diff options
Diffstat (limited to 'info/impatient/concepts.tex')
| -rw-r--r-- | info/impatient/concepts.tex | 4097 |
1 files changed, 4097 insertions, 0 deletions
diff --git a/info/impatient/concepts.tex b/info/impatient/concepts.tex new file mode 100644 index 0000000000..2b38675345 --- /dev/null +++ b/info/impatient/concepts.tex @@ -0,0 +1,4097 @@ +% This is part of the book TeX for the Impatient. +% Copyright (C) 2003, 2014 Paul W. Abrahams, Kathryn A. Hargreaves, Karl Berry. +% See file fdl.tex for copying conditions. + +\input macros +\chapter{Concepts} + +\chapterdef{concepts} + +This part of the book contains definitions and explanations of +the concepts that we use in describing \TeX. +The concepts include both +technical terms that we use in explaining the commands and +important topics that don't fit elsewhere in the book. + +The concepts are arranged alphabetically. +The inside back cover of the book contains a complete list of +concepts and the pages on which they are explained. +We suggest that you make a copy of the inside back cover and keep it nearby +so that you'll be able to identify and look up an unfamiliar +concept immediately. +As far as possible, we've kept our terminology consistent with that of +\texbook.\idxref{\texbook} + +\beginconcepts + +\conceptindex{active characters} +\concept {active character} + +An \defterm{active character} is a \refterm{character} +that has a definition, e.g., a macro definition, associated with it. +^^{macros//named by active characters} +You can think of an active character as a special kind of control sequence. +When \TeX\ encounters an active character, it +executes the definition associated with the character. +If \TeX\ encounters an active character that does not have +an associated definition, it will complain about an +undefined control sequence. + +An active character has a \refterm{category code} of $13$ (the value +of ^|\active|). +To define an active character, you should first +use the ^|\catcode| command +\ctsref{\catcode} to make it active +and then provide the definition of the character, using +a command such as |\def|, |\let|, or |\chardef|. +The definition of an active character has the same form as +the definition of a \refterm{control sequence}. +^^{category codes//of active characters} +If you try to define the macro for an active character +before you make the character active, \TeX\ will complain about a +missing control sequence. + +For example, the tilde character (|~|) is defined as an active character +in \plainTeX. It +produces a space between two words but links those words so that +\TeX\ will not turn the space into a \refterm{line break}. +\refterm{\PlainTeX:\plainTeX} defines `|~|' by the commands: + +\csdisplay +\catcode `~ = \active \def~{\penalty10000\!visiblespace} +| +(The |\penalty| inhibits a line break and the `|\!visiblespace|' +inserts a space.) +\endconcept + + +\conceptindex{alignments} +\concept alignment + +\bix^^{tables} +An \defterm{alignment} is a construct for aligning material, such as a +table, in columns or rows. To form an alignment you need to +(a)~describe the layout of the columns or rows and (b)~tell \TeX\ what +material goes into the columns or rows. A tabbing alignment or a +horizontal alignment is organized as a sequence of rows; a vertical +alignment is organized as a sequence of columns. We first describe +tabbing and horizontal alignments and then more briefly describe +vertical alignments. + +Tabbing alignments are defined by \plainTeX. They are simpler but less +flexible than horizontal alignments. Tabbing and horizontal alignments +differ principally in how you describe their layouts. + +\bix^^|\settabs| +\bix\ctsidxref{+} +\bix\ctsidxref{cr} + +To construct a tabbing alignment you first issue a |\settabs| command +\ctsref{\settabs} that specifies how \TeX\ should divide the available +horizontal space into columns. Then you provide a sequence of rows for +the table. Each row consists of a |\+| control sequence \ctsref{\@plus} +followed by a list of ``entries'', i.e., row\slash column intersections. +^^{entry (column or row)} +Adjacent entries in a row are separated by an ampersand (|&|). +\xrdef{@and} +\ttidxref{&} +The end of a row is indicated by ^|\cr| after its +last entry. +If a row has fewer entries than there are columns in the alignment, +\TeX\ effectively fills out the row with blank entries. + +As long as it's preceded by a |\settabs| command, you can put a row of a +tabbing alignment anywhere in your document. In particular, you can put +other things between the rows of a tabbing alignment or describe several +tabbing alignments with a single |\settabs|. Here's an example of a +tabbing alignment: + +\xrdef{tabbedexample}\csdisplay +{\hsize = 1.7 in \settabs 2 \columns +\+cattle&herd\cr +\+fish&school\cr +\+lions&pride\cr} +| +The |\settabs 2 \columns| command in this example \ctsref{\settabs} +tells \TeX\ to produce two equally wide columns. +The line length is $1.7$ inches. +The typeset alignment looks like this: + +{\def\+{\tabalign}% so it isn't \outer. +\vdisplay{% +\hsize 1.7 in \settabs 2 \columns +\+cattle&herd\cr +\+fish&school\cr +\+lions&pride\cr} +}% + +\margin{Missing explanation added here.} +There's another form of tabbing alignment in which you specify the column +widths with a template. The column widths in the template +determine the column widths in the rest of the alignment: +\csdisplay +{\settabs\+cattle\quad&school\cr +\+cattle&herd\cr +\+fish&school\cr +\+lions&pride\cr} +| +Here's the result: +{\def\+{\tabalign}% so it isn't \outer. +\vdisplay{% +\settabs\+cattle\quad&school\cr +\+cattle&herd\cr +\+fish&school\cr +\+lions&pride\cr} +}% + +\eix^^|\settabs| +\eix\ctsidxref{+} +\bix^^|\halign| +Horizontal alignments are constructed with |\halign| \ctsref\halign. +\TeX\ adjusts the column widths of a horizontal alignment according to +what is in the columns. When \TeX\ encounters the |\halign| command +that begins a horizontal alignment, it first examines all the rows of +the alignment to see how wide the entries are. It then sets each column +width to accommodate the widest entry in that column. + +A horizontal alignment governed by |\halign| consists of a +``\pix^{preamble}'' that indicates the row layout followed by the rows +themselves. +\ulist +\li The preamble consists of a sequence of \pix^{template}s, one for each +column. The template for a column specifies how the text for that +column should be typeset. Each template must include a single |#| +character +\ttidxref{#}\xrdef{@asharp} +to indicate where \TeX\ should substitute the text of an entry into the +template. The templates are separated by ampersands (|&|), \ttidxref{&} +and the end of the preamble is indicated by |\cr|. By providing an +appropriate template you can obtain effects such as centering a column, +left or right justifying a column, or setting a column in a particular +\refterm{font}. + +\li The rows have the same form as in a tabbing alignment, except that +you omit the |\+| at the beginning of each row. +As before, entries are separated by |&| and the end of the row +is indicated by |\cr|. +\TeX\ treats each entry as a +\refterm{group}, so any +font-setting command or other \refterm{assignment} +in a column template is in effect only for the entries in that column. +\endulist +\noindent The preamble and the rows must all be enclosed in the braces +that follow |\halign|. Each |\halign| alignment must include +its own preamble. + +For example, the horizontal alignment: +\csdisplay +\tabskip=2pc +\halign{\hfil#\hfil &\hfil#\hfil &\hfil#\hfil \cr + &&\it Table\cr +\noalign{\kern -2pt} + \it Creature&\it Victual&\it Position\cr +\noalign{\kern 2pt} + Alice&crumpet&left\cr + Dormouse&muffin&middle\cr + Hatter&tea&right\cr} +| + +\noindent produces the result: + +\xrdef{halignexample} +\vdisplay{% +\tabskip=2pc \halign{\hfil#\hfil &\hfil#\hfil &\hfil#\hfil \cr + &&\it Table\cr +\noalign{\kern -2pt} + \it Creature&\it Victual&\it Position\cr +\noalign{\kern 2pt} + Alice&crumpet&left\cr + Dormouse&muffin&middle\cr + Hatter&tea&right\cr} +}% +\noindent The ^|\tabskip| \ctsref{\tabskip} in this example +tells \TeX\ to insert |2pc| of +\refterm{glue} between the columns. +The ^|\noalign| \ctsref{\noalign} commands tell \TeX\ to insert +\refterm{vertical mode} material between two rows. +In this example we've +used |\noalign| to produce some extra space between the title rows and +the data rows, and also to bring ``Table'' and ``Position'' closer together. +(You can also use |\noalign| before the first row or after the +last row.) +\eix^^|\halign| + +You can construct a vertical alignment with the ^|\valign| command +\ctsref{\valign}. A vertical alignment is organized as a series of +columns rather than as a series of rows. A vertical alignment follows +the same rules as a horizontal alignment except that the roles of rows +and columns are interchanged. For example, the vertical alignment: + +\csdisplay +{\hsize=0.6in \parindent=0pt +\valign{#\strut&#\strut&#\strut\cr + one&two&three\cr + four&five&six\cr + seven&eight&nine\cr + ten&eleven\cr}} +| +\noindent yields: +\vdisplay{% +{\hsize=0.6in \parindent=\listleftindent % Because lists and displays + % are not indented just by \parindent. +\valign{#\strut&#\strut&#\strut\cr + one&two&three\cr + four&five&six\cr + seven&eight&nine\cr + ten&eleven\cr}} +} +The ^|\strut| commands \ctsref{\strut} +in the template are necessary to get the entries in each row +to line up properly, i.e., to have a common \refterm{baseline}, +and to keep the distance between baselines uniform. +\eix\ctsidxref{cr} +\eix^^{tables} + +\endconcept + + +\concept{\anatomy} + +\texbook\ describes the way that {\TeX} processes its input in terms of \TeX's +``digestive tract''---its ``^{eyes}'', ``^{mouth}'', +``^{gullet}'', ``^{stomach}'', and ``^{intestines}''. Knowing how this +processing works can be helpful when you're trying to understand subtle +aspects of \TeX's behavior as it's digesting a document. + +\ulist + +\li Using its ``\pix^{eyes}'', \TeX\ reads \refterm{characters:character} from +^{input files} and passes them to its mouth. Since an input file +can contain ^|\input| commands \ctsref{\input}, +\TeX\ can in effect ``shift its gaze'' from one file to another. + +\li Using its ``\pix^{mouth}'', {\TeX} assembles the characters into +\refterm{tokens:token} and passes them to its gullet. +Each token is either a \refterm{control sequence} or a single +character. A control sequence always