Copyright 1994, LaTeX3 Project Team.
All rights reserved.
June 1994
This document describes the new font selection features of the LaTeX Document Preparation System. It is intended for package writers who want to write font-loading packages similar to times or latexsym. This document is only a brief introduction to the new facilities and is intended for package writers who are familiar with TeX fonts and LaTeX packages. It is neither a user-guide nor a reference manual for fonts in LaTeX2e.
The most important difference between LaTeX 2.09 and LaTeX2e is the way that fonts are selected. In LaTeX 2.09, the Computer Modern fonts were built into the LaTeX format, and so customizing LaTeX to use other fonts was a major effort.
In LaTeX2e, very few fonts are built into the format, and there are commands to load new text and math fonts. Packages such as times or latexsym allow authors to access these fonts. This document describes how to write similar font-loading packages.
The LaTeX2e font selection system was first released as the `New Font Selection Scheme' (NFSS) in 1989, and then in release 2 in 1993. LaTeX2e includes NFSS release 2 as standard.
This document contains an overview of the new font commands of LaTeX.
For a general introduction to LaTeX, including the new features of LaTeX2e, you should read LaTeX: A Document Preparation system, Leslie Lamport, Addison Wesley, 2nd ed, 1994.
A more detailed description of the LaTeX font selection scheme is to be found in The LaTeX Companion, Goossens, Mittelbach and Samarin, Addison Wesley, 1994.
The LaTeX font selection scheme is based on TeX, which is described by its developer in The TeX book, Donald E. Knuth, Addison Wesley, 1986, revised in 1991 to include the features of TeX 3.
Sebastian Rahtz's psnfss software contains the software for using a large number of Type 1 fonts (including the Adobe Laser Writer 35 and the Monotype CD-ROM fonts) in LaTeX. It should be available from the same source as your copy of LaTeX.
The psnfss software uses fonts generated by Alan Jeffrey's fontinst software. This can convert fonts from Adobe Font Metric format into a format readable by LaTeX, including the generation of the .fd files described in Section here. The fontinst software should be available from the same source as your copy of LaTeX.
Whenever practical, LaTeX uses the font naming scheme described in Filenames for fonts, Karl Berry, TUGboat 11(4), 1990.
The class-writer's guide LaTeX2e for Class and Package Writers describes the new LaTeX features for writers of document classes and packages and is kept in clsguide.tex.
We are gradually turning the source code for LaTeX into a LaTeX document LaTeX: the program. This document includes an index of LaTeX commands and can be typeset from source2e.tex.
For more information about TeX and LaTeX, please contact your local TeX Users Group, or the international TeX Users Group, P. O.\ Box 869, Santa Barbara, CA 93102-0869, USA, Fax: +1 805 963 8358, E-mail: tug@tug.org.
This section describes the commands available to class and package writers for specifying and selecting fonts.
Every text font in LaTeX has five attributes:
OT1 & TeX text
T1 & TeX extended text
OML & TeX math italic
OMS & TeX math symbols
OMX & TeX math large symbols
U & Unknown
L<xx> & A local encoding
The `local' encodings are intended for font encodings which are only locally available, for example a font containing a university logo in various sizes.
