply add the footnote to the section title.

Margin Notes

A margin note.

Margin Notes are useful during the editorial process, to exchange comments among authors. To insert a margin note use \marginpar{margin text}. For one-sided layout (simplex), the text will be placed in the right margin, starting from the line where it is defined. For two-sided layout (duplex), it will be placed in the outside margin and for two-column layout it will be placed in the nearest margin.

To swap the default side, use \reversemarginpar and margin notes will then be placed on the opposite side, which would be the inside margin for two-sided layout.

If the text of your marginpar depends on which margin it is put in (say it includes an arrow pointing at the text or refers to a direction as in "as seen to the left..."), you can use \marginpar[left text]{right text} to specify the variants.

To insert a margin note in an area that \marginpar can't handle, such as footnotes or equation environments, use the package marginnote.

Another option for adding colored margin notes in a fancy way provides the package todonotes by using \todo{todo note}. It makes use of the package pgf used for designing and drawing with a huge tool database.

The packages mparhack and marginnote can be used if the native \marginpar command does not meet your needs.

Margin geometry (bottom margin H not shown).

The marginnote and geometry package can set the widths of the margins and marginnotes as follows.

In the preamble, insert


and use the geometry package with custom sizes:

\usepackage[top=Bcm, bottom=Hcm, outer=Ccm, inner=Acm, heightrounded, marginparwidth=Ecm, marginparsep=Dcm]{geometry}

where A, B, C, D, E, F, G, H are all numbers in cm (of course other units than cm can be used).

In the main text, employ the marginnote package according to:

\marginnote{typeset text here...}[Fcm]


  • marginparwidth (E) is the width of the margin note,
  • marginparsep (D) is the separation between the paragraph and the margin note,
  • F is the downwards vertical offset from the first line the margin note was written (negative values of F shift the margin note upwards), and
  • the value G = C − (D + E) is the separation between the edge of the margin note and the edge.

The example on the right was typeset by the following:



\usepackage[top=1.5cm, bottom=1.5cm, outer=5cm, inner=2cm, heightrounded, marginparwidth=2.5cm, marginparsep=2cm]{geometry}



\section{Margin notes}

\marginnote{This is a margin note using the geometry package, set at 0cm vertical offset to the first line it is typeset.}[0cm]
\marginnote{This is a margin note using the geometry package, set at 5cm vertical offset to the first line it is typeset.}[5cm]

Additionally, the minimum vertical gap between margin notes can be adjusted with \marginparpush, such as with \setlength{\marginparpush}{0pt}.

Notes and References

  1. "LaTeX footnotes". Retrieved 2016-01-14.

This page uses material from Andy Roberts' Getting to grips with LaTeX with permission from the author.


LaTeX enables typesetting of hyperlinks, useful when the resulting format is PDF, and the hyperlinks can be followed. It does so using the package hyperref.


The package hyperref[1] provides LaTeX the ability to create hyperlinks within the document. It works with pdflatex and also with standard "latex" used with dvips and ghostscript or dvipdfm to build a PDF file. If you load it, you will have the possibility to include interactive external links and all your internal references will be turned to hyperlinks. The compiler pdflatex makes it possible to create PDF files directly from the LaTeX source, and PDF supports more features than DVI. In particular PDF supports hyperlinks. Moreover, PDF can contain other information about a document such as the title, the author, etc., which can be edited using this same package.


The basic usage with the standard settings is straightforward. Just load the package in the preamble:


This will automatically turn all your internal references into hyperlinks. It won't affect the way to write your documents: just keep on using the standard \label-\ref system (discussed in the chapter on Labels and Cross-referencing); with hyperref those "connections" will become links and you will be able to click on them to be redirected to the right page. Moreover the table of contents, list of figures/tables and index will be made of hyperlinks, too. The hyperlinks will not show up if you are working in draft mode.


The package provides some useful commands for inserting links pointing outside the document.



\hyperref[label_name]{''link text''}

This will have the same effect as \ref{label_name} but will make the text link text a full link, instead. The two can be combined. If the lemma labelled as mainlemma was number 4.1.1 the following example would result in

We use \hyperref[mainlemma]{lemma \ref*{mainlemma} }.

We use lemma 4.1.1.

with the hyperlink as expected. Note the "*" after \ref for avoiding nested hyperlinks.




It will show the URL using a mono-spaced font and, if you click on it, your browser will be opened pointing at it.




It will show the string description using standard document font but, if you click on it, your browser will be opened pointing at my_url. Here is an example:

\href{}{Wikibooks home}

Both point at the same page, but in the first case the URL will be shown, while in the second case the URL will be hidden. Note that, if you print your document, the link stored using \href will not be shown anywhere in the document.

Other possibilities

Apart from linking to websites discussed above, hyperref can be used to provide mailto links, links to local files, and links to anywhere within the PDF output file.

E-mail address

A possible way to insert email links is by


It just shows your email address (so people can know it even if the document is printed on paper) but, if the reader clicks on it, (s)he can easily send you an email. Or, to incorporate the url package's formatting and line breaking abilities into the displayed text, use[2]

\href{}{\nolinkurl{} }

When using this form, note that the \nolinkurl command is fragile and if the hyperlink is inside of a moving argument, it must be preceded by a \protect command.

Local file

Files can also be linked using the url or the href commands. You simply have to add the string run: at the beginning of the link string:

\href{run:/path/to/my/file.ext}{text displayed}

Following the version with url does not always work, but href does.

It is possible to use relative paths to link documents near the location of your current document; in order to do so, use the standard Unix-like notation (./ is the current directory, ../ is the previous directory, etc.)


\href{run:D:/my_folder/my_subfolder/my_file.txt}{Windows, absolute path}
\href{run:my_subfolder/my_file.txt}{Windows, relative path}
\href{run:./my_subfolder/my_file.txt}{Windows, relative path, variant}
\href{run:/my_folder/my_subfolder/my_file.txt}{Linux, absolute path}
\href{run:./my_subfolder/my_file.txt}{Linux, relative path}

Hyperlink and Hypertarget

It is also possible to create an anchor anywhere in the document (with or without caption) and to link to it. To create an anchor, use:

\hypertarget{label}{target caption}

and to link to it, use:

\hyperlink{label}{link caption}

where the target caption and link caption are the text that is displayed at the target location and link location respectively.