starts with an \refterm{escape +character}. Note that spaces and ends-of-line are characters in their +own right, although \TeX\ compresses a sequence of input spaces into a single +space token. See \knuth{pages~46--47} for the rules by which \TeX\ assembles +characters into tokens. +^^{tokens//assembled from characters} + +\li Using its ``\pix^{gullet}'', {\TeX} expands any macros, conditionals, and +^^{macros//expanded in \TeX's stomach} +^^{tokens//passed to \TeX's stomach} +similar constructs that it finds (see \knuth{pages~212--216}) and passes +the resulting sequence of \refterm{tokens:token} +to \TeX's stomach. Expanding one token +may yield other tokens that in turn need to be expanded. {\TeX} carries +out this expansion from left to right unless the order is modified by +a command such as |\expandafter| \ctsref{\expandafter}. +In other words, \TeX's gullet always expands the leftmost un\-ex\-panded +token that it has not yet sent to \TeX's stomach. + +\li Using its ``\pix^{stomach}'', {\TeX} processes the tokens in +groups. +Each group contains a primitive command followed by its arguments, if any. +Most of the commands are of the ``typeset this character'' variety, +so their groups consist of just one token. +Obeying the instructions given by the commands, +\TeX's stomach assembles larger and larger +units, starting with +characters and ending with pages, +and passes the pages to \TeX's intestines. +^^{pages//assembled in \TeX's stomach} +\TeX's stomach handles the tasks of \refterm{line break}ing---% +^^{line breaking} +i.e., breaking each paragraph into a sequence of lines---% +and of \refterm{page break}ing---i.e., breaking a continuous sequence of lines +and other vertical mode material +into pages. + +\li Using its ``\pix^{intestines}'', \TeX\ transforms the pages produced by its +stomach into a form intended for processing +by other programs. It then sends the transformed output to the +\dvifile. +^^{\dvifile//created by \TeX's intestines} + +\endulist + +Most of the time you can think of the processes that take place in \TeX's +eyes, mouth, gullet, stomach, and intestines +as happening one after the other. But the +truth of the matter is that commands executed in \TeX's stomach can +influence the earlier stages of digestion. For instance, when \TeX's stomach +encounters the |\input| command \ctsref{\input}, +its eyes start reading from a different +file; when \TeX's stomach encounters a |\catcode| command + \ctsref{\catcode} specifying a category code +for a character $c$, the interpretation of $c$ by \TeX's mouth is affected. +And when +\TeX's stomach encounters a \refterm{macro} definition, the expansions carried +out in \TeX's gullet are affected. + +You can understand how the processes interact by imagining that each +process eagerly gobbles up the output of its predecessor as soon as it +becomes available. For instance, once \TeX's stomach has seen +the last character of the filename in an |\input| command, \TeX's gaze +immediately shifts to the first character of the specified input file. +\endconcept + +\pagebreak +\conceptindex{arguments} +\concept argument + +^^{commands//arguments of} +An \defterm{argument} contains text that is passed to a +\refterm{command}. +The arguments of a command complete the description of what +the command is supposed to do. +The command can either be a \refterm{primitive} command or a +\refterm{macro}. + +Each primitive command ^^{primitive//command} +has its own convention about the form of its +arguments. For instance, the sequence of \refterm{tokens:token}: + +\csdisplay +\hskip 3pc plus 1em +| +consists of the command `|\hskip|' and the arguments +`|3pc plus 1em|'. But if you were to write: + +\csdisplay +\count11 3pc plus 1em +| +you'd get an entirely different effect. +\TeX\ would treat `|\count11|' as a command with argument `|3|', +followed by the ordinary text tokens `|pc plus 1em|' +(because count registers expect a number to be assigned to them)% +---probably not what +you intended. The effect of the command, by the way, would be to +assign $3$ to count register $11$ (see the discussion of ^|\count|, +\xref\count). + +Macros, on the other hand, all follow the same convention +for their arguments. +^^{macros//arguments of} +Each argument passed to a macro +corresponds to a \refterm{parameter} +^^{parameters//and arguments} +in the definition of that +macro. ^^{macros//parameters of} +A macro parameter is either ``delimited'' or ``undelimited''. +The macro definition determines the number and nature of the macro parameters +and therefore the number and nature of the macro arguments. + +The difference between a delimited argument and an undelimited argument +lies in the way that \TeX\ decides where the argument ends. +^^{delimited arguments} +^^{undelimited arguments} +\ulist +\li A delimited argument consists of the tokens +from the start of the argument up to, but not including, the +particular sequence of tokens that serves as the delimiter for that argument. +The delimiter is specified in the macro definition. Thus you +supply a delimited argument to a macro by writing the argument itself +followed by the delimiter. A delimited argument can be empty, i.e., have +no text at all in it. Any braces in a delimited argument must be paired +properly, i.e., every left brace must have a corresponding right brace +and vice versa. + +\li An undelimited argument consists of a single token or a sequence of +tokens enclosed in braces, like this: +`|{Here is {the} text.}|'. Despite appearances, the outer braces don't +form a \refterm{group}---\TeX\ uses them only to determine what the +argument is. Any inner braces, such as the ones around `|the|', must be +paired properly. If you make a mistake and put in too many right +braces, \TeX\ will complain about an unexpected right brace. \TeX\ will +also complain if you put in too many left braces, but you'll probably +get \emph{that} complaint long after the place where you intended to +end the argument (see \xref{mismatched}). +\endulist +\noindent +See \conceptcit{macro} for more information +about parameters and arguments. You'll find the precise rules pertaining +to delimited and undelimited arguments in \knuth{pages~203--204}. +\endconcept + + +\concept{ASCII} + +\defterm{\ascii} is the abbreviation of ``American Standard Code for +Information Interchange''. There are $256$ \ascii\ +^{characters}, each with its own code number, but +the meanings of only the first~$128$ have been standardized. You can +find these meanings in an \ascii\ ``code table'' such as the one on +\knuth{page~367}. Characters $32$--$126$ are ``printable characters'', +^^{printable characters} such as letters, numbers, and punctuation +marks. The remaining characters are ``^{control characters}'' that are +typically used (in the computer industry, not in \TeX) to control +input\slash output and data communications devices. For instance, +\ascii\ code $84$ corresponds to the letter `T', while \ascii\ +code~$12$ corresponds to the ``form feed'' function (interpreted by most +printers as ``start a new page''). Although the \ascii\ standard +specifies meanings for the control characters, many manufacturers of +equipment such as modems and printers have used the control characters +for purposes other than the standard ones. + +The meaning of a +character in \TeX\ is usually consistent with its meaning in standard \ascii, +and \refterm{fonts:font} that contain \ascii\ +printable characters usually have those characters in the same positions as +their \ascii\ counterparts. +But some fonts, notably those used for math, replace the \ascii\ +printable characters by other characters unrelated to the \ascii\ characters. +For instance, the Computer Modern math font +^^{Computer Modern fonts} +|cmsy10| has the math symbol +`{$\forall$}' in place of the \ascii\ digit `8'. + + +\endconcept + +\conceptindex{assignments} +\concept assignment + +An \defterm{assignment} is a construct that tells \TeX\ to assign a +value to a register, +^^{registers//assignment to} +to one of its internal +\refterm{parameters:parameter}, +^^{parameters//assignments to} +to an entry in one of its internal tables, +or to a \refterm{control sequence}. Some examples of assignments are: + +\csdisplay +\tolerance = 2000 +\advance\count12 by 17 +\lineskip = 4pt plus 2pt +\everycr = {\hskip 3pt \relax} +\catcode\`@ = 11 +\let\graf = \par +\font\myfont = cmbx12 +| + +The first assignment indicates that \TeX\ should assign the numeric value +|2000| to the numeric parameter |\tolerance|, i.e., make the value of +|\tolerance| be $2000$. The other assignments are similar. The `|=|' +and the spaces are optional, so you could also write the first +assignment more tersely as: + +\csdisplay +\tolerance2000 +| + +See \knuth{pages~276--277} for the detailed syntax of assignments. +\endconcept + + +\concept badness + +The \defterm{badness} of a line is a measure of how far the interword +spaces ^^{interword spacing} +in the line deviate from their natural values, +i.e., the values specified in the \refterm{fonts:font} used in the line. +The greater the +deviation, the greater the badness. Similarly, the badness of a page is +a measure of how far the spaces between the boxes that +make up the page deviate from their ideal values. (Ordinarily, most of these +boxes are single lines of paragraphs.) + +More precisely, the badness +is a measure of how much the \refterm{glue} associated with these spaces needs +to stretch or shrink to fill the line or page exactly. +\TeX\ computes the badness as approximately $100$ +times the cube of the ratio by which it must stretch or shrink the glue +in order to compose a line or a page of the required size. +^^{line breaks//badness for}^^{page breaks//badness for} +For example, stretching the glue by twice its stated stretch yields a ratio of +$2$ and a badness of $800$; stretching it by half its stated stretch yields +a ratio of $.5$ and a badness of $13$. +\TeX\ treats a badness greater than $10000$ as +equal to $10000$. + +\TeX\ uses the badness of a line when it's breaking a paragraph into lines +\seeconcept{line break}. It uses this information in two stages: + +\olist +\li When \TeX\ is choosing line breaks, +it will eventually accept lines whose badness is less than or equal to +the value of |\tolerance| (\xref \tolerance). If \TeX\ cannot avoid setting +a line whose badness exceeds this +value, it will set it as an underfull or overfull \refterm{hbox}. +\TeX\ will set +an overfull