There are far too many font families to list them all, but some common ones are:
cmr & Computer Modern Roman
cmss & Computer Modern Sans
cmtt & Computer Modern Typewriter
cmm & Computer Modern Math Italic
cmsy & Computer Modern Math Symbols
cmex & Computer Modern Math Extensions
ptm & Adobe Times
phv & Adobe Helvetica
pcr & Adobe Courier
The most common values for the font series are:
m & Medium
b & Bold
bx & Bold extended
sb & Semi-bold
c & Condensed
The most common values for the font shape are:
n & Normal (that is `upright' or `roman')
it & Italic
sl & Slanted (or `oblique')
sc & Caps and small caps
The font size is specified as a dimension, for example 10pt or 1.5in or 3mm. These five parameters specify every LaTeX font, for example:
LaTeX specification &
Font &
TeX font name
OT1 & cmr & m & n & 10pt &
Computer Modern Roman 10pt &
cmr10
OT1 & cmss & m & sl & 12pt &
Computer Modern Sans Oblique 12pt &
cmssi12
OML & cmm & m & it & 10pt &
Computer Modern Math Italic 10pt &
cmmi10
T1 & ptm & b & it & 18pt &
Adobe Times Bold Italic 18pt &
ptmbq at 18pt
These five parameters are displayed whenever LaTeX gives an overfull box warning, for example:
Overfull \hbox (3.80855pt too wide) in paragraph at lines 314--318 []\OT1/cmr/m/n/10 Normally [] and [] will be iden-ti-cal,
The author commands for fonts set the five attributes:
Author command &
Attribute &
Value in article class
\textrm{..} or \rmfamily & family & cmr
\textsf{..} or \sffamily & family & cmss
\texttt{..} or \ttfamily & family & cmtt
\textmd{..} or \mdseries & series & m
\textbf{..} or \bfseries & series & bx
\textup{..} or \upshape & shape & n
\textit{..} or \itshape & shape & it
\textsl{..} or \slshape & shape & sl
\textsc{..} or \scshape & shape & sc
\tiny & size & 5pt
\scriptsize & size & 7pt
\footnotesize & size & 8pt
\small & size & 9pt
\normalsize & size & 10pt
\large & size & 12pt
\Large & size & 14.4pt
\LARGE & size & 17.28pt
\huge & size & 20.74pt
\Huge & size & 24.88pt
The values used by these commands are determined by the document class, using the parameters defined in Section here.
Note that there are no author commands for selecting new encodings. These should be provided by packages, such as the fontenc package.
This section does not explain how LaTeX font specifications are turned into TeX font names. This is described in Section here.
The low-level commands used to select a text font are as follows.
\fontencoding {encoding}
\fontfamily {family}
\fontseries {series}
\fontshape {shape}
\fontsize {size} {baselineskip}
Each of these commands sets one of the font attributes; \fontsize also sets \baselineskip. The actual font in use is not altered by these commands, but the current attributes are used to determine which font to use after the next \selectfont command.
\selectfont
Selects a text font, based on the current values of the font attributes.
Warning: there must be a \selectfont command immediately after any settings of the font parameters by (some of) the five \font<parameter> commands, before any following text. For example, it is legal to say:
\fontfamily{ptm}\fontseries{b}\selectfont Some text.
but it is not legal to say:
\fontfamily{ptm} Some \fontseries{b}\selectfont text.
You may get unexpected results if you put text between a \font<parameter> command and a \selectfont.
\usefont {encoding} {family} {series} {shape}
A short hand for the equivalent \font\ldots commands followed by a call to \selectfont.
The current values of the font attributes are held in internal macros.
\f@encoding
\f@family
\f@series
\f@shape
\f@size
\f@baselineskip
\tf@size
\sf@size
\ssf@size
These hold the current values of the encoding, the family, the series, the shape, the size, the baseline skip, the main math size, the script math size and the scriptscript math size. The last four are accessible only within a formula; outside of it they may contain arbitrary values.
For example, to set the size to 12 without changing the baseline skip:
\fontsize{12}{\f@baselineskip}
However, you should never alter the values of the internal commands directly; they must only be modified using the low-level commands like \fontfamily, \fontseries, etc. If you disobey this warning you might produce code that loops.
The parameter values set by author commands such as \textrm and \rmfamily, etc. are not hard-wired into LaTeX; instead these commands use the values of a number of parameters set by the document class and packages. For example, \rmdefault is the name of the default family selected by \textrm and \rmfamily. Thus to set a document in Adobe Times, Helvetica and Courier, the document designer specifies:
\renewcommand{\rmdefault}{ptm}
\renewcommand{\sfdefault}{phv}
\renewcommand{\ttdefault}{pcr}
\encodingdefault
\familydefault
\seriesdefault
\shapedefault
The encoding, family, series and shape of the main body font. By default these are OT1, \rmdefault, m and n. Note that since the default family is \rmdefault, this means that changing \rmdefault will change the main body font of the document.
\rmdefault
\sfdefault
\ttdefault
The families selected by \textrm, \rmfamily, \textsf, \sffamily, \texttt and \ttfamily. By default these are cmr, cmss and cmtt.