Note also that if you put a hypertarget, when clicking a link to that hypertarget, it may actually direct to the line after the hypertarget, which is not desirable. Therefore if this occurs, you can report the bug and refer to here for a solution.

Viewing in a browser

You can also get an external URL to the hypertarget by appending #label to the URL for the file, or right clicking one of the hyperlinks to the target and copying the URL, or getting the link from the headings in the sidebar of the PDF, or through a process of deduction from viewing (e.g. subsubsection 11.5.1 would have the label subsubsection.11.5.1). This can be useful e.g. for academic and pedagogical purposes. The URL will then direct to the target if you enable a PDF viewer that is compatible with PDF 1.5 in a browser, such as PDF Viewer for Chrome or Chromium browsers, and No PDF Download for Firefox; and on Android the Xodo app works best with links as it renders them with the correct border as with PDF desktop programs, followed by the Foxit PDF app which renders links highlighted in grey (while in other apps links may also work but without any special rendering, such as the Dropbox PDF viewer app; and yet others do not work with links, e.g Drive PDF viewer, PDF reader and Google PDF viewer). You may need to open the URL to the PDF in a new tab, otherwise it may prompt to download it (i.e. if you are clicking on the URL, don't left click, right click and select to open it in a new tab).


The standard settings should be fine for most users, but if you want to change something, that is also possible. There are several variables and two methods to pass those to the package. Options can be passed as an argument of the package when it is loaded (the standard way packages work), or the \hypersetup command can be used as follows:

\hypersetup{<option1> [, ...]}

you can pass as many options as you want; separate them with a comma. Options have to be in the form:


exactly the same format has to be used if you pass those options to the package while loading it, like this:

\usepackage[<option1, option2>]{hyperref}

Here is a list of the possible variables you can change (for the complete list, see the official documentation). The default values are written in an upright font:

Checkout 3.8 Big list at hyperref-manual at

variable values comment
bookmarks =true,false show or hide the bookmarks bar when displaying the document
unicode =false,true allows to use characters of non-Latin based languages in Acrobat’s bookmarks
pdfborder ={RadiusH RadiusV Width [Dash-Pattern]} set the style of the border around a link. The first two parameters (RadiusH, RadiusV) have no effect in most pdf viewers. Width defines the thickness of the border. Dash-Pattern is a series of numbers separated by space and enclosed by box-brackets. It is an optional parameter to specify the length of each line & gap in the dash pattern. For example, {0 0 0.5 [3 3]} is supposed to draw a square box (no rounded corners) of width 0.5 and a dash pattern with a dash of length 3 followed by a gap of length 3. There is no uniformity in whether/how different pdf viewers render the dash pattern.
pdftoolbar =true,false show or hide Acrobat’s toolbar
pdfmenubar =true,false show or hide Acrobat’s menu
pdffitwindow =true,false resize document window to fit document size
pdfstartview ={FitH},{FitV},etc[3]. fit the width of the page to the window
pdftitle ={text} define the title that gets displayed in the "Document Info" window of Acrobat
pdfauthor ={text} the name of the PDF’s author, it works like the one above
pdfsubject ={text} subject of the document, it works like the one above
pdfcreator ={text} creator of the document, it works like the one above
pdfproducer ={text} producer of the document, it works like the one above
pdfkeywords ={text} list of keywords, separated by commas, example below
pdfnewwindow (=true,false) define if a new PDF window should get opened when a link leads out of the current document. NB: This option is ignored if the link leads to an http/https address.
pagebackref (=false,true) activate back references inside bibliography. Must be specified as part of the \usepackage{} statement.
colorlinks (=false,true) surround the links by color frames (false) or colors the text of the links (true). The color of these links can be configured using the following options (default colors are shown):
hidelinks hide links (removing color and border)
linkcolor =red color of internal links (sections, pages, etc.)
linktoc =none,section,page,all defines which part of an entry in the table of contents is made into a link
citecolor =green color of citation links (bibliography)
filecolor =cyan color of file links
urlcolor =magenta color of URL links (mail, web)
linkbordercolor ={1 0 0} color of frame around internal links (if colorlinks=false)
citebordercolor ={0 1 0} color of frame around citations
urlbordercolor ={0 1 1} color of frame around URL links

Please note, that explicit RGB specification is only allowed for the border colors (like linkbordercolor etc.), while the others may only assigned to named colors (which you can define your own, see Colors). In order to speed up your customization process, here is a list with the variables with their default value. Copy it in your document and make the changes you want. Next to the variables, there is a short explanations of their meaning:

    bookmarks=true,         % show bookmarks bar?
    unicode=false,          % non-Latin characters in Acrobat’s bookmarks
    pdftoolbar=true,        % show Acrobat’s toolbar?
    pdfmenubar=true,        % show Acrobat’s menu?
    pdffitwindow=false,     % window fit to page when opened
    pdfstartview={FitH},    % fits the width of the page to the window
    pdftitle={My title},    % title
    pdfauthor={Author},     % author
    pdfsubject={Subject},   % subject of the document
    pdfcreator={Creator},   % creator of the document
    pdfproducer={Producer}, % producer of the document
    pdfkeywords={keyword1, key2, key3}, % list of keywords
    pdfnewwindow=true,      % links in new PDF window
    colorlinks=false,       % false: boxed links; true: colored links
    linkcolor=red,          % color of internal links (change box color with linkbordercolor)
    citecolor=green,        % color of links to bibliography
    filecolor=cyan,         % color of file links
    urlcolor=magenta        % color of external links

If you don't need such a high customization, here are some smaller but useful examples. When creating PDFs destined for printing, colored links are not a good thing as they end up in gray in the final output, making it difficult to read. You can use color frames, which are not printed:


or make links black:


or use \usepackage{hyperref} \hypersetup{hidelinks}

When you just want to provide information for the Document Info section of the PDF file, as well as enabling back references inside bibliography:

\usepackage[pdfauthor={Author's name},%
pdftitle={Document Title},%

By default, URLs are printed using mono-spaced fonts. If you don't like it and you want them to be printed with the same style of the rest of the text, you can use this:



Problems with Links and Equations 1

Messages like the following

! pdfTeX warning (ext4): destination with the same identifier (name{
equation.}) has been already used, duplicate ignored

appear, when you have done something like


The error disappears, if you use instead this form:


Beware that the shown line number is often completely different from the erroneous line.