or underfull hbox only as a last resort, i.e., only if there's no +other way to break the paragraph into lines. +\li Assuming that all lines are tolerably bad, \TeX\ uses the badness of lines +in order to evaluate the different ways of breaking the paragraph into lines. +During this evaluation it associates ``demerits'' with each potential line. +The badness increases the number of \refterm{demerits}. +\TeX\ then +breaks the paragraph into lines in a way that minimizes the +total demerits for the paragraph. +Most +often \TeX\ arranges the paragraph in a way that minimizes the badness of the +worst line. See \knuth{pages~97--98} for the details of how \TeX\ +breaks a paragraph into lines. +\endolist + +\TeX's procedure for assembling a sequence of lines and other vertical +mode material into pages is similar to its procedure for line breaking. +However, assembling pages is +not as complicated because \TeX\ only considers one page at a time +when it looks for page breaks. +Thus the only decision it must make is where to end the current page. +In contrast, when \TeX\ is choosing line breaks it +considers several of them simultaneously. +(Most word processors choose line breaks one at a time, +and thus don't do as good a job at it as \TeX\ does.) +See \knuth{pages~111--113} for the details of how \TeX\ chooses its +page breaks. +\endconcept + + +\conceptindex{baselines} +\concept baseline + +The \defterm{baseline} of a \refterm{box} is an imaginary line that runs +across the box. +^^{boxes//baselines of} +When \TeX\ is +assembling the boxes of a \refterm{horizontal list} into a larger box, +it lines up the boxes in the list so that their baselines coincide. +As an analogy, think of writing on a pad of ruled paper. Each letter +that you write has +an implicit baseline. +In order to line up the letters horizontally, +you place them on the pad so that their baselines +agree with the light guidelines that are printed on the pad. + +A box can and often does extend below its baseline. +For instance, the letter `g' extends below the baseline of its box because +it has a descender (the bottom loop of the `g'). +\endconcept + + +\conceptindex{boxes} +\concept box + +A \defterm{box} is a rectangle of material to be typeset. A single +\refterm{character} is a box by itself, +and an entire page is also a box. +\TeX\ forms a page as a nest of boxes within boxes within boxes. The +outermost box is the page itself, the innermost boxes are mostly +single characters, and single lines are boxes that are +somewhere in the middle. + +\TeX\ carries out most of its box-building activities implicitly as it +constructs paragraphs and pages. +You can construct boxes explicitly +using a number of \TeX\ \refterm{commands}, notably +^|\hbox| \ctsref{\hbox}, +^|\vbox| \ctsref{\vbox}, and +^|\vtop| \ctsref{\vtop}. +The ^|\hbox| command +constructs a box by appending smaller boxes horizontally from left to right; +it operates on a \refterm{horizontal list} and yields +an \refterm{hbox} ^^{hboxes} (horizontal box). +^^{horizontal lists} +The ^|\vbox| and |\vtop| commands +construct a box by appending smaller boxes vertically from top to bottom; +^^{vboxes} +they operate on a \refterm{vertical list} +and yield a \refterm{vbox} ^^{vboxes} (vertical box). +^^{vertical lists} +These horizontal and vertical lists can include not just smaller boxes but +several other kinds of entities as well, e.g., \refterm{glue} and +kerns. +^^{kerns//as list items} + +A box has \refterm{height}, \refterm{depth}, and \refterm{width}, +^^{height} ^^{depth} ^^{width} +like this: +\vdisplay{\offinterlineskip\sevenrm + \halign{#&#&\kern3pt \hfil#\hfil\cr + &\hrulefill\cr + &\vrule + \vtop to .7in{\vss \hbox to .9in{\hss baseline\hss}\vskip4pt}% + \vrule + &\labelledheight{.7in}{height}\cr + % + \vbox to 0pt{ + \vss + \hbox{reference point \hbox to 15pt{\rightarrowfill}% + \hskip3pt}% + \kern-4.5pt}&{\box\refpoint}\hrulefill\cr + % + \omit + &\vrule\hfil\vrule + &\labelledheight{.4in}{depth}\cr + % + &\hrulefill\cr + % + \noalign{\vskip3pt}% + &\leftarrowfill { width }\rightarrowfill\cr +}} + +^^{baselines} +The \refterm{baseline} is like one of +the light guidelines on a pad of ruled paper. +The boxes for letters such as `g' +extend below the baseline; the boxes for letters such as `h' don't. +The height of a box is the distance that the box extends above its +baseline, while its depth is the distance that it extends below its +baseline. \bix^^{reference point} +The \minref{reference point} +of a box is the place where its baseline intersects its left edge. + +{\tighten +\TeX\ builds an hbox $H$ from a horizontal list by assuming +a reference point for $H$ and then appending the items in the list to $H$ +one by one from left to right. +Each box in the list is placed so that its baseline coincides with the +baseline of $H$\kern-2pt, +i.e., the component boxes are lined up horizontally.% +\footnote{If a box is moved up or down with ^|\raise| or +^|\lower|, \TeX\ uses its reference point before the move when +placing it.} +The height of $H$ is the +height of the tallest box in the list, and the depth of $H$ is the depth +of the deepest box in the list. +The width of $H$ is the sum of the +widths of all the items in the list. +If any of these items are \refterm{glue} and \TeX\ needs to stretch or shrink +the glue, +the width of $H$ will be larger or smaller accordingly. +See \knuth{page~77} for the~details. +\par} + +Similarly, \TeX\ builds a vbox $V$ from a vertical list by assuming a +temporary reference point for $V$ and then appending the items in the +list to $V$ one by one from top to bottom. Each box in the list is +placed so that its reference point is lined up vertically with the +reference point of \Vperiod.\footnote{If a box is moved left or right with +^|\moveleft| or ^|\moveright|, \TeX\ uses its reference point before the +move when placing it.} As each box other than the first one is added to +\Vcomma, \TeX\ puts \minref{interline glue} just above it. (This +^{interline glue} has no analogue for hboxes.) The width of $V$ is the +width of the widest box in the list, and the vertical extent (height +plus depth) of $V$ is the sum of the vertical extents of all the +items in the list. + +\bix^^|\vbox| +\bix^^|\vtop| +The difference between |\vbox| and |\vtop| is in how they partition +the vertical extent of $V$ into a height and a depth. +Choosing the reference point of $V$ determines that partition. +\ulist +\li For |\vbox|, \TeX\ places the reference point on a horizontal line +with the reference point of the last component box +or rule of \Vcomma, except +that if the last box (or rule) is followed by glue or a kern, \TeX\ places the +reference point at the very bottom of \Vperiod.% +\footnote{The depth is limited by +the parameter ^|\boxmaxdepth| \ctsref{\boxmaxdepth}.} + +\li For |\vtop|, \TeX\ places the reference point on a horizontal line +with the reference point of the first component box or rule of \Vcomma, +except that if the first box (or rule) +is preceded by glue or a kern, \TeX\ places +the reference point at the very top of \Vperiod. + +\endulist +\noindent +Roughly speaking, then, |\vbox| puts the reference point near the bottom +of the vbox and |\vtop| puts it near the top. +When you want to align a +row of vboxes so that their tops line up horizontally, +you should usually use |\vtop| rather than |\vbox|. +See \knuth{pages~78 and 80--81} for the +details of how \TeX\ builds vboxes. +\eix^^|\vbox| +\eix^^|\vtop| +\eix^^{reference point} + +You have quite a lot of freedom in constructing boxes. The typeset +material in a box can extend beyond the boundaries of the box as it does +for some letters (mostly italic or slanted ones). The component boxes +of a larger box can overlap. A box can have negative width, depth, or +height, though boxes like that are not often needed. + +You can save a box in a box \refterm{register} and retrieve it later. +Before using a box register, +^^{box registers} +you should reserve it and give it a name +with the ^|\newbox| command \ctsref{\@newbox}. See +\conceptcit{register} for more information about box +registers. +\endconcept + + +\conceptindex{category codes} +\concept {category code} + +The \defterm{category code} of a \refterm{character} determines that +character's role in \TeX. +^^{characters//category code of} +For instance, \TeX\ assigns a certain role to +letters, another to space characters, and so forth. \TeX\ attaches a +code to each character that it reads. When \TeX\ reads the +letter `|r|', for example, it ordinarily +attaches the category code $11$ (letter) +to it. For simple \TeX\ applications you won't need to worry about +category codes, but they become important when you are trying to achieve +special effects. + +Category codes apply only to characters that \TeX\ reads from input +files. Once a character has gotten past \TeX's ^{gullet} +\seeconcept{\anatomy} and been interpreted, its category code no +longer matters. A character that you produce with the ^|\char| command +\ctsref{\char} does not have a category code because |\char| +is an instruction to \TeX\ to produce a certain character in a certain +\refterm{font}. For instance, the ^{\ascii} code for `|\|' +(the usual escape character) is $92$. If +you type `|\char92 grok|', it is \emph{not} equivalent to |\grok|. +Instead it tells \TeX\ to +typeset `$c$grok', where $c$ is the character in position $92$ +of the code table for the current font. + +You can use the ^|\catcode| command \ctsref{\catcode} to reassign the +category code of any character. By changing category codes you can +change the roles of various characters. For instance, if you type +`|\catcode`\@ = 11|', the category code of the at sign (|@|) will be set +to ``letter''. You then can use `|@|' in the name of a control +sequence. + +Here is a list of the category codes as they're defined in +\refterm{\plainTeX} (see \xref{twocarets} for an explanation of +the |^^| notation), +together with the characters in each category: + +\xrdef{catcodes} +\vskip\abovedisplayskip +%k \vskip 0pt plus 2pt % to fix bad page break +{%k \interlinepenalty = 10000 +\halign{\indent\hfil\strut#&\qquad#\hfil\cr +\it Code&\it Meaning\cr +\noalign{\vskip\tinyskipamount} +0&Escape character \quad |\| ^^{escape character//category code of} + {\recat!ttidxref[\//category code of]] + \cr +1&Beginning of group \quad |{| ^^{groups} + {\recat!ttidxref[{//category code of]] + \cr +2&End of group \quad |}| + {\recat!ttidxref[}//category code of]] + \cr +3&Math shift \quad |$| ^^{math shift} + {\recat!ttidxref[$//category code of]] + \cr +4&Alignment tab \quad |&| ^^{tabs} ^^{alignments//tab character for} + \ttidxref{&//category code of} \cr +5&End of line \quad |^^M| \tequiv \ascii\ \asciichar{return} + ^^{end of line} \ttidxref{^^M//category code of}\cr +6&Macro parameter \quad |#| + ^^{macros//parameters of} + ^^{parameters//indicated by \b\tt\#\e} + \ttidxref{#//category code of} \cr +7&Superscript \quad |^| and |^^K| ^^{superscripts} + \ttidxref{^^K} + \ttidxref{^//category code of} \cr +8&Subscript \quad |_| and |^^A| ^^{subscripts} + \ttidxref{^^A} + \ttidxref{_//category code of} \cr +9&Ignored character \quad |^^@| \tequiv \ascii\ \asciichar{null} + ^^{ignored characters} \indexchar ^^@ \cr +10&Space \quad \visiblespace\ and |^^I| \tequiv \ascii\ + \asciichar{horizontal\ tab} + ^^{horizontal tab} + ^^{space characters//category code of} \indexchar ^^I + {\recat!ttidxref[ ]] + \cr +11&Letter \quad |A| \dots |Z| and |a| \dots |z| ^^{letter}\cr +12&Other character \quad (everything not listed above or below) + ^^{other characters}\cr +13&Active character \quad |~| and |^^L| \tequiv\ascii\ \asciichar{form\ feed} + ^^{active characters} ^^{form feed} \indexchar ~ \indexchar ^^L \cr +14&Comment character \quad |%| ^^{comments} + {\recat!ttidxref[%//category code of]] + \cr +15&Invalid character \quad |^^?