\bfdefault
\mddefault
The series selected by \textbf, \bfseries, \textmd and \mdseries. By default these are bx and m. These values are suitable for the default families used. If other fonts are used as standard document fonts, for example, some PostScript fonts, it might be necessary to adjust the value of \bfdefault to b since only a few such families have `bold extended' series. An alternative (taken for the fonts provided by psnfss) is to define silent substitutions from bx series to b series with special \DeclareFontShape declarations and the ssub size function, see Section here.
\itdefault
\sldefault
\scdefault
\updefault
The shapes selected by \textit, \itshape, \textsl, \slshape, \textsc, \scshape, \textup and \upshape. By default these are it, sl, sc and n.
Note that there are no parameters for the size commands. These should be defined directly in class files, for example:
\renewcommand{\normalsize}{\fontsize{10}{12}\selectfont}
More elaborate examples (setting additional parameters when the text size is changed) can be found in classes.dtx the source documentation for the classes article, report, and book.
\DeclareFixedFont {cmd} {ENC} {family} {series} {shape} {size}
Declares command <cmd> to be a font switch which selects the font that is specified by the attributes <ENC>, <family>, <series>, <shape>, and <size>.
The font is selected without any adjustments to baselineskip and other surrounding conditions.
This example makes {\picturechar .} select a small dot very quickly:
\DeclareFixedFont{\picturechar}{OT1}{cmr}{m}{n}{5}
\DeclareTextFontCommand {cmd} {font-switches}
Declares command <cmd> to be a font command with one argument. The current font attributes are locally modified by <font-switches> and then the argument of <cmd> is typeset in the resulting new font.
Commands defined by \DeclareTextFontCommand automatically take care of any necessary italic correction (on either side).
The following example shows how \textrm is defined by the kernel.
\DeclareTextFontCommand{\textrm}{\rmfamily}
To define a command that always typeset its argument in the italic shape of the main document font you could declare:
\DeclareTextFontCommand{\normalit}{\normalfont\itshape}
This declaration can be used to change the meaning of a command; if <cmd> is already defined, a log that it has been redefined is put in the transcript file.
\DeclareOldFontCommand {cmd} {text-switch} {math-switch}
Declares command <cmd> to be a font switch (i.e. used with the syntax {<cmd>...}) having the definition <text-switch> when used in text and the definition <math-switch> when used in a formula. Math alphabet commands, like \mathit, when used within <math-switch> should not have an argument. Their use in this argument causes their semantics to change so that they here act as a font switch, as required by the usage of the <cmd>.
This declaration is useful for setting up commands like \rm to behave as they did in LaTeX 2.09. We strongly urge you not to misuse this declaration to invent new font commands.
The following example defines \it to produce the italic shape of the main document font if used in text and to switch to the font that would normally be produced by the math alphabet \mathit if used in a formula.
\DeclareOldFontCommand{\it}{\normalfont\itshape}{\mathit}
This declaration can be used to change the meaning of a command; if <cmd> is already defined, a log that it has been redefined is put in the transcript file.
This section describes the commands available to class and package writers for specifying math fonts and math commands.
The selection of fonts within math mode is quite different to that of text fonts.
Some math fonts are selected explicitly by one-argument commands such as \mathsf{max} or \mathbf{vec}: these are called math alphabets. These math alphabet commands affect only the font used for letters and symbols of type \mathalpha (see Section here); other symbols within the argument will be left unchanged. The predefined math alphabets are:
Alphabet & Description & Example
\mathnormal & default & $abcXYZ$
\mathrm & roman & $\mathrm{abcXYZ}$
\mathbf & bold roman & $\mathbf{abcXYZ}$
\mathsf & sans serif & $\mathsf{abcXYZ}$
\mathit & text italic & $\mathit{abcXYZ}$
\mathtt & typewriter & $\mathtt{abcXYZ}$
\mathcal & calligraphic & $\mathcal{XYZ}$
Other math fonts are selected implicitly by TeX for symbols, with commands such as \oplus (producing $\oplus$) or with straight characters like >> or +. Fonts containing such math symbols are called math symbol fonts. The predefined math symbol fonts are:
Symbol font & Description & Example
operators & symbols from \mathrm & $[\;+\;]$
letters & symbols from \mathnormal & $<<\star>>$
symbols & most LaTeX symbols & $\leq*\geq$
largesymbols & large symbols & $\sum\prod\int$
Some math fonts are both math alphabets and math symbol fonts, for example \mathrm and operators are the same font, and \mathnormal and letters are the same font.