Possible solution: Place the amsmath package before the hyperref package.

Problems with Links and Equations 2

Messages like the following

! Runaway argument?
{\@firstoffive }\fi ), Some text from your document here (\ref {re\ETC.
Latex Error: Paragraph ended before \Hy@setref@link was complete.

appear when you use \label inside an align environment.

Possible solution: Add the following to your preamble:


Note: The same error appears if you use a colon ":" as part of a label, i.e. \label{lst:program01}. Replacing that will help.

Problems with Links and Pages

Messages like the following:

! pdfTeX warning (ext4): destination with the same
identifier (name{page.1}) has been already used,
duplicate ignored

appear when a counter gets reinitialized, for example by using the command \mainmatter provided by the book document class. It resets the page number counter to 1 prior to the first chapter of the book. But as the preface of the book also has a page number 1 all links to "page 1" would not be unique anymore, hence the notice that "duplicate has been ignored." The counter measure consists of putting plainpages=false into the hyperref options. This unfortunately only helps with the page counter. An even more radical solution is to use the option hypertexnames=false, but this will cause the page links in the index to stop working.

The best solution is to give each page a unique name by using the \pagenumbering command:

\pagenumbering{alph}    % a, b, c, ...
... titlepage, other front matter ...
\pagenumbering{roman}   % i, ii, iii, iv, ...
... table of contents, table of figures, ...
\pagenumbering{arabic}  % 1, 2, 3, 4, ...
... beginning of the main matter (chapter 1) ...

Another solution is to use \pagenumbering{alph} before the command \maketitle, which will give the title page the label page.a. Since the page number is suppressed, it won't make a difference to the output.

By changing the page numbering every time before the counter is reset, each page gets a unique name. In this case, the pages would be numbered a, b, c, i, ii, iii, iv, v, 1, 2, 3, 4, 5, ...

If you don't want the page numbers to be visible (for example, during the front matter part), use \pagestyle{empty} ... \pagestyle{plain}. The important point is that although the numbers are not visible, each page will have a unique name.

Another more flexible approach is to set the counter to something negative:

... titlepage, other front matter ...
\pagenumbering{roman}   % i, ii, iii, iv, ...
... table of contents, table of figures, ...
\pagenumbering{arabic}  % 1, 2, 3, 4, ...
... beginning of the main matter (chapter 1) ...

which will give the first pages a unique negative number.

The problem can also occur with the algorithms package: because each algorithm uses the same line-numbering scheme, the line identifiers for the second and follow-on algorithms will be duplicates of the first.

The problem occurs with equation identifiers if you use \nonumber on every line of an eqnarray environment. In this case, use the *'ed form instead, e.g. \begin{eqnarray*} ... \end{eqnarray*} (which is an unnumbered equation array), and remove the now unnecessary \nonumber commands.

If your url's are too long and running off of the page, try using the breakurl package to split the url over multiple lines. This is especially important in a multicolumn environment where the line width is greatly shortened.

Problems with bookmarks

The text displayed by bookmarks does not always look like you expect it to look. Because bookmarks are "just text", much fewer characters are available for bookmarks than for normal LaTeX text. Hyperref will normally notice such problems and put up a warning:

Package hyperref Warning:
Token not allowed in a PDFDocEncoded string:

You can now work around this problem by providing a text string for the bookmarks, which replaces the offending text:

\texorpdfstring{''TEX text''}{''Bookmark Text''}

Math expressions are a prime candidate for this kind of problem:

\section{ \texorpdfstring{$E=mc^2$}{E=mc2} }

which turns \section{$E=mc^2$} to E=mc2 in the bookmark area. Color changes also do not travel well into bookmarks:

\section{ \textcolor{red}{Red !} }

produces the string "redRed!". The command \textcolor gets ignored but its argument (red) gets printed. If you use:

\section{ \texorpdfstring{\textcolor{red}{Red !}}{Red\ !} }

the result will be much more legible.

If you write your document in unicode and use the unicode option for the hyperref package you can use unicode characters in bookmarks. This will give you a much larger selection of characters to pick from when using \texorpdfstring.

Problems with tables and figures

The links created by hyperref point to the label created within the float environment, which, as previously described, must always be set after the caption. Since the caption is usually below a figure or table, the figure or table itself will not be visible upon clicking the link[4]. A workaround exists by using the package hypcap [19] with:


Be sure to call this package after loading hyperref.

If you use the wrapfig package[5] mentioned in the "Wrapping text around figures" section of the "Floats, Figures and Captions" chapter, or other similar packages that define their own environments, you will need to manually include \capstart in those environments, e.g.:

  \caption{\label{labelname}a figure}

Problems with long caption and \listoffigures or long title

There is an issue when using \listoffigures with hyperref for long captions or long titles. This happens when the captions (or the titles) are longer than the page width (about 7-9 words depending on your settings). To fix this, you need to use the option breaklinks when first declaring:


This will then cause the links in the \listoffigures to word wrap properly.

Problems with already existing .toc, .lof and similar files

The format of some of the auxiliary files generated by latex changes when you include the hyperref package. One can therefore encounter errors like

! Argument of \Hy@setref@link has an extra }.

when the document is typeset with hyperref for the first time and these files already exist. The solution to the problem is to delete all the files that latex uses to get references right and typeset again.

Problems with footnotes and special characters

See the relevant section.

Problems with Beamer

Using the command

\hyperref[some_label]{some text}

is broken when pointed at a label. Instead of sending the user to the desired label, upon clicking the user will be sent to the first frame. A simple work around exists; instead of using


to label your frames, use


and reference it with

\hyperlink{some_label}{some text}

Problems with draft mode

WARNING! Please note that if you have activated the "draft"-option in your \documentclass declaration the hyperlinks will not show up in the table of contents, or anywhere else for that matter!!!