| \tequiv \ascii\ \asciichar{delete} + ^^{invalid character} \indexchar ^^? \cr +}} +\vskip\belowdisplayskip +%k \vskip 0pt plus 2pt % to fix bad page break + +\noindent Except for categories $11$--$13$, +all the characters in a particular category produce the same effect. +\margin{Misleading material removed.} +For instance, suppose +that you type: +\csdisplay +\catcode`\[ = 1 \catcode`\] = 2 +| +Then the left and right bracket characters become +beginning-of-group and end-of-group characters equivalent to +the left and right brace characters. With these definitions `|[a b]|' +is a valid group, and so are \hbox{`|[a b}|'} and~\hbox{`|{a b]|'}. + +The characters in categories $11$ (letter) and $12$ +(other character) act as \refterm{commands:command} +that mean +``produce a \refterm{box} containing this character +typeset in the current font''. +The only distinction between letters and ``other'' characters is +that letters can appear in \refterm{control word}s but +``other'' characters~can't. + +A character in category $13$ (active) acts like a control sequence +all by itself. \TeX\ complains if it encounters an active character that +doesn't have a definition associated with it. +^^{active characters} + +If \TeX\ encounters an ^{invalid character} (category $15$) +in your input, it will complain about it. + +The `|^^K|' and `|^^A|' characters have been included in categories +$8$ (subscript) and $9$ (superscript), even though these meanings +don't follow the standard \refterm{\ascii} interpretation. +That's because some keyboards, notably some at Stanford +University where \TeX\ originated, +have down arrow and up arrow keys that generate these characters. +\ttidxref{^^A} +\ttidxref{^^K} + +There's a subtle point about the way \TeX\ assigns category codes that +can trip you up if you're not aware of it. \TeX\ sometimes needs to +look at a character twice as it does its initial scan: first to find the +end of some preceding construct, e.g., a control sequence, and later to +turn that character into a token. \TeX\ doesn't assign the category +code until its \emph{second} look at the character. For example: + +\csdisplay +\def\foo{\catcode`\$ = 11 }% Make $ be a letter. +\foo$ % Produces a `$'. +\foo$ % Undefined control sequence `foo$'. +| +\noindent +This bit of \TeX\ code produces `\$' in the typeset output. When +\TeX\ first sees the `|$|' on the second line, +it's looking for the end of a control sequence name. Since +the `|$|' isn't yet a letter, it marks the end of `|\foo|'. Next, +\TeX\ expands the `|\foo|' macro and changes the category code of `|$|' +to $11$ (letter). Then \TeX\ reads the `|$|' ``for real''. Since +`|$|' is now a letter, \TeX\ produces a box +containing the `|$|' character in the current font. +When \TeX\ sees the third line, it treats `|$|' as a letter and thus +considers it to be part of the control sequence name. +As a result it complains about an undefined control sequence |\foo$|. + +\TeX\ behaves this way even when the terminating character is an +end of line. For example, suppose that the macro |\fum| activates the +end-of-line character. Then if |\fum| appears on a line $\ell$ by +itself, \TeX\ will first interpret the end of line of $\ell$ as +the end of the |\fum| control sequence and then will \emph{reinterpret} +the end of line of $\ell$ as an active character. +\endconcept + + +\conceptindex{characters} +\concept character + +{\tighten +\TeX\ works with \defterm{characters} in two contexts: +as input characters, which it reads, and as output characters, +which it typesets. +\TeX\ transforms most input characters +into the output characters that depict them. +For example, it normally +transforms the input letter `|h|' into the letter `h' typeset in the current +font. +That is not the case, however, for an input character such as `|$|' that has a +special meaning. +\par} + +\TeX\ gets its input characters by reading them from input files (or from your +terminal) and by expanding \refterm{macros:macro}. These are the +\emph{only} ways that \TeX\ can acquire an input character. +Each input character has a code number corresponding to its position in the +\refterm{\ascii} code table. ^^{\ascii} +For instance, the letter `|T|' has \ascii\ code~$84$. + +When \TeX\ reads +a character, it attaches a \refterm{category code} +^^{category codes//attached during input} +to it. The category code affects how \TeX\ interprets the +character once it has been read in. \TeX\ determines +(and remembers) the category codes of the characters in a macro when it +reads the macro's definition. As \TeX\ reads characters with its eyes +\seeconcept{\anatomy} it does some ``filtering'', +such as condensing +sequences of spaces to a single space. See \knuth{pages~46--48} for the +details of this filtering. + +The \ascii\ +``^{control characters}'' have codes $0$--$31$ and $127$--$255$. +They either don't +show up or cause strange behavior on most terminals if you try to +display them. Nonetheless they are sometimes needed in \TeX\ input, +so \TeX\ has a special notation for them. +\xrdef{twocarets} +If you type `|^^|$c$', where $c$ is any character, you get the character +whose \ascii\ code is either $64$ greater or $64$ less than $c$'s +\ascii\ code. The largest acceptable code value using this notation +is $127$, so the notation is unambiguous. +Three particularly common instances of this +notation are `|^^M|' (the \ascii\ \asciichar{return} character), +`|^^J|' (the \ascii\ \asciichar{line\ feed} character) and `|^^I|' +(the \ascii\ \asciichar{horizontal\ tab} character). +\ttidxref{^^M}\ttidxref{^^J}\ttidxref{^^I} + +{\tighten +\TeX\ also has another notation for indicating \ascii\ code values +that works for all character codes from $0$ to $255$. +\xrdef{hexchars} +If you type `|^^|$xy$', where $x$ +and $y$ are any of the ^{hexadecimal digit}s `|0123456789abcdef|', +you get the single character with the specified code. +(Lowercase letters are required here.) +\TeX\ opts for the ``hexadecimal digits'' +interpretation whenever it has a choice, so you must not follow a character +like `|^^a|' with a lowercase hexadecimal digit---if you do, you'll get the +wrong interpretation. +If you need to use this +notation you'll find it handy to have a table of \ascii\ codes. +\par} + +An output character is a character to be typeset. +A command for producing an output character has the meaning +``Produce a \refterm{box} containing +character number $n$ from the current \refterm{font}'', +where $n$ is determined by the command. +\TeX\ produces your typeset document by combining such boxes with +other typographical \hbox{elements} and arranging them +on the page. + +An input character whose category code is $11$ (^{letter}) or $12$ (other) +^^{other characters} +acts as a command to produce the corresponding output character. In +addition you can get \TeX\ to produce character $n$ by issuing the +command `|\char |$n$' \ctsref{\char}, ^^|\char| where $n$ is a +\refterm{number} between $0$ and $255$. The commands `|h|', +|\char`h|, and |\char104| all have the same effect. ($104$ is the +\ascii\ code for `h'.) + +\endconcept + + +\concept class + +The \defterm{class} of a \refterm{character} specifies that character's +role in math formulas. The class of a character is encoded in its +\refterm{mathcode}. ^^{mathcodes//class encoded in} For example, the +equals sign `|=|' is in class $3$ (Relation). \TeX\ uses its knowledge +of character classes to decide how much space to put between different +components of a math formula. \margin{clarifying material added} For +example, here's a math formula shown first as \TeX\ normally prints it +and then with the class of each character randomly changed: +$$\strut a+(b-a)=a \qquad + \mathopen{a}\mathord{+}\mathrel{(}\mathclose{b}\mathclose{-} + \mathop{a}\mathopen{)}\mathord{=}\mathopen{a}$$ + +See \xrefpg\mathord\ of this book for a list of the classes and +\knuth{page~154} for their meanings. + +\endconcept + + +\conceptindex{commands} +\concept command + +A \defterm{command} instructs \TeX\ to carry out a certain action. +Every \refterm{token} that reaches \TeX's stomach \seeconcept{\anatomy} +acts as a command, except for those that are parts of arguments to +other commands (see below). +^^{tokens//as commands} +A command can be invoked by a +\refterm{control sequence}, by an \refterm{active character}, or by an +ordinary character. It might seem odd that \TeX\ treats an ordinary +character as a command, but in fact that's what it does: +when \TeX\ sees +an ordinary character +it constructs a \refterm{box} containing that character typeset in +the current font. + +A command can have arguments. +The arguments of a command are single tokens or +groups of tokens that complete the description of what +the command is supposed to do. +For example, the command `|\vskip 1in|' tells \TeX\ to skip +$1$ inch vertically. It has an argument `|1in|', +which consists of three tokens. +The description of what |\vskip| is supposed to do would be incomplete +without specifying how far it is supposed to skip. +The tokens in the