Math fonts in LaTeX have the same five attributes as text fonts: encoding, family, series, shape and size. However, there are no commands that allow the attributes to be individually changed. Instead, the conversion from math fonts to these five attributes is controlled by the math version. For example, the normal math version maps:
Math font &
External font
Alphabets & Symbol fonts &
Attributes
\mathnormal & letters & OML & cmm & m & it
\mathrm & operators & OT1 & cmr & m & n
\mathcal & symbols & OMS & cmsy & m & n
& largesymbols & OMX & cmex & m & n
\mathbf & & OT1 & cmr & bx & n
\mathsf & & OT1 & cmss & m & n
\mathit & & OT1 & cmr & m & it
\mathtt & & OT1 & cmtt & m & n
The bold math version is similar except that it contains bold fonts. The command \boldmath selects the bold math version.
Math versions can only be changed outside of math mode.
The two predefined math versions are:
normal & the default math version
bold & the bold math version
Packages may define new math alphabets, math symbol fonts, and math versions. This section describes the commands for writing such packages.
There are no commands for selecting symbol fonts. Instead, these are selected indirectly through symbol commands like \oplus. Section here explains how to define symbol commands.
\mathnormal{<math>}
\mathcal{<math>}
\mathrm{<math>}
\mathbf{<math>}
\mathsf{<math>}
\mathit{<math>}
\mathtt{<math>}
Each math alphabet is a command which can only be used inside math mode. For example, $x + \mathsf{y} + \mathcal{Z}$ produces $x + \mathsf{y} + \mathcal{Z}$.
\mathversion{<version-name>}
This command selects a math version; it can only be used outside math mode. For example, \boldmath is defined to be \mathversion{bold}.
\DeclareMathVersion {version-name}
Defines <version-name> to be a math version.
The newly declared version is initialised with the defaults for all symbol fonts and math alphabets declared so far (see the commands \DeclareSymbolFont and \DeclareMathAlphabet).
If used on an already existing version, an information message is written to the transcript file and all previous \SetSymbolFont or \SetMathAlphabet declarations for this version are overwritten by the math alphabet and symbol font defaults, i.e. one ends up with a virgin math version.
Example:
\DeclareMathVersion{normal}
\DeclareMathAlphabet {math-alph} {encoding} {family} {series} {shape}
Defines <math-alph> to be a new math alphabet.
The arguments <encoding> <family> <series> <shape> are the default values for this math alphabet in all math versions; these can be reset later for a particular math version by a \SetMathAlphabet command. If <shape> is empty then the <math-alph> is declared to be invalid in all versions, unless it is set by a later \SetMathAlphabet command.
Checks that <math-alph> can be used and that <encoding> is a valid encoding scheme.
In these examples, \foo is defined everywhere but \baz, by default, is defined nowhere.
\DeclareMathAlphabet{\foo}{OT1}{cmtt}{m}{n}
\DeclareMathAlphabet{\baz}{OT1}
\SetMathAlphabet {math-alph} {version-name}
\null\hfill{encoding} {family} {series} {shape}
Changes, or sets, the font for the math alphabet <math-alph> in math version <version-name> to <encoding><family><series><shape>.
Checks that <math-alph> is a math alphabet, <version-name> is a math version and <encoding> is a known encoding scheme.
This example defines \baz for the `normal' math version only:
\SetMathAlphabet\baz{normal}{OT1}{cmss}{m}{n}
Note that this declaration is not used for all math alphabets: Section here describes \DeclareSymbolFontAlphabet, which is used to set up math alphabets contained in fonts which have are declared as symbol fonts.
\DeclareSymbolFont {sym-font-name} {encoding} {family} {series} {shape}
Defines <sym-font-name> to be a new symbol font.
The arguments <encoding> <family> <series> <shape> are the default values for this symbol font in all math versions; these can be reset later for a particular math version by a \SetSymbolFont command.
Checks that <sym-font-name> can be used and that <encoding> is a declared encoding scheme.