The hyperlinks can be re-enabled by using the "final=true" option in the following initialization of the hyperref package, just after the package was included:


A good source of further options for the hyperref package can be found here [6].

Notes and References

  1. Hyperref package webpage in CTAN
  2. "Email link with hyperref, url packages". comp.text.tex User Group. {{cite web}}: Unknown parameter |accessmonthday= ignored (help); Unknown parameter |accessyear= ignored (|access-date= suggested) (help)
  3. Other possible values are defined in the hyperref manual
  5. Wrapfig package webpage in CTAN
  6. [1]Hyperref - Hyperlinks With LaTeX Page

Labels and Cross-referencing


In LaTeX, you can easily reference almost anything that can be numbered, and have LaTeX automatically updating the numbering for you whenever necessary. The objects which can be referenced include chapters, sections, subsections, footnotes, theorems, equations, figures and tables[1]. The commands to be used do not depend on what you are referencing, and they are:

Used to give the object you want to reference a marker — a name which can be used to refer to that object later.
Used to reference an object with the specified marker. This will print the number that was assigned to the object.
Used to print the page number where the object with the specified marker is found.

LaTeX will calculate the right numbering for the objects in the document; the marker you have used to label the object will not be shown anywhere in the document. Instead, LaTeX will replace the string "\ref{marker}" with the right number that was assigned to the object. If you reference a marker that does not exist, the compilation of the document will be successful but LaTeX will return a warning:

LaTeX Warning: There were undefined references.

and it will replace "\ref{unknown-marker}" with "??" — so that it will be easier to find in the document.

As you may have noticed, this way of cross-referencing is a two-step process: first the compiler has to store the labels with the right number to be used for referencing, then it has to replace the \ref with the right number.

Because of that, you would have to compile your document twice to see the output with the proper numbering. If you only compile it once, then LaTeX will use the older information collected in previous compilations (which might be outdated), and the compiler will inform you by printing the following message at the end of the compilation:

LaTeX Warning: Label(s) may have changed. Rerun to get cross-references right.

Using the command \pageref{} you can help the reader to find the referenced object by providing also the page number where it can be found. You could write something like:

See figure~\ref{fig:test} on page~\pageref{fig:test}.

Since you can use exactly the same commands to reference almost anything, you might get a bit confused after you have introduced a lot of references. It is common practice among LaTeX users to add a few letters to the label to describe what you are referencing. Some packages, such as fancyref, rely on this meta information. Here is an example:

ch: chapter
sec: section
subsec: subsection
fig: figure
tab: table
eq: equation
lst: code listing
itm: enumerated list item
alg: algorithm
app: appendix subsection

Following this convention, the label of a figure will look like \label{fig:my_figure}, etc. You are not obligated to use these prefixes, and can in fact use any string as an argument of \label{...}, but these prefixes can become increasingly useful as your document grows in size.

Another suggestion: try to avoid using numbers within labels. You are better off describing what the object is about. This way, if you change the order of the objects, you will not have to rename all your labels and their references.

If you want to be able to see the markers you are using in the output document as well, you can use the showkeys package; this can become very useful as you develop your document. For more information see the Packages section.


Here are some practical examples, but you will notice that they are all the same because they all use the same commands.





I greeted in section~\ref{sec:greetings}.


You could place the label anywhere in the section; however, in order to avoid confusion, it is better to place it immediately after the beginning of the section. Note how the marker starts with sec:, as suggested before. The label is then referenced in a different section, where the tilde (~) indicates a non-breaking space.


You can reference a picture by inserting it in the figure floating environment.

  \caption{Close-up of a gull}
Figure~\ref{fig:gull} shows a photograph of a gull.


When a label is declared within a float environment, the \ref{...} will return the respective figure/table number, but it must occur after the caption. When declared outside, it will give the section number. To be completely safe, the label for any picture or table can go within the \caption{} command, as follows:

\caption{Close-up of a gull\label{fig:gull}}

For more, see the Floats, Figures and Captions section about the figure and related environments.

Fixing wrong labels

The command \label must appear after (or inside) \caption. Otherwise, it will pick up the current section or the list number instead of what is intended.

  \caption{Close-up of a gull} \label{fig:gull} 

Issues with links to tables and figures handled by hyperref

In case you use the package hyperref to create a PDF, the link to a table or a figure will point to its caption instead, which is always below the table or the figure itself[2]. As a result, the table or the figure will not be visible if it is above the pointer, which means that some scrolling-up would be required. If you want the link to point to the top of the image, you can give the option hypcap to the caption package[3]:

\usepackage{caption} % hypcap is true by default so [hypcap=true] is optional in \usepackage[hypcap=true]{caption}


Here is an example showing how to reference formulae:

\begin{equation} \label{eq:solve}
x^2 - 5 x + 6 = 0

x_1 = \frac{5 + \sqrt{25 - 4 \times 6}}{2} = 3

x_2 = \frac{5 - \sqrt{25 - 4 \times 6}}{2} = 2

and so we have solved equation~\eqref{eq:solve}


Here, notice the eq: prefix in the label — and that the label is placed soon after the beginning of the math mode. To reference a formula, an environment with counter would have to be used. Most of the times, you will be using the equation environment, as that's usually the best choice for one-line formulae whether you are using amsmath or not.


The amsmath package adds a new command for referencing formulae; it is \eqref{}. It works exactly like \ref{}, but adds parentheses so that instead of printing a plain number as 5, it will print (5). This can be useful to help the reader distinguish between formulae and other things, without the need to repeat the word "formula" before any reference.[4] Its output can be changed as desired; for more information see the amsmath documentation.


The \tag{eqnno} command is used to manually set equation numbers, where eqnno is the text string you want to display instead of the usual equation number. It is normally better to use labels, but sometimes hard-coded equation numbers might offer a useful work-around — such as the case where you want to repeat an equation that has already been used before (e.g. \tag{\ref{eqn:before}}).


The amsmath package adds the \numberwithin{countera}{counterb} command, which replaces the simple countera with a more sophisticated counterb.countera. For example, \numberwithin{equation}{section} in the preamble will prepend the section number to all equation numbers.