arguments to a command are not themselves considered +to be commands. + +Some examples of different kinds of \TeX\ commands are: +\ulist\compact +\li Ordinary characters, such as `|W|', which instructs \TeX\ +to produce a box containing a typeset `W' +\li Font-setting commands, +such as |\bf|, which begins boldface type +\li Accents, such as |\`|, which produces a grave accent as in `\`e' +\li Special symbols and ligatures, such as |\P| (\P) and |\ae| (\ae) +\li Parameters, such as |\parskip|, the amount of glue that +\TeX\ puts between paragraphs +\li Math symbols, such as |\alpha| ($\alpha$) and |\in| ($\in$) +\li Math operators, such as |\over|, which produces a fraction +\endulist +\endconcept + + +\conceptindex{conditional tests} +\concept {conditional test} + +A \defterm{conditional test} is a command that tests whether or not a certain +condition is true and +causes \TeX\ either to expand or to skip some text, depending on +the outcome. +The general form of a conditional test is either: +\display{ +{\tt \\if}$\alpha$\<true text>{\tt \\else}\<false text>{\tt \\fi}} +^^|\else|^^|\fi| +\noindent or:\hfil\ +\display{ +{\tt \\if}$\alpha$\<true text>{\tt \\fi}} +\noindent where $\alpha$ specifies the particular test. +For example, |\ifvmode| tests the condition that \TeX\ +is currently in a \refterm{vertical mode}. +If the condition is true, \TeX\ expands \<true text>. +If the condition is false, \TeX\ expands \<false text> (if it's present). +Conditional tests are interpreted in \TeX's gullet +\seeconcept{\anatomy}, so any expandable \minref{token}s in +the interpreted text are expanded after the test has been resolved. +The +various conditional tests are explained in \headcit{Conditional tests}% +{conds}. + +\endconcept + + +\conceptindex{control sequences} +\concept{control sequence} + +A \defterm{control sequence} is a name for a \TeX\ \refterm{command}. +A control sequence always starts with an ^{escape character}, usually +a backslash (|\|). +\indexchar \ +A control sequence takes one of two forms: + +\ulist + +\li A \refterm{control word} is a control sequence consisting of an +\refterm{escape character} followed by one or more letters. +^^{control words} +The control +word ends when \TeX\ sees a nonletter. For instance, when \TeX\ reads +`\hbox{|\hfill!visiblespace,!visiblespace!.the|}', it sees six +\refterm{tokens:token}: +the control sequence `|\hfill|', comma, space, `|t|', `|h|', `|e|'. The +space after `|\hfill|' ends the control sequence and +is absorbed by \TeX\ when it scans the control sequence. +(For the text `|\hfill,!visiblespace!.the|', on the other hand, +the comma both ends the control sequence and counts as a character in its +own right.) + +\li A \refterm{control symbol} +^^{control symbols} +is a control sequence consisting of an +^{escape character} followed by any character other than a letter---% +even a space or an end of line. +A control symbol is self-delimited, i.e., \TeX\ knows where it ends without +having to look at what character comes after it. +The character after a control symbol is never absorbed by +the control symbol. +\endulist +\noindent See \xrefpg{spaces} for more information about spaces after control +sequences. + +\TeX\ provides a great many predefined control sequences. The +\refterm{primitive} control sequences are built into the \TeX\ computer +program and thus are available in all forms of \TeX. +^^{primitive//control sequence} +Other +predefined control sequences are provided by \refterm{\plainTeX}, the +form of \TeX\ described in this book. + +You can augment the predefined control sequences with ones of your own, +using commands such as ^|\def| and ^|\let| to define them. +\chapterref{eplain} of this book contains a +collection of control sequence definitions that you may find +useful. In addition, your computing facility may +be able to provide a collection of +locally developed \TeX\ macros. +\endconcept + + +\conceptindex{control symbols} +\concept{control symbol} + +A \defterm{control symbol} is a \refterm{control sequence} that consists +of an \refterm{escape character} followed by any character other than a +letter---even a space or end of line. +^^{escape character} +\endconcept + + +\conceptindex{control words} +\concept{control word} + +A \defterm{control word} is a \refterm{control sequence} that consists +of an \refterm{escape character} followed by one or more +letters.\footnote{A ``letter'' here has the strict meaning of a +character with category code $11$.} +\TeX\ +ignores any spaces or ends-of-line that follow a control word, except to +note that they end the control word. +^^{escape character} +\endconcept + + + +\concept {decimal constant} + +See \conceptcit{number}. +\endconcept + +\conceptindex{delimiters} +\concept delimiter + +A \defterm{delimiter} is a character +that is intended to be used as +a visible boundary of a math formula. +The essential property of a delimiter is that \TeX\ can adjust +its size according +to the vertical size (\refterm{height} plus \refterm{depth}) +of the subformula. +However, \TeX\ performs the adjustment only if the delimiter +appears in a ``delimiter context'', namely, as an argument to +one of the commands ^|\left|, +^|\right|, +|\over!-with!-delims|, +|\atop!-with!-delims|, +or |\above!-with!-delims| +^^|\overwithdelims| +^^|\atopwithdelims| +^^|abovewithdelims| +\margin{Footnote deleted} +(see \pp\xrefn{\overwithdelims},~\xrefn{\left}). +The delimiter contexts also include any \refterm{argument} +to a \refterm{macro} that uses the argument in a delimiter context. + +For example, the left and right +parentheses are delimiters. +If you use ^{parentheses} in a delimiter context +around a formula, \TeX\ makes the parentheses big +enough to enclose the \refterm{box} that contains the formula (as long as the +\refterm{fonts:font} you're using have big enough parentheses). +For example: +\csdisplay +$$ \left( a \over b \right) $$ +| +yields: +\centereddisplays $$\left (a \over b \right ) $$ +Here \TeX\ has made the parentheses big enough to accommodate the fraction. +But if you write, instead: +\csdisplay +$$({a \over b})$$ +| +you'll get: +$$({a \over b})$$ +Since the parentheses aren't in a delimiter context, +they are \emph{not} enlarged. + +Delimiters come in pairs: +an opening delimiter at the left of the subformula +and a closing delimiter at its right. +You can explicitly choose a larger height for a +delimiter with the commands ^|\bigl|, ^|\bigr|, and their +relatives \ctsref{\bigl}.\footnote +{\PlainTeX\ defines the various |\big| commands by using |\left| and |\right| +to provide a delimiter context. It sets the size by +constructing an empty formula with the desired height.} +For instance, in order to get the +displayed formula: +$$\bigl(f(x) - x \bigr) \bigl(f(y) - y \bigr)$$ + +\noindent in which the outer parentheses are a little bigger than the inner +ones, you should write: + +\csdisplay +$$\bigl( f(x) - x \bigr) \bigl( f(y) - y \bigr)$$ +| + +The $22$ \plainTeX\ delimiters, shown at their normal size, are: +\display{% +$( \>) \>[ \>] \>\{ \>\} +\>\lfloor \>\rfloor \>\lceil \>\rceil +\>\langle \>\rangle \>/ \>\backslash +\>\vert \>\Vert +\>\uparrow \>\downarrow \>\updownarrow +\>\Uparrow \>\Downarrow \>\Updownarrow$} +^^|)| ^^|)| ^^|[| ^^|]| ^^|\lbrack| ^^|\rbrack| +^^|\{| ^^|\}| ^^|\lbrace| ^^|\rbrace| +^^|\lfloor| ^^|\rfloor| ^^|\lceil| ^^|\rceil| +^^|\langle| ^^|\rangle| ^^|/| ^^|\backslash| +^^|\vert| ^^|\Vert| +^^|\uparrow| ^^|\downarrow| ^^|\updownarrow| +^^|\Uparrow| ^^|\Downarrow| ^^|\Updownarrow| +\noindent +Here they are at the largest size provided explicitly by \plainTeX\ + (the |\Biggl|, |\Biggr|, etc., versions): +\nobreak\vskip .5\abovedisplayskip +\display{% +$\Biggl( \>\Biggr) \>\Biggl[ \>\Biggr] +\>\Biggl\{ \>\Biggr\} \>\Biggl\lfloor \>\Biggr\rfloor +\>\Biggl\lceil \>\Biggr\rceil +\>\Biggl\langle \>\Biggr\rangle +\>\Biggm/ \>\Biggm\backslash +\>\Biggm\vert \>\Biggm\Vert +\>\Biggm\uparrow \>\Biggm\downarrow \>\Biggm\updownarrow +\>\Biggm\Uparrow \>\Biggm\Downarrow \>\Biggm\Updownarrow$} +\vskip .5\belowdisplayskip +\noindent +The delimiters (except for `|(|', `|)|', and +`|/|') +are among the symbols listed on +pages~\xrefn{\lbrace}--\xrefn{\Uparrow}. +They are listed in one place on \knuth{page~146}. + +A delimiter can belong to any class. +^^{class//of a delimiter} +For a delimiter that you enlarge with +|\bigl|, |\bigr|, etc., the class is determined by the command: +``opener'' for |l|-commands, ``closer'' for |r|-commands, +``relation'' for |m|-commands, and ``ordinary symbol'' for |g|-commands, +e.g., |\Big|. + +You can obtain a delimiter in two different ways: +\olist +\li You can make a character be a delimiter by assigning it a +nonnegative delimiter code +\bix^^{delimiter codes} +(see below) with the ^|\delcode| command (\xref\delcode). +Thereafter the character acts as a delimiter whenever you use it in a +delimiter context.\footnote{% +It's possible to use a character with a nonnegative delimiter code in +a context where it isn't a delimiter. In this case \TeX\ doesn't perform the +search; instead it just uses the character in the ordinary way +(see \knuth{page~156}).