For example, the following sets up the first four standard math symbol fonts:
\DeclareSymbolFont{operators}{OT1}{cmr}{m}{n}
\DeclareSymbolFont{letters}{OML}{cmm}{m}{it}
\DeclareSymbolFont{symbols}{OMS}{cmsy}{m}{n}
\DeclareSymbolFont{largesymbols}{OMX}{cmex}{m}{n}
\SetSymbolFont {sym-font-name} {version name}
\null\hfill {encoding} {family} {series} {shape}
Changes the symbol font <sym-font-name> for math version <version name> to <encoding> <family> <series> <shape>.
Checks that <sym-font-name> is a symbol font, <version name> is a known math version and <encoding> is a declared encoding scheme.
For example, the following come from the set up of the `bold' math version:
\SetSymbolFont{operators}{bold}{OT1}{cmr}{bx}{n}
\SetSymbolFont{letters}{bold}{OML}{cmm}{b}{it}
\DeclareSymbolFontAlphabet {math-alph} {sym-font-name}
Allows the previously declared symbol font <sym-font-name> to be also the math alphabet <id> (in all math versions).
This declaration should be used in preference to \DeclareMathAlphabet and \SetMathAlphabet when a math alphabet is the same as a symbol font; this is because it makes better use of the limited number (only 16) of TeX's math groups.
Checks that <math-alph> can be defined and that <sym-font-name> is a symbol font.
Example:
\DeclareSymbolFontAlphabet{\mathrm}{operators}
\DeclareSymbolFontAlphabet{\mathcal}{symbols}
\DeclareMathSymbol {symbol} {type} {sym-font-name} {slot}
The <symbol> can be either a single character such as `>>', or a macro name, such as \sum.
Defines the <symbol> to be a math symbol of type <type> in slot <slot> of symbol font <sym-font-name>. The <type> can be given as a number or as a command:
Type & Meaning & Example
0 or \mathord & Ordinary & $\alpha$
1 or \mathop & Large operator & $\sum$
2 or \mathbin & Binary operation & $\times$
3 or \mathrel & Relation & $\leq$
4 or \mathopen & Opening & $\langle$
5 or \mathclose & Closing & $\rangle$
6 or \mathpunct & Punctuation & $;$
7 or \mathalpha & Alphabet character & $A$
Only symbols of type \mathalpha will be affected by math alphabet commands: within the argument of a math alphabet command they will produce the character in slot <slot> of that math alphabet's font. Symbols of other types will always produce the same symbol (within one math version).
\DeclareMathSymbol allows a macro <symbol> to be redefined only if it was previously defined to be a math symbol. It also checks that the <sym-font-name> is a declared symbol font.
Example:
\DeclareMathSymbol{\alpha}{0}{letters}{"0B}
\DeclareMathSymbol{\lessdot}{\mathbin}{AMSb}{"0C}
\DeclareMathSymbol{\foo}{\mathalpha}{AMSb}{"0C}
\DeclareMathDelimiter {cmd} {type} {sym-font-name-1} {slot-1}
\null\hfill {sym-font-name-2} {slot-2}
Defines <cmd> to be a math delimiter where the small variant is in slot <slot-1> of symbol font <sym-font-name-1> and the large variant is in slot <slot-2> of symbol font <sym-font-name-2>. Both symbol fonts must have been declared previously.
Checks that <sym-font-name-i> are both declared symbol fonts.
If TeX is not looking for a delimiter, <cmd> is treated just as if it had been defined with \DeclareMathSymbol using <type>,<sym-font-name-1> and <slot-1>. In other words, if a command is defined as a delimiter then this automatically defines it as a math symbol.
Example:
\DeclareMathDelimiter{\langle}{\mathopen}{symbols}{"68}
{largesymbols}{"0A}
\DeclareMathAccent {cmd} {type} {sym-font-name} {slot}
Defines <cmd> to act as a math accent.
The accent character comes from slot <slot> in <sym-font-name>. The <type> can be either \mathord or \mathalpha; in the latter case the accent character changes font when used in a math alphabet.
Example:
\DeclareMathAccent{\acute}{\mathalpha}{operators}{"13}
\DeclareMathAccent{\vec}{\mathord}{letters}{"7E}
\DeclareMathRadical {cmd} {sym-font-name-1} {slot-1}
\null\hfill {sym-font-name-2} {slot-2}
Defines <cmd> to be a radical where the small variant is in slot <slot-1> of symbol font <sym-font-name-1> and the large variant is in slot <slot-2> of symbol font <sym-font-name-2>. Both symbol fonts must have been declared previously.