The cases package adds the \numcases and the \subnumcases commands, which produce multi-case equations with a separate equation number and a separate equation number plus a letter, respectively, for each case.

The varioref package

The varioref package introduces a new command called \vref{}. This command is used exactly like the basic \ref, but it has a different output according to the context. If the object to be referenced is in the same page, it works just like \ref; if the object is far away, it will print something like "5 on page 25", i.e. it adds the page number automatically. If the object is close, it can use more refined sentences such as "on the next page" or "on the facing page" automatically — according to the context and the document class.

This command has to be used very carefully. It outputs more than one word, so it may happen that its output falls on two different pages. In this case, the algorithm can get confused and cause a loop.

For example, you could label an object on page 23 and the \vref output could happen to stay between page 23 and 24. If it were on page 23, it would print like the basic ref, if it were on page 24, it would print "on the previous page", but it is on both, and this may cause some strange errors at compiling time that could be very difficult to fix.

And while for small documents, these situations might happen very rarely, for long documents spanning hundreds of references, these situations are more likely to happen. One way to avoid these problems during document preparation is to use the standard ref all the way through at first, convert all to vref when the document is close to its final version — before making the adjustment to fix any possible problem.

The hyperref package


The hyperref package introduces another useful command; \autoref{}. This command creates a reference with additional text corresponding to the target's type, all of which will be a hyperlink. For example, the command \autoref{sec:intro} would create a hyperlink to the \label{sec:intro} command, wherever it is. Assuming that this label is pointing to a section, the hyperlink would contain the text "section 3.4", or similar (the full list of default names can be found here). Note that, while there's an \autoref* command that produces an unlinked prefix (useful if the label is on the same page as the reference), no alternative \Autoref command is defined to produce capitalized versions (useful, for instance, when starting sentences); but since the capitalization of autoref names was chosen by the package author, you can customize the prefixed text by redefining \typeautorefname to the prefix you want, as in:


This renaming trick can, of course, be used for other purposes as well.

  • If you would like to have a hyperlink reference without the predefined text \autoref{} provides, then you change this with a command such as \hyperref[sec:intro]{Appendix~\ref*{sec:intro}}. Note that you can disable the creation of hyperlinks in hyperref, and just use these commands for automatic text.
  • Keep in mind that the \label must be placed inside an environment with a counter, such as a table or a figure. Otherwise, not only will the number refer to the current section (as mentioned above), but the name might refer to the previous environment with a counter as well. For example, if you put a label after closing a figure, the label will still say "figure n", on which n is the current section number.


The hyperref package also automatically includes the nameref package, and a similarly named command. It is similar to \autoref{}, but inserts text corresponding to the section name, for example.


\section{MyFirstSection} \label{sec:marker}
In section~\nameref{sec:marker} we defined...


In section MyFirstSection we defined...

Anchor manual positioning

When you define a \label outside a figure, a table, or other floating objects, the label points to the current section. In some cases, this behavior is not what you'd like and you'd prefer the generated link to point to the line where the \label is defined. This can be achieved with the command \phantomsection as in this example:

%The link location will be placed on the line below.

The cleveref package

The cleveref package introduces the new command \cref{} which includes the type of referenced object like \autoref{} does. The alternate \labelcref{} command works more like standard \ref{}. References to pages are handled by the \cpageref{} command.

  • Use \crefrange{}{} and \cpagerefrange{} commands for start and end label in either order and to provide a natural language (babel enabled) range.
  • Use \cref{} command for multiple or single references (on single line), it will sort them and group into ranges automatically in very convenient format like ` item 2 to 4 and 6 to 19.`.[5]
  • Use \labelcref{} where the numbers would be a references to the labels and [sorted] output as in previous paragraph

The format can be specified in the preamble.

Interpackage interactions for varioref , hyperref , and cleveref

Because varioref,hyperref, and cleveref redefine the same commands, they can produce unexpected results when their \usepackage commands appear in the preamble in the wrong order. For example, using hyperref,varioref, then cleveref can cause \vref{} to fail as though the marker were undefined.[6] The following order generally seems to work:

  1. varioref
  2. hyperref
  3. cleveref[6]

See also

Notes and References

Errors and Warnings

LaTeX describes what it is typesetting while it does it. If it encounters something it doesn't understand or can't do, it will display a message saying what is wrong. It may also display warnings for less serious conditions.

Don't panic if you see error messages: it is very common to mistype or misspell commands, forget curly braces, type a forward slash instead of a backslash, or use a special character by mistake. Errors are easily spotted and easily corrected in your editor, and you can then run LaTeX again to check you have fixed everything. Some of the most common errors are described in the next sections.

Error messages

The format of an error message is always the same. Error messages begin with an exclamation mark at the start of the line, and give a description of the error, followed by another line starting with the number, which refers to the line-number in your document file which LaTeX was processing when the error was spotted. Here's an example, showing that the user mistyped the \tableofcontents command:

! Undefined control sequence.
l.6 \tableofcotnetns

When LaTeX finds an error like this, it displays the error message and pauses. You must type one of the following letters to continue:

Key Meaning
x Stop immediately and exit the program.
q Carry on quietly as best you can and don't bother me with any more error messages.
e Stop the program but re-position the text in my editor at the point where you found the error (This only works if you're using an editor which LaTeX can communicate with).
h Try to give me more help.
i (followed by a correction) means input the correction in place of the error and carry on (This is only a temporary fix to get the file processed. You still have to make that correction in the editor).
r run in non-stop mode. Plow through any errors, unless too many pile up and it fails (100 errors).

Some systems (Emacs is one example) run LaTeX with a "nonstop" switch turned on, so it will always process through to the end of the file, regardless of errors, or until a limit is reached.


Warnings don't begin with an exclamation mark: they are just comments by LaTeX about things you might want to look into, such as overlong or underrun lines (often caused by unusual hyphenations, for example), pages running short or long, and other typographical niceties (most of which you can ignore until later). Unlike other systems, which try to hide unevennesses in the text (usually unsuccessfully) by interfering with the letter spacing, LaTeX takes the view that the author or editor should be able to contribute. While it is certainly possible to set LaTeX's parameters so that the spacing is sufficiently sloppy that you will almost never get a warning about badly-fitting lines or pages, you will almost certainly just be delaying matters until you start to get complaints from your readers or publishers.