} +\li You can produce a delimiter explicitly with the ^|\delimiter| command +(\xref\delimiter), in analogy to the way that you can produce an ordinary +character with the |\char| command or a math character with the |\mathchar| +command. +The |\delimiter| command uses the same delimiter codes that are used in a +|\delcode| table entry, but with an extra digit in front to indicate a +class. +It's rare to use |\delimiter| outside of a macro definition. +\endolist + +A delimiter code tells +\TeX\ how to search for an appropriate output character to represent +a delimiter. +The rules for this search are rather complicated +(see \knuth{pages~156 and 442}). +A complete understanding of these rules requires knowing +about the organization of font ^{metrics file}s, a topic that is not just beyond +the scope of this book but beyond the scope of \texbook\ as well. + +In essence the search works like this. The delimiter code specifies a +``small'' output character and a ``large'' output character by +providing a \refterm{font} position and a font \refterm{family} for each +(see \xref\delcode). +Using this information, \TeX\ can find (or construct) +larger and larger versions of the delimiter. \TeX\ first tries +different sizes (from small to large) +of the ``small'' character in the ``small'' font +and then +different sizes (also from small to large) +of the ``large'' character in the ``large'' font, +seeking one whose height plus depth is sufficiently large. +If none of the characters it finds are large +enough, it uses the largest one that it finds. +It's possible that +the small character, the large character, or both have been left unspecified +(indicated by a zero in the appropriate part of the delimiter code). +If only one character +has been specified, \TeX\ uses that one. If neither has been specified, +it replaces the delimiter by a space of width ^|\nulldelimiterspace|. +\eix^^{delimiter codes} + +\endconcept + + +\concept demerits + +\TeX\ uses \refterm{demerits} as a measure of how undesirable a line is +when it's breaking a paragraph into lines \seeconcept{line break}. +^^{line breaks//demerits for} +The demerits of a line are affected both by the \refterm{badness} of the line +and by \refterm{penalties:penalty} associated with the line. +^^{badness} +\TeX's goal in choosing a particular arrangement of lines is to minimize the +total demerits for the paragraph, which it computes by adding up the demerits +for the individual lines. +See \knuth{pages~97--98} for the details of how \TeX\ +breaks a paragraph into lines. +\TeX\ does not use demerits when it's choosing page breaks; instead, it uses +a similar measure known as the ``cost'' of a particular page break. +\endconcept + +\concept depth + +^^{boxes//depth of} +The \defterm{depth} of a \refterm{box} is the distance that the box extends +below its \refterm{baseline}. +\endconcept + + +\conceptindex{dimensions} +\concept dimension + +A \defterm{dimension} specifies a distance, that is, a linear measure of +space. You use dimensions to specify sizes of things, such as the length +of a line. Printers in English-speaking countries traditionally measure +distance in points and picas, while printers in continental Europe +traditionally measure distance in did\^ot points and ciceros. You can +use these units or others, such as inches, that may be more +familiar to you. The font-independent +^{units of measure} that \TeX\ understands are: + +\xrdef{dimdefs} +\nobreak\vskip\abovedisplayskip +\halign{\indent\hfil\tt #\qquad&#\hfil\cr +pt&^{point} (72.27 points = 1 inch)\cr +pc&^{pica} (1 pica = 12 points)\cr +bp&big point (72 big points = 1 inch)\cr +in&^{inch}\cr +cm&^{centimeter} (2.54 centimeters = 1 inch)\cr +mm&^{millimeter} (10 millimeters = 1 centimeter)\cr +dd&^{\didotpt} (1157 {\didotpt}s = 1238 points)\cr +cc&^{cicero} (1 cicero = 12 {\didotpt}s)\cr +sp&^{scaled point} (65536 scaled points = 1 point)\cr +} +\vskip\belowdisplayskip + +Two additional units of measure are associated with every font: `^|ex|', +a vertical measure usually about the height of the letter `x' +in the font, and `^|em|', a +horizontal measure usually equal to the point size of the font and +about the width of the letter `M' in the font. Finally, +\TeX\ provides three ``infinite'' units of measure: `^|fil|', `^|fill|', and +`^|filll|', in increasing order of~strength. + +A dimension is written as a ^{factor}, i.e, a multiplier, +followed by a unit of measure. +^^{units of measure} +The factor can be either a whole \refterm{number} or +a \refterm{decimal constant} containing a decimal point +or decimal comma. +The factor can be preceded by a plus or minus sign, so a dimension +can be positive or negative. +^^{dimensions//negative} +The unit of measure must be there, even if the number is +zero. Spaces between the number and the unit of measure are permitted +but not required. You'll find a precise definition of a +dimension on \knuth{page~270}. Here are some examples of dimensions: + +\csdisplay +5.9in 0pt -2,5 pc 2fil +| +The last of these represents a first-order infinite distance. + +An infinite distance outweighs any finite distance or any weaker infinite +distance. If you add |10in| to |.001fil|, you get |.001fil|; if you add +|2fil| to |-1fill| you get |-1fill|; and so forth. +\TeX\ accepts infinite distances +only when you are specifying the \refterm{stretch} and \refterm{shrink} +of \refterm{glue}. + +\TeX\ multiplies all dimensions in your document by a +\refterm{magnification} factor $f/1000$, +where $f$ is the value of the ^|\mag| parameter. +^^{magnification} +Since the default value of +|\mag| is $1000$, the normal case is that your document is +typeset just as specified. You can specify a dimension as it will be +measured in the final document independent of magnification by putting +`|true|' in front of the unit. For instance, `|\kern 8 true pt|' +produces a kern of $8$ points whatever the magnification. +\endconcept + + +\concept {display math} + +The term \defterm{display math} refers to a math formula that \TeX\ +places on a line by itself with extra space above and below +so as to set it off from the surrounding text. +A display math formula is enclosed by `|$$|'s. +\ttidxref{$$} +\TeX\ reads display math in display math \refterm{mode}. +\endconcept + + +\concept{escape character} + +An \defterm{escape character} introduces a control sequence. The escape +character in \refterm{\plainTeX} is the backslash (|\|). +\indexchar \ +You can change the escape character from $c_1$ to $c_2$ +by reassigning the category codes of $c_1$ and $c_2$ +with the ^|\catcode| command \ctsref{\catcode}. +You can also define additional escape characters similarly. +If you want to typeset material containing literal escape characters, you must +either +(a) define a control sequence that stands for the printed escape character or +(b) temporarily +disable the escape character by changing its category code, using the +method shown on \xrefpg{verbatim}. The definition: + +\csdisplay +\def\\{$\backslash$} +| +is one way of creating a control sequence that stands for `$\backslash$' +(a backslash typeset in a math font). + +You can use the ^|\escapechar| parameter \ctsref{\escapechar} to specify +how the escape character is represented in synthesized control sequences, +e.g., those created by |\string| and |\message|. + + +\endconcept + + +\concept family + +A \defterm{family} is a group of three related \refterm{fonts:font} used +when \TeX\ is in \refterm{math mode}. +^^{fonts//families of} +Outside of math mode, families +have no effect. The three fonts in a family are used for normal symbols +(\refterm{text size}), subscripts and superscripts (\refterm{script +size}), and sub-subscripts, super-superscripts, etc.\ +(\refterm{scriptscript size}). +^^{text size} +^^{script size} +^^{scriptscript size} +For example, the numeral `|2|' set in +these three fonts would give you `$2$', `$\scriptstyle 2$', and +`$\scriptscriptstyle 2$' (in \plainTeX). +Ordinarily you would set up the +three fonts in a family as different point sizes of the same typeface, +but nothing prevents you from using different typefaces for the three +fonts as well or using the same font twice in a family. + +{\tighten +\TeX\ provides for up to sixteen families, numbered $0$--$15$. For +example, family $0$ in \refterm{\plainTeX} consists of $10$-point roman +for text, $7$-point roman for script, and $5$-point roman for +scriptscript. +^^{\plainTeX//font families in} +\PlainTeX\ also defines family $1$ to consist of math +italic fonts and reserves families $2$ and $3$ for ^{special symbols} and +^{math extensions} respectively.\footnote{Families $2$ and $3$ are special +in that their font metric files must include parameters for math +spacing.} If you need to define a family for yourself, you should use +the ^|\newfam| command \ctsref{\@newfam} to get the number of a family that +isn't in use, and the ^|\textfont|, ^|\scriptfont|, +and ^|\scriptscriptfont| commands \ctsref{\scriptscriptfont} +to assign fonts to that family. +\par} + +\endconcept + + +\conceptindex{files} +\concept file + +A \defterm{file} is a stream of information that \TeX\ interprets or +creates. Files are managed by the ^{operating system} that supervises your +\TeX\ run. \TeX\ deals with files in four different contexts: +\olist +\li A ``^{source file}'' is one that \TeX\ reads with its ``eyes'' +\seeconcept{\anatomy} and interprets according to its ordinary rules. +Your primary input file---the one you specify after `|**|' or +on the command line when +you invoke \TeX---is a source file, and so is any file that you call for +with an ^|\input| command \ctsref \input. + +\li A ``^{result file}'' is one that contains the results of +running \TeX. A \TeX\ run creates two result files: the +\dvifile\ and the log file. +^^{\dvifile//as a result file} +^^{log file//as a result file} +The \dvifile\ contains the information needed to print your document; +the +log file contains a record of what happened during the run, including any +error messages that \TeX\ generated. +If your primary source file is named +|screed.tex|, your \dvifile\ and log file will be named |screed.dvi| +and |screed.log|.\footnote{This is the usual convention, but +particular implementations of \TeX\ are free to change it.} + +\li To read from a file with the ^|\read| +command \ctsref{\read} you need to associate the file with an input stream. +^^{input streams//reading with \b\tt\\read\e} +You can have up to $16$ input streams active +at once, numbered $0$--$15$. +The |\read| command reads a single line and makes it the value of a +designated \refterm{control sequence}, so reading with +|\read| is very different from reading with ^|\input| (which brings in an +entire file). +\TeX\ takes any input stream number not between +$0$ and $15$ to refer to the terminal, +so `|\read16|', say, reads the next line that you type at the terminal. + +\li To write to a file with the |\write| +command \ctsref \write\ you need to associate the file +with an output stream. +^^|\write//output stream for| +^^{output streams} +You can have up to $16$ output streams active +at once, numbered $0$--$15$. +Input and output streams are independent. +Anything sent to an output stream with a negative number goes to the log +file; anything sent to an output stream with a number greater than $15$ +goes both to the log file and to the terminal. +Thus `|\write16|', say, writes a line on the terminal and also sends +that line +to the log~file. + +\endolist + +You must open a stream file before you can use it. +An input stream file is opened with an ^|\openin| +command \ctsref \openin\ and an output stream file is opened with an +^|\openout| command \ctsref\openout. +For tidiness +you should close a stream file when you're done with it, although +\TeX\ will do that at the end of the run if you don't. +The two commands for closing a stream file are ^|\closein| \ctsref\closein\ +and ^|\closeout| \ctsref\closeout. +An advantage of closing a stream when +you're done with it is that you can then reuse the stream for a different file. +Doing this can be essential when you're reading a long sequence of files. + +Although you can assign numbers yourself to input and output streams, +it's better to do it with the ^|\newread| and +^|\newwrite| \ctsref{\@newwrite} commands. +You can have more than one stream associated with a particular file, +but you'll get (probably undiagnosed) garbage unless all of the streams +are input streams. Associating more than one stream with an input file +can be useful when you want to use the same input file for two different +purposes. + +\TeX\ ordinarily defers the actions of opening, writing to, or closing +an output stream until it ships out a page with ^|\shipout| +(see \knuth{page~227} +for the details). This behavior applies even to messages written to the +terminal with |\write|. But you can get \TeX\ to perform an action +on an output stream immediately by preceding the action command with +^|\immediate| \ctsref\immediate. For example: +\csdisplay +\immediate\write16{Do not pass GO!! Do not collect $200!!