Example (probably the only use for it!):
\DeclareMathRadical{\sqrt}{symbols}{"70}{largesymbols}{"70}
\DeclareMathSizes {t-size} {mt-size} {s-size} {ss-size}
Declares that <mt-size> is the math text size, <s-size> is the script size and <ss-size> the scriptscript size to be used in math, when <t-size> is the current text size. For text sizes for which no such declaration is given the script and scriptscript size will be calculated and then fonts are loaded for the calculated sizes or the best approximation (this may result in a warning message).
Normally, <t-size> and <mt-size> will be identical; however, if, for example, PostScript text fonts are mixed with bit-map math fonts then you may not have available a <mt-size> for every <t-size>.
Example:
\DeclareMathSizes{13.82}{12.4}{10}{7}
This section explains how LaTeX's font attributes are turned into TeX font specifications.
The description of how LaTeX font attributes are turned into TeX fonts is usually kept in a font definition (.fd) file. The .fd file for family <family> in encoding <encoding> is called <ENC><family>.fd, for example OT1cmr.fd for Computer Modern Roman or T1ptm.fd for Adobe Times.
Whenever LaTeX encounters an encoding/family combination that it does not know (e.g. if the document designer says \fontfamily{ptm}\selectfont) then LaTeX attempts to load the appropriate .fd file. ``Not known'' means: there was no \DeclareFontFamily declaration issued for this encoding/family combination.
The declarations in the .fd file are responsible for telling LaTeX how to load fonts for that encoding/family combination. If the .fd file could not be found, a warning is issued and font substitutions are made.
The .fd files should contain only commands from this section. Note that those commands can also be used outside a .fd file: they can be put in package or class files, or even in the preamble of a document.
\ProvidesFile{<file-name>}[<release-info>]
The .fd file should announce itself with a \ProvidesFile command, as described in LaTeX2e for Class and Package Writers. For example:
\ProvidesFile{T1ptm.fd}[1994/06/01 Adobe Times font definitions]
\DeclareFontFamily {encoding} {family} {loading-settings}
Declares a font family <family> to be available in encoding scheme <encoding>.
The <loading-settings> are executed immediately after loading any font with this encoding and family.
Checks that <encoding> was previously declared.
This example refers to the Computer Modern Typewriter font family in the Cork encoding:
\DeclareFontFamily{T1}{cmtt}{\hyphenchar\font=-1}
Each .fd file should contain exactly one \DeclareFontFamily command, and it should be for the appropriate encoding/family combination.
\DeclareFontShape {encoding} {family} {series} {shape}
\null\hfill {loading-info} {loading-option}
Declares a font shape combination; here <loading-info> contains the information that combines sizes with external fonts. The syntax is complex and is described in Section here below.
The <loading-settings> are executed after loading any font with this font shape. They are executed immediately after the `loading-settings' which were declared by \DeclareFontFamily and so they can be used to overwrite the settings made at the family level.
Checks that the combination <encoding><family> was previously declared via \DeclareFontFamily.
Example:
\DeclareFontShape{OT1}{cmr}{m}{sl}{%
<<5-8>> sub * cmr/m/n
<<8>> cmsl8
<<9>> cmsl9
<<10>> <<10.95>> cmsl10
<<12>> <<14.4>> <<17.28>> <<20.74>> <<24.88>> cmsl12
}
An .fd file can contain any number of \DeclareFontShape commands, which should be for the appropriate <encoding> and <family>.
The information which tells LaTeX exactly which font ({\tt .tfm}) files to load is contained in the <loading-info> part of a \DeclareFontShape declaration. This part consists of one or more <fontshape-decl>s, each of which has the following form:
<fontshape-decl> & <size-infos> <font-info>
<size-infos> & <size-infos> <size-info> $\mid$
<size-info>
<size-info> & ``<<'' <number-or-range> ``>>''
<font-info> & $[$ <size-function> ``*'' $]$
$[$ ``['' <optarg> ``]'' $]$ <fontarg>
The <number-or-range> denotes the size or size-range for which this entry applies.