Only a few common error messages are given here: those most likely to be encountered by beginners. If you find another error message not shown here, and it's not clear what you should do, ask for help.

Most error messages are self-explanatory, but be aware that the place where LaTeX spots and reports an error may be later in the file than the place where it actually occurred. For example if you forget to close a curly brace which encloses, say, italics, LaTeX won't report this until something else occurs which can't happen until the curly brace is encountered (e.g. the end of the document!) Some errors can only be righted by humans who can read and understand what the document is supposed to mean or look like.

Newcomers should remember to check the list of special characters: a very large number of errors when you are learning LaTeX are due to accidentally typing a special character when you didn't mean to. This disappears after a few days as you get used to them.

Too many }'s

! Too many }'s.
l.6 \date December 2004}

The reason LaTeX thinks there are too many }'s here is that the opening curly brace is missing after the \date control sequence and before the word December, so the closing curly brace is seen as one too many (which it is!). In fact, there are other things which can follow the \date command apart from a date in curly braces, so LaTeX cannot possibly guess that you've missed out the opening curly brace until it finds a closing one!

Undefined control sequence

! Undefined control sequence.
l.6 \dtae
{December 2004}

In this example, LaTeX is complaining that it has no such command ("control sequence") as \dtae. Obviously it's been mistyped, but only a human can detect that fact: all LaTeX knows is that \dtae is not a command it knows about: it's undefined. Mistypings are the most common source of errors. Some editors allow common commands and environments to be inserted using drop-down menus or icons, which may be used to avoid these errors.

Not in Mathematics Mode

! Missing $ inserted

A character that can only be used in the mathematics mode was inserted in normal text. If you intended to use mathematics mode, then use $...$ or \begin{math}...\end{math} or use the 'quick math mode': \ensuremath{...}. If you did not intend to use mathematics mode, then perhaps you are trying to use a special character that needs to be entered in a different way; for example _ will be interpreted as a subscript operator in mathematics mode, and you need \_ to get an underscore character.

This can also happen if you use the wrong character encoding, for example using utf8 without "\usepackage[utf8]{inputenc}" or using iso8859-1 without "\usepackage[latin1]{inputenc}", there are several character encoding formats, make sure to pick the right one.

Runaway argument

Runaway argument?
{December 2004 \maketitle
! Paragraph ended before \date was complete.
<to be read again>

In this error, the closing curly brace has been omitted from the date. It's the opposite of the error of too many }'s, and it results in \maketitle trying to format the title page while LaTeX is still expecting more text for the date! As \maketitle creates new paragraphs on the title page, this is detected and LaTeX complains that the previous paragraph has ended but \date is not yet finished.

Underfull hbox

Underfull \hbox (badness 1394) in paragraph
at lines 28--30
[][]\LY1/brm/b/n/10 Bull, RJ: \LY1/brm/m/n/10
Ac-count-ing in Busi-

This is a warning that LaTeX cannot stretch the line wide enough to fit, without making the spacing bigger than its currently permitted maximum. The badness (0-10,000) indicates how severe this is (here you can probably ignore a badness of 1394). It says what lines of your file it was typesetting when it found this, and the number in square brackets is the number of the page onto which the offending line was printed. The codes separated by slashes are the typeface and font style and size used in the line. Ignore them for the moment.

This comes up if you force a linebreak, e.g., \\, and have a return before it. Normally TeX ignores linebreaks, providing full paragraphs to ragged text. In this case it is necessary to pull the linebreak up one line to the end of the previous sentence.

This warning may also appear when inserting images. It can be avoided by using the \textwidth or possibly \linewidth options, e.g. \includegraphics[width=\textwidth]{image_name}

Overfull hbox

Overfull \hbox (9.11617pt too wide) in paragraph
at lines 860--861
[]\LY1/brm/m/n/10 Windows, \LY1/brm/m/it/10 see
\LY1/brm/m/n/10 X Win-

An overfull \hbox means that there is a hyphenation or justification problem: moving the last word on the line to the next line would make the spaces in the line wider than the current limit; keeping the word on the line would make the spaces smaller than the current limit, so the word is left on the line, but with the minimum allowed space between words, and which makes the line go over the edge.

The warning is given so that you can find the line in the code that originates the problem (in this case: 860-861) and fix it. The line on this example is too long by a shade over 9pt. The chosen hyphenation point which minimizes the error is shown at the end of the line (Win-). Line numbers and page numbers are given as before. In this case, 9pt is too much to ignore (over 3mm), and a manual correction needs making (such as a change to the hyphenation), or the flexibility settings need changing.

If the "overfull" word includes a forward slash, such as "input/output", this should be properly typeset as "input\slash output". The use of \slash has the same effect as using the "/" character, except that it can form the end of a line (with the following words appearing at the start of the next line). The "/" character is typically used in units, such as "mm/year" character, which should not be broken over multiple lines.

The warning can also be issued when the \end{document} tag was not included or was deleted.

Easily spotting overfull hboxes in the document

To easily find the location of overfull hbox in your document, you can make latex add a black bar where a line is too wide:


Missing package

! LaTeX Error: File `paralisy.sty' not found.
Type X to quit or <RETURN> to proceed,
or enter new name. (Default extension: sty)
Enter file name:

When you use the \usepackage command to request LaTeX to use a certain package, it will look for a file with the specified name and the filetype .sty. In this case the user has mistyped the name of the paralist package, so it's easy to fix. However, if you get the name right, but the package is not installed on your machine, you will need to download and install it before continuing. If you don't want to affect the global installation of the machine, you can simply download from Internet the necessary .sty file and put it in the same folder of the document you are compiling.

Package babel Warning: No hyphenation patterns were loaded for the language X

Although this is a warning from the Babel package and not from LaTeX, this error is very common and (can) give some strange hyphenation (word breaking) problems in your document. Wrong hyphenation rules can decrease the neatness of your document.