} +| +\endconcept + + +\conceptindex{file names} +\concept {file name} + +A \defterm{file name} names a +\refterm{file} that is known to the ^{operating system} +that in turn +supervises your \TeX\ run. The syntax of a file name does \emph{not} +follow the usual rules of \TeX\ syntax, and in fact it is different +in different implementations of \TeX. +In particular, most \TeX\ implementations consider a file name to be +terminated by a blank or an end of line. Thus \TeX\ is likely to +misinterpret `|{\input chapter2}|' +by taking the right brace as part of the file name. +As a general rule, you should follow a file name by a blank or the +end of the line as in `|{\input chapter2!visiblespace}|'. + +\endconcept + + +\eject +\conceptindex{fonts} +\concept font + +A \defterm{font} in \TeX\ is a collection of up to $256$ output +characters, usually having the same typeface design, style (roman, +italic, bold, condensed, etc.), +and point size.\footnote{\PlainTeX\ uses a special +font for constructing ^{math symbols} in which the characters have +different sizes. Other special fonts are often useful for applications +such as typesetting ^{logos}.} The ^{Computer Modern fonts} that +generally come with \TeX\ have only $128$ characters. The colophon on +the last page of this book describes the typefaces that we used to set +this book. + +For instance, here is the alphabet in the Palatino Roman $10$ point font: +^^{Palatino fonts} +\vskip\abovedisplayskip{\narrower\tenpal +\noindent ABCDEFGHIJKLMNOPQRSTUVWXYZ\hfil\break +abcdefghijklmnopqrstuvwxyz\par +}\vskip\belowdisplayskip +\noindent +And here it is in the Computer Modern Bold Extended $12$ +point font: +^^{Computer Modern fonts} +\vskip\abovedisplayskip{\narrower\font\twelvebf=cmbx12\twelvebf +\noindent ABCDEFGHIJKLMNOPQRSTUVWXYZ\hfil\break +abcdefghijklmnopqrstuvwxyz\par +}\vskip\belowdisplayskip +The characters in a font are numbered. +The numbering usually agrees with the ^{\ascii} numbering +for those characters that exist in the \ascii\ character set. +The code table for each font indicates what the character +with code $n$ looks like in that font. +Some fonts, such as the ones used for mathematical symbols, have no +letters at all in them. You can produce a \refterm{box} containing the +character numbered $n$, typeset in the current font, by writing `|\char |$n$' + \ctsref{\char}. + +In order to use a font in your document, +you must first name it with a control sequence and load it. Thereafter you +can select it by typing +that control sequence whenever you want to use it. +\PlainTeX\ provides a number of fonts that are already named and~loaded. + +You name and load a font as a single operation, using a +command such as `|\font\twelvebf=cmbx12|'. Here `|\twelvebf|' is the +control sequence that you use to name the font +and `|cmbx12|' identifies the font metrics file +|cmbx12.tfm| +in your computer's file system. +You then can start to use the font by typing +`|\twelvebf|'. After that, the font will be in effect until +either (a)~you select another font or (b)~you terminate the +\refterm{group}, if any, in which you started the +font. For example, the input: + +\csdisplay +{\twelvebf white rabbits like carrots} +| +will cause the |cmbx12| font to be in effect just for the +text `|white rabbits like carrots|'. + +You can use \TeX\ with fonts other than +Computer Modern (look at the example on \xrefpg{palatino} and +at the page headers). +The files for such fonts need to be +installed in your computer's file system in a place where \TeX\ can find +them. \TeX\ and its companion programs generally need two files for each font: +one to give its metrics (|cmbx12.tfm|, +^^{\tfmfile} +for example) and another to +give the shape of the characters (|cmbx12.pk|, for example). +\TeX\ itself uses only the metrics +file. Another program, the device driver, +^^{device drivers} +converts the \dvifile\ +^^{\dvifile//converted by driver} +produced by \TeX\ to a form that your printer +or other output device can handle. The driver +uses the shape file (if it exists). + +The font metrics file contains the information that \TeX\ needs in +order to allocate space for each typeset character. +Thus it includes the size of each character, the ligatures and +kerns that pertain to sequences of adjacent characters, and so on. +What the metrics file +\emph{doesn't} include is any information about the shapes +of the characters in the font. + +{\tighten +The shape (pixel) file \xrdef{shape} +^^{pixel file}^^{shape file} +may be in any of several +formats. The extension part of the name (the part after the dot) +tells the driver which format the shape file is in. For example, +|cmbx12.pk| ^^{\pkfile} might be the shape file for font |cmbx12| in +packed format, while |cmbx12.gf| ^^{\gffile} might be the shape file +for font |cmbx12| in generic font format. +A shape file may not be needed for a font that's resident in your +output device. +\par} + +\endconcept + + +\conceptindex{footers} +\concept footer + +A \defterm{footer} is material that \TeX\ puts at the bottom of every page, +below the text of that page. +The default footer in \plainTeX\ is a centered page number. +Ordinarily a footer consists of a single line, which you can set by +assigning a token list to ^|\footline| \ctsref\footline. +See \xrefpg{bighead} for a method of producing multiline footers. + +\endconcept + +%k \vskip 0pt plus 2pt % to solve page break problem + +\concept {format file} + +{\tighten +A \defterm{format file} is a file that contains an image of +\TeX's memory in a form in which it can be reloaded quickly. +A format file can be created with the ^|\dump| command \ctsref\dump. +The image contains a complete record +of the definitions (of \refterm{fonts:font}, \refterm{macros:macro}, etc.) +that were present when the dump took place. +By using ^|virtex|, a special ``virgin'' form of \TeX, +you can then reload the format file at high speed and +continue in the same state that \TeX\ was in at the time of the dump. +The advantage of a format file over an ordinary input +file containing the same information is that \TeX\ can load it much +faster. +\par} + +Format files can only be created by ^|initex|, another special +form of \TeX\ intended just for that purpose. +Neither |virtex| nor |initex| has any +facilities other than the primitives built into the +\TeX\ program itself. + +A ^{preloaded} form of \TeX\ is one that has a format file already +loaded and is ready to accept user input. +The form of \TeX\ that's called |tex| +often has the \plainTeX\ definitions preloaded. +(\PlainTeX\ is ordinarily available in two other forms as well: +as a format file and as a \TeX\ source file. +In some environments, |tex| is equivalent to calling |virtex| +and then loading |plain|.) +Creating preloaded forms of \TeX\ requires a special program; +it cannot be done using only the facilities of \TeX\ itself. + +\endconcept + +\concept global + +A \defterm{global} definition is effective +until the end of the +document or until it is overridden by another definition, +even when it occurs within a \refterm{group}. +Thus a global definition is unaffected by group boundaries. +You can make any definition global by prefixing it with the |\global| +command \ctsref{\global} unless ^|\globaldefs| \ctsref\globaldefs{} +is negative. +^^|\global| + +There's a special way of making a \refterm{macro} definition global. +^^{macros//global} +Normally you define a macro using either the |\def| command +or the |\edef| command \ctsref{\edef}. +^^|\edef//making global| +^^|\def//making global| +If you use |\gdef| or |\xdef| +instead of |\def| and |\edef| respectively, the macro definition will +be global. That is, `^|\gdef|' is equivalent to `|\global\def|' and +`^|\xdef|' is equivalent to `|\global\edef|'. +\endconcept + + +\concept glue + +\bix^^{stretch} +\bix^^{shrink} +\defterm{Glue} is blank space that can stretch or shrink. +Glue gives \TeX\ the flexibility that it needs in order to produce +handsome +documents. Glue comes in two flavors: horizontal glue and vertical glue. +Horizontal glue occurs within \refterm{horizontal lists:horizontal list}, +while vertical glue occurs within \refterm{vertical lists:vertical list}. +^^{horizontal lists} +^^{vertical lists} +You +can produce a glue item either implicitly, e.g., with an interword space, or +explicitly, e.g., with the ^|\hskip| command. +^^{spaces//interword} +\TeX\ itself produces many glue +items as it typesets your document. +We'll just describe horizontal glue---vertical glue is analogous. + +When \TeX\ assembles a list of boxes and glue into a larger +unit, +^^{boxes//glue with} +it adjusts the size of the glue to meet the space requirements of the +larger unit. For instance, \TeX\ ensures that the ^{right margin} of a page +is uniform by adjusting the horizontal glue within lines. +Similarly, it ensures that different pages have the +same ^{bottom margin} +by adjusting the glue between blocks of text such as paragraphs and +math displays +(where the change is least likely to be conspicuous). + +A glue item has a natural space---the size it ``wants to be''. Glue +also has two other attributes: its \refterm{stretch} and its +\refterm{shrink}. You can produce a specific amount of horizontal glue +with the ^|\hskip| \refterm{command} \ctsref{\hskip}. The command +|\hskip 6pt plus 2pt minus 3pt| +produces a horizontal glue item whose natural +size is $6$ points, whose stretch is $2$ points, and whose shrink is +$3$ points. Similarly, you can produce a specific amount of vertical +glue with the ^|\vskip| command \ctsref{\vskip}. + +The best way to understand what stretch and shrink are about +is to see an example of glue at work. +Suppose you're constructing an \refterm{hbox} from three boxes and two glue +items, as in this picture: +\gluepicture + 29 {\picbox 4 \gluebox 6 4 1 6 \picbox 5 \gluebox 10 8 3 10 \picbox 4 } +\noindent +The units of measurement here could be points, millimeters, or anything else. +If the hbox is allowed to assume its natural width, then each glue item in the +box also assumes its natural width. The total width of the hbox is then the +sum of the widths of its parts, namely, $29$ units. + +Next, suppose that the hbox is required to be wider than $29$ units, say +$35$ units. This +could happen, for example, if the hbox is required to occupy an entire +line and the line width is $35$ units. +Since the boxes can't change their width, +\TeX\ produces the necessary extra space by making the glue items wider. +The picture now looks like this: +\gluepicture + 35 {\picbox 4 \gluebox 6 4 2 8 \picbox 5 \gluebox 10 8 6 14 \picbox 4 } +The glue items don't become wider equally; they became wider in proportion to +their stretch. Since the second glue item +has twice as much stretch as the first one, +it gets wider by four units while the first glue item gets wider by only +two units. +Glue can be stretched as far as necessary, although \TeX\ is +somewhat reluctant to +stretch it beyond the amount of stretch given in its definition. + +Finally, suppose that the hbox is required to be narrower than $29$ units, say +$25$ units. In this case \TeX\ makes the glue items narrower. +The picture looks like this: +\gluepicture + 25 {\picbox 4 \gluebox 6 4 2 5 \picbox 5 \gluebox 10 8 6 7 \picbox 4 } +The glue items become narrower in proportion to their shrink. +The first glue item becomes narrower by one unit, while the second glue item +becomes narrower by three units. Glue cannot shrink by a distance +less than the amount of shrink +given in its definition even though the distance it can stretch is +unlimited. In this important sense the shrink and +the stretch behave differently. + +A good rule of thumb for glue is to set the natural size to the amount +of space that looks best, the stretch to the largest amount of space +that \TeX\ can add before the document starts to look bad, and the +shrink to the largest amount of space that \TeX\ can take away before +the document starts to look bad. You may need to set the values by +experiment. + +You can produce glue that is infinitely stretchable by specifying +its stretch in units of `^|fil|', `^|fill|', or `^|filll|'. Glue measured in +`|fill|' is infinitely more stretchable than glue measured in `|fil|', and +glue measured in `|filll|' is infinitely more stretchable than glue measured +in `|fill|'. You should rarely have any need for `|filll|' glue. Glue whose +stretch is |2fil| has twice as much stretch as glue whose stretch is |1fil|, +and similarly for the other kinds of infinitely stretchable glue. + +When \TeX\ is +apportioning extra space among glue items, the infinitely stretchable +ones, if there +are any, get all of it. Infinitely stretchable glue is particularly useful for +setting text flush left, flush right, or centered: + +\ulist\compact +\li To set text ^{flush left}, put infinitely stretchable +horizontal glue to the right of it. +That glue will consume all the +extra space that's available on the line. +You can use the ^|\leftline| command \ctsref{\leftline} +or the |\raggedright| command \ctsref{\raggedright} to do~this. +\li To set text ^{flush right}, put infinitely +stretchable horizontal glue to the left of it. +As before, that glue will consume all the extra space on the line. +You can use the ^|\rightline| command \ctsref{\rightline} to do~this. +\li To set ^{centered text}, put identical infinitely stretchable +horizontal glue items +on both sides of it. +These two glue items will divide all the extra space on the line +equally between them. +You can use the ^|\centerline| command \ctsref{\centerline} to do~this. +\endulist + +You can also specify infinitely shrinkable glue +^^{glue//infinitely shrinkable} +in a similar way. Infinitely shrinkable glue can act as negative space. +Note that |fil|, etc., can be used only +to specify the stretch and shrink of glue---they can't be used to specify +its natural size. +\eix^^{stretch} +\eix^^{shrink} +\endconcept + + +\conceptindex{groups} +\concept group + +A \defterm{group} is a part of your manuscript that \TeX\ treats as a unit. +You indicate a group by enclosing it in the braces +`|{|' and `|}|' (or any other characters with the appropriate +\refterm{category codes}). +^^|{//starting a group| +^^|}//ending a group| + +The most important property of a group is that any nonglobal +definition or assignment that you make inside a group disappears when +the group ends. For instance, if you write: + +\csdisplay +Please don't pour {\it any} more tea into my hat. +| +the |\it| \refterm{control sequence} causes \TeX\ to set the word +`|any|' in italic type but does not affect the rest of the text. +As another example, if you use the |\hsize| parameter +\ctsref{\hsize} to change the line length within a group, the line length +reverts to its previous value once \TeX\ has gotten past the group. + +Groups are also useful as a way of controlling spacing. For instance, if you +write: + +\csdisplay +\TeX for the Impatient and the Outpatient too. +| +\noindent +you'll get: +\display{% +\TeX for the Impatient and the Outpatient too. +} +\noindent +since the control sequence |\TeX| (which produces the \TeX\ +logo) absorbs the following space. +What you probably want is: +\display{% +{\TeX} for the Impatient and the Outpatient too. +} +\noindent +One way to get it is to enclose `|\TeX|' in a group: +\csdisplay +{\TeX} for the Impatient and the Outpatient too. +| +The right brace prevents the control sequence from absorbing the space. +\endconcept + + +\conceptindex{hboxes} +\concept hbox + +An \defterm{hbox} (horizontal box) is a \refterm{box} that \TeX\ +constructs by placing the items of a \refterm{horizontal list} one after +another, left to right. +^^{horizontal lists//hboxes formed from} +An hbox, taken as a unit, is neither +inherently horizontal nor inherently vertical, i.e., it can appear as an +item of either a horizontal list or a \refterm{vertical list}. You can +construct an hbox with the |\hbox| command \ctsref{\hbox}. +\endconcept + + +\conceptindex{headers} +\concept header + +A \defterm{header} is material that \TeX\ puts at the top of every page, +above the text of that page. +The header for a simple report +might consist of the title on the left side of +the page and the text ``Page $n$'' on the right side of the page. +Ordinarily a header consists of a single line, which you can set by +assigning a token list to ^|\headline| \ctsref\headline. +The default \plainTeX\ header is blank. +It's possible to produce multiline headers too; see \xrefpg{bighead} for +how to do it. +\endconcept + + +\concept height + +The \defterm{height} of a \refterm{box} is the distance that the box +extends above its \refterm{baseline}. +^^{boxes//height of} +\endconcept + + +\conceptindex{horizontal lists} +\concept{horizontal list} + +A \defterm{horizontal list} is a list of items +that \TeX\ has produced while it is +in one of its \refterm{horizontal modes:horizontal mode}, i.e., assembling +either a paragraph or an hbox. See ``horizontal mode'' below. +\endconcept + + +\concept {horizontal mode} + +^^{hboxes//horizontal mode for} +When \TeX\ is assembling a paragraph or an \refterm{hbox}, it is in one +of two \defterm{horizontal modes}: ^{ordinary horizontal +mode} for assembling paragraphs and ^{restricted horizontal mode} for +assembling hboxes. Whenever \TeX\ is in a horizontal mode its stomach +\seeconcept{\anatomy} is constructing a \refterm{horizontal +list} of items (boxes, glue, penalties, etc.). +\TeX\ typesets the items in the list +one after another, left to right. + +A horizontal list can't contain any +items produced by inherently vertical commands, e.g., |\vskip|. +^^{horizontal lists//can't contain vertical commands} + +\ulist +\li If \TeX\ is assembling a horizontal list in ordinary horizontal mode and +encounters an inherently vertical command, \TeX\ ends the paragraph and +enters \refterm{vertical mode}. + +\li If \TeX\ is assembling a horizontal list in restricted horizontal +mode and encounters an inherently vertical command, it complains. +\endulist Two commands that you might at first think are inherently +horizontal are in fact inherently vertical: |\halign| \ctsref{\halign} +and |\hrule| \ctsref{\hrule}. +^^|\hrule//inherently vertical| +^^|\halign//inherently vertical| +See \knuth{page~286} for a list +of the inherently vertical commands. + +{\tighten +You should be aware of a subtle but important property of restricted +horizontal mode: \emph{you can't enter ordinary horizontal mode +when you're in restricted horizontal mode}. What this means in practice is that +when \TeX\ is assembling an hbox it +won't handle paragraph-like text, i.e., text for which it does +\refterm{line breaking}. You can get +around this restriction by enclosing the paragraph-like text in a +\refterm{vbox} within the hbox. The same method works if you want to +put, say, a horizontal \refterm{alignment} inside an~hbox. +}% end scope of tighten + +\endconcept + + +\concept{hyphenation} + +\TeX\ automatically hyphenates words as it is processing your document. +\TeX\ is not eager to insert hyphens, preferring instead to find good +line breaks by adjusting the spacing between words and moving words +from one line to another. +\TeX\ is clever enough to understand +hyphens that are already in words. + +You can control \TeX's hyphenation in several ways: +\ulist +\li You can tell \TeX\ to +allow a hyphen in a particular place by inserting a +discretionary hyphen +^^{discretionary hyphens} +with the ^|\-| command \ctsref{\@minus}. +\li You can tell \TeX\ how to +hyphenate particular words throughout your document with the ^|\hyphenation| +command \ctsref{\hyphenation}. +