If it contains a hyphen it is a range: lower bound on the left (if missing, zero implied), upper bound on the right (if missing, $\infty$ implied). For ranges, the upper bound is not included in the range and the lower bound is.
Examples:
<<10>> & simple size& 10pt only
<<-8>> & range& all sizes less than 8pt
<<8-14.4>> & range& all sizes greater than or equal to 8pt
& & but less than 14.4pt
<<14.4->> & range& all sizes greater than or equal 14.4pt
If more than one <size-info> entry follows without any intervening <font-info>, they all share the next <font-info>.
The <size-function>, if present, handles the use of <font-info>. If not present, the `empty' <size-function> is assumed.
All the <size-info>s are inspected in the order in which they appear in the font shape declaration. If a <size-info> matches the requested size, its <size-function> is executed. If \external@font is non-empty afterwards this process stops, otherwise the next <size-info> is inspected. (See also \DeclareSizeFunction.)
If this process does not lead to a non-empty \external@font, LaTeX tries the nearest simple size. If the entry contains only ranges an error is returned.
LaTeX provides the following size functions, whose `inputs' are <fontarg> and <optarg> (when present).
Examples for the use of most of the above size functions can be found in the file cmfonts.fdd---the source for the standard .fd files describing the Computer Modern fonts by Donald Knuth.
\DeclareSizeFunction {name} {code}
Declares a size-function <name> for use in \DeclareFontShape commands. The interface is still under development but there should be no real need to a define new size functions.
The <code> is executed when the size or size-range in \DeclareFontShape matches the user-requested size.
The arguments of the size-function are automatically parsed and placed into \mandatory@arg and \optional@arg for use in <code>. Also available, of course, is \f@size, which is the use- requested size.
To signal success <code> must define the command \external@font to contain the external name and any scaling options (if present) for the font to be loaded.
This example sets up the `empty' size function (simplified):
\DeclareSizeFunction
{\edef\external@font{\mandatory@arg\space at\f@size}
This section explains how to declare and use new font encodings and how to declare commands for use with particular encodings.
Users can select new font encodings using the fontenc package. The fontenc package has options for encodings; the last option becomes the default encoding. For example, to use the OT2 (Washington University Cyrillic encoding) and T1 encodings, with T1 as the default, an author types:
\usepackage[OT2,T1]{fontenc}
This package loads the encoding definition files (<ENC>enc.def files) for each font encoding <ENC> given as an option but not already declared; it also sets \encodingdefault to be the last encoding in the option list.
LaTeX currently predefines the OT1 and T1 text encodings, and provides the files OT1enc.def and T1enc.def. Other encoding set-ups might be added to the distribution at a later stage.
Thus the example above loads the file OT2enc.def and sets \encodingdefault to T1.
{\bf Warning:} Some aspects of the contents of font definition files are still under development. Therefore, the current versions of the files OT1enc.def and T1enc.def are temporary versions and should not be used as models for producing further such files. For further information you should read the documentation in ltoutenc.dtx.
The declarations in the encoding definition file are responsible for declaring the encoding and telling LaTeX how to produce characters in this encoding.
The <ENC>enc.def files should contain only commands from this section. As with the font definition file commands, it is also possible (although normally not necessary) to use these declarations directly within a class or package file.
\ProvidesFile{<file-name>}[<release-info>]
The <ENC>enc.def file should announce itself with a \ProvidesFile command, described in LaTeX2e for Class and Package Writers. For example:
\ProvidesFile{OT2enc.def}
[1994/06/01 Washington Univ. Cyrillic encoding]
\DeclareFontEncoding {encoding} {text-settings} {math-settings}
Declares a new encoding scheme <encoding>.
The <text-settings> are declarations which are executed every time \selectfont changes the encoding to be <encoding>.
The <math-settings> are similar but are for math alphabets. They are executed whenever a math alphabet with this encoding is called.
Spaces within the arguments are ignored to avoid surplus spaces in the document. If a real space is necessary use \space.
Example:
\DeclareFontEncoding{OT1}
Some author commands need to change their definition depending on which encoding is currently in use. For example, in the OT1 encoding, the letter `\AE' is in slot "1D, whereas in the T1 encoding it is in slot "C6. So the definition of \AE has to change depending on whether the current encoding is OT1 or T1. The following commands allow this to happen.