Package babel Warning: No hyphenation patterns were loaded for
(babel)                the language `Latin'
(babel)                I will use the patterns loaded for \language=0 instead.

This can happen after the usage of: (see LaTeX/Internationalization)


The solution is not difficult, just install the used language in your LaTeX distribution.

Package babel Error: You haven't loaded the option X yet.

If you previously set the X language, and then decided to switch to Y, you will get this error. This may seem awkward, as there is obviously no error in your code if you did not change anything. The answer lies in the .aux file, where babel defined your language. If you try the compilation a second time, it should work. If not, delete the .aux file, then everything will work as usual.

No error message, but won't compile

One common cause of (pdf)LaTeX getting stuck is forgetting to include \end{document}

Software that can check your .tex Code

There are several programs capable of checking LaTeX source, with the aim of finding errors or highlighting bad practice, and providing more help to (particularly novice) users than the built-in error messages.




Counters are an essential part of LaTeX: they allow you to control the numbering mechanism of everything (sections, lists, captions, etc.). To that end each counter stores an integer value in the range of long integer, i.e., from   to  . [20]

Counter manipulation

In LaTeX it is fairly easy to create new counters and even counters that reset automatically when another counter is increased (think subsection in a section for example). With the command


you create a new counter that is automatically set to zero. If you want the counter to be reset to zero every time another counter is increased, use:


For example, if you want to enumerate the equations in each chapter independently, you can create something like an "equationschapter" counter that will be automatically reset at the begin of each section.

    \section{First Section}
    I present one equation:
    \stepcounter{equationschapter} $a=b+c$ (Eq. \arabic{section}.\arabic{equationschapter})
    \section{Second Section}
    I present more equations:
    \stepcounter{equationschapter} $a=c+d$ (Eq. \arabic{section}.\arabic{equationschapter})
    \stepcounter{equationschapter} $d=e$ (Eq. \arabic{section}.\arabic{equationschapter})

To add to an existing counter another counter causing a reset when increased, use:


If this doesn't work it might be because of an old LaTeX version, the following should work in that case:


To undo this effect one can use:




To increase the counter, either use



\refstepcounter{NameOfTheNewCounter} % used for labels and cross referencing



here the number can also be negative. For automatic resetting you need to use \stepcounter.

To set the counter value explicitly, use


Counter access

There are several ways to get access to a counter.

  • \theNameOfTheNewCounter will print the formatted string related to the counter (note the "the" before the actual name of the counter).
  • \value{NameOfTheNewCounter} will return the counter value which can be used by other counters or for calculations. It is not a formatted string, so it cannot be used in text.
  • \arabic{NameOfTheNewCounter} will print the formatted counter using arabic numbers.

Note that \arabic{NameOfTheNewCounter} may be used as a value too, but not the others.

Strangely enough, LaTeX counters are not introduced by a backslash in any case, even with the \the command. plainTeX equivalents \count and \newcount\mycounter do abide by the backslash rule.

Counter style

The following internal LaTeX commands will convert numeric value of specified counter into printable string and insert string into document:

Numbers from   to   inclusive converted to strings «-2147483648», «-2147483647», …, «-1», «0», «1», …, «2147483646», «2147483647».
Example: 1, 2, 3, …
Numbers from 1 to 26 inclusive converted to strings «a», «b», …, «z». Other numbers (negative numbers, zero, 27, 28, …,  ) converted to empty string.
Example: a, b, c, …
Same as \alph, but upper case letters used.
Example: A, B, C, …
Numbers from 1 to 4999 inclusive converted to strings «i» (1), «ii» (2), …, «mmmmcmxcix» (4999), where «i» — 1, «v» — 5, «x» — 10, «l» — 50, «c» — 100, «d» — 500, «m» — 1000. Numbers from 5000 to   inclusive converted to strings «mmmmm» (5000), «mmmmmi» (5001), …. Other numbers (negative numbers, zero) converted to empty string.
Example: i, ii, iii, …
Same as \roman, but upper case letters used.
Example: I, II, III, …
Aimed at footnotes; prints a sequence of symbols.
Number Symbol(s)
4 §
7 ∗∗
8 ††
9 ‡‡
Other numbers Empty string
Example: ∗, †, ‡, …

LaTeX default counters

  • part
  • chapter
  • section
  • subsection
  • subsubsection
  • paragraph
  • subparagraph
  • page
  • figure
  • table
  • footnote
  • mpfootnote

For the enumerate environment:

  • enumi
  • enumii
  • enumiii
  • enumiv

For the eqnarray environment:

  • equation

Book with parts, sections, but no chapters

Here follows an example where we want to use parts and sections, but no chapters in the book class :

\renewcommand{\thesection}{\thepart .\arabic{section}}

\part{My Part}                                                                
\section{My Section}
\subsection{My Subsection}

Custom enumerate

See the List Structures chapter.

Custom sectioning

Here is an example for recreating something similar to a section and subsection counter that already exist in LaTeX:

\addtocounter{mysection}{2} % set them to some other numbers than 0
\addtocounter{mysubsection}{10} % same
Blah blah

Blah blah

Blah blah

Blah blah and more blah blah


TeX boxes and glue overview

A box is the TeX term for an invisible container that can hold a visible element, nothing, or other boxes. Glue is the TeX term for an invisible connector that determines the relative position of joined boxes. Each separate visible element contained within a TeX document is contained within a box. A visible element can be a letter, image, geometric shape, etc. TeX builds pages by gluing boxes together according to the default TeX rules, default LaTeX rules, or document commands. In a typical document, letter boxes are glued to other letter boxes to form words, which are then elastically glued to other words to form sentences. Sentences are broken into lines and placed in paragraph boxes. Elastic glue is squeezed or stretched to fully justify lines within paragraph boxes. Paragraph boxes are glued to diagram boxes, and so on.

While it is true that boxes can hold other boxes, not all commands that can generate boxes can be used within all other commands that can generate boxes. There are often workarounds for these limitations.

The size of a box is typically related to the size and position of its contents, but it doesn't have to be. Many box commands accept custom widths and/or heights, and there are other commands that affect the shape and position of boxes. Boxes are placed relative to other boxes, while visible elements are placed relative to the boxes which contain them.

A more complete description of boxes and glue can be found in chapters 11 and 12 of Donald E. Knuth's, 'The TeXbook'. A list of basic LaTeX box and glue commands can be found at .


character boxes

TeX character boxes have three dimensional properties:

  • The height is the length between the baseline and the top of the box.
  • The depth is the length between the baseline and the bottom of the box.
  • The width is the width of the box.

Character boxes are glued together at the baseline.

parbox, minipage, and pbox

A \parbox is a box of specific width formatted in paragraph mode. In paragraph mode, text is broken into lines and lines are broken into pages.


width defines the width of the paragraph box. Text will be broken into lines so that it fits within this width. Besides fixed lengths, you can also provide user defined length macros or TeX/LaTeX defined length macros and primitives such as \width, \height, \depth and \totalheight. (See character box above for explanations of depth and height.)

height defines the height of the \parbox.

pos selects which baseline to join. It can be top, bottom, or center. This parameter is often confusing to new users! See the special note below.

contentpos positions the contents of the box within the box. It can be one of center, top, bottom or spread. Note that contentpos has no effect if the box is not larger than the text it contains.

\pbox is available in the pbox package. A \pbox has the same parameters as a \parbox, but if the user provided width parameter is larger than the actual contents of the pbox the pbox shrinks to fit the content. This is only useful if the content contains manual line breaks and you wish to fit the resulting material.


The minipage environment takes the same parameters as a \parbox, and behaves nearly identically to it. The difference between a minipage and a \parbox is that a \parbox can only contain a single paragraph and you cannot use all commands and environments inside it, while a minipage may contain multiple paragraphs and, in fact, almost anything.

\begin{minipage}[pos][height][contentpos]{width} text \end{minipage}

You can make use of minipage, \parbox, and \pbox to embed paragraphs in non-paragraph boxes. For instance:

    Some very long text\\
    that would not be allowed\\
    in an fbox.

special notes on the pos parameter


The pos alignment parameter does not refer to the \parbox's borders when contentpos is either missing or equal to pos and anytime pos= center! Under any of these circumstances the alignment parameter selects which line of text within the \parbox will be used to align the paragraph box. The \parbox is placed so the baseline of that chosen line of text is aligned to the baseline of the box that the \parbox is glued to. Thus, if the pos is set top, the baseline of the top line of text in the \parbox will line up with the baseline outside of the \parbox. In the special case of a parbox that has only one line of text, that one and only line of text is the top, bottom, and center line of text simultaneously, and changing pos will appear to do nothing unless additional text is added.

If the contentpos is present and not the same as pos and pos is not center, the \parbox will align at its borders.

makebox and mbox

Makebox creates a single-line box, optionally of fixed width, but otherwise large enough to hold its contents. Note that the width does not have to be wider than the contents: for instance, setting width to 0 typesets the content without changing the current position. (E.g., this would allow for an overstrike.) Makebox is typically used to prevent hyphenation (see Hyphenation) or simply to keep text that belongs together from being placed on separate lines. You cannot place line breaks (\\) within a Makebox. mbox is the shorthand no-option version of Makebox.


The pos parameter takes a one letter value: center, flushleft, flushright, or spread the text to fill the box.

\makebox[0pt]{Some text} over this text

\makebox[15ex][s]{Censored text}\hspace{-15ex}\makebox[15ex][s]{X X X X X}

Text \makebox[2\width][r]{running away}

framebox and fbox

The command \framebox behaves identically to \makebox except that it additionally draws a box around its contents.


The following example shows you some things you could do with the \makebox and \framebox commands:

\makebox[\textwidth]{c e n t r a l} \par
\makebox[\textwidth][s]{s p r e a d} \par
\framebox[1.1\width]{Guess I'm framed now!} \par
\framebox[0.8\width][r]{Bummer, I am too wide} \par
\framebox[1cm][l]{never mind, so am I}
Can you read this?


You can tweak the following frame lengths.

  • \fboxsep: the distance between the frame and the content.
  • \fboxrule: the thickness of the rule.

This prints a thick and more distant frame:

\fbox{A frame.}

This shows the box frame of a letter.


The framed package is available that adds the framed environment which provides an an easy way to frame a paragraph within a document:

% ...

This is an easy way to box text within a document!


A savebox is a non-rendered box that is saved for later repeated rendering via the usebox command.

\savebox{\boxname}{some content}

The command \newsavebox creates a placeholder for storing content; the command \savebox stores the specified content in the placeholder without rendering it in the document; and \usebox renders the content of the placeholder into the document.

colorbox and fcolorbox

See Colors. \fcolorbox can also be tweaked with \fboxsep and \fboxrule.


The fancybox package provides additional boxes.

  • \doublebox
  • \ovalbox
  • \shadowbox

box modifiers


Now that we control the horizontal, the obvious next step is to go for the vertical. No problem for LaTeX. The


command lets you define the vertical properties of a box. You can use \width, \height, \depth and \totalheight in the first three parameters, in order to act upon the size of the box inside the text argument. The two optional parameters set for the height and depth of the raisebox. For instance you can observe the difference when embedded in a framebox.

he shouted but not even the next
one in line noticed that something
terrible had happened to him.



See Rotations.

resizebox and scalebox

The graphicx package features additional boxes.

\resizebox{10ex}{2\baselineskip}{Dunhill style}

Rules and Struts


The \rule command in normal use produces a simple black box:


The parameter thickness determines the height, whereas width determines the width of the produced rule. With the optional parameter raise, you can optionally raise or lower the produced rule above or below the baseline.

Here is an example (the thin lines are located at the baseline):



This is useful for drawing vertical and horizontal lines.


A special case is a rule with no width but a certain height. In professional typesetting, this is called a strut. It is used to guarantee that an element on a page has a certain minimal height. You could use it in a tabular environment or in boxes to make sure a row has a certain minimum height.

In LaTeX a strut is defined as


Stretched rules

LaTeX provides the \hrulefill command, which work like a stretched horizontal space. See the Lengths chapter.



Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry

Template:Print entry Template:Print entry

Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry Template:Print entry