\DeclareTextCommand {cmd} {encoding} [num] [default] {definition}
This command is like \newcommand, except that it defines a command which is specific to one encoding. For example, the definition of \aa in the OT1 encoding is:
\DeclareTextCommand{\aa}{OT1}{{\accent23a}}
\DeclareTextCommands takes the same optional arguments as \newcommand.
The resulting command is robust, even if the code in <definition> is fragile.
It does not produce an error if the command has already been defined but logs the redefinition in the transcript file.
\DeclareTextSymbol {cmd} {encoding} {slot}
This command defines a text symbol with slot <slot> in the encoding. For example, the definition of \ss in the OT1 encoding is:
\DeclareTextSymbol{\ss}{OT1}{25}
It does not produce an error if the command has already been defined but logs the redefinition in the transcript file.
\DeclareTextAccent {cmd} {encoding} {slot}
This command declares a text accent, with the accent taken from slot <slot> in the encoding. For example, the definition of \" in the OT1 encoding is:
\DeclareTextAccent{\"}{OT1}{127}
It does not produce an error if the command has already been defined but logs the redefinition in the transcript file.
\DeclareTextComposite {cmd} {encoding} {letter} {slot}
This command declares that the composite letter formed from applying <cmd> to <letter> is defined to be simply slot <slot> in the encoding. The <letter> should be a single letter (such as a) or a single command (such as \i).
For example, the definition of \'{a} in the T1 encoding could be declared like this:
\DeclareTextComposite{\'}{T1}{a}{225}
The <cmd> should have been previously declared with \DeclareTextAccent, or as a one-argument \DeclareTextCommand.
\DeclareFontEncodingDefaults {text-settings} {math-settings}
Declares <text-settings> and <math-settings> for all encoding schemes. These are executed before the encoding scheme dependent ones are executed so that one can use the defaults for the major cases and overwrite them if necessary using \DeclareFontEncoding.
If \relax is used as an argument, the current setting of this default is left unchanged.
This example is used by amsfonts.sty for accent positioning; it changes only the math settings:
\DeclareFontEncodingDefaults{\relax}{\def\accentclass}
\DeclareFontSubstitution {encoding} {family} {series} {shape}
Declares the default values for font substitution which will be used when a font with encoding <encoding> should be loaded but no font can be found with the current attributes.
These substitutions are local to the encoding scheme because the encoding scheme is never substituted! They are tried in the order <shape> then <series> and finally <family>.
If no defaults are set up for an encoding, the values given by \DeclareErrorFont are used.
The font specification for <encoding><family><series><shape> must have been defined by \DeclareFontShape before the is reached.
Example:
\DeclareFontSubstitution{T1}{cmr}{m}{n}
This section covers the remaining font commands of LaTeX.
\DeclareErrorFont {encoding} {family} {series} {shape} {size}
Declares <encoding><family><series><shape> to be the font shape used in cases where the standard substitution mechanism fails (i.e. would loop). For the standard mechanism see the command \DeclareFontSubstitution above.
The font specification for <encoding><family><series><shape> must have been defined by \DeclareFontShape before the is reached.
Example:
\DeclareErrorFont{OT1}{cmr}{m}{n}{10}
\fontsubfuzz
This parameter is used to decide whether or not to produce a terminal warning if a font size substitution takes place. If the difference between the requested and the chosen size is less than \fontsubfuzz the warning is only written to the transcript file. The default value is 0.4pt. This can be redefined with \renewcommand, for example:
\renewcommand{\fontsubfuzz}{0pt} % always warn
\DeclarePreloadSizes {encoding} {family} {series} {shape} {size-list}
Specifies the fonts that should be pre-loaded by the format. These commands should be put in a preload.cfg file, which is read in when the LaTeX format is being built. Read preload.dtx for more information on how to built such a configuration file.
Example:
\DeclarePreloadSizes{OT1}{cmr}{m}{sl}{10,10.95,12}
Encoding schemes which are local to a site should start with L.
Whenever possible, you should use the series and shape names suggested in The LaTeX Companion since this will make it easier to combine new fonts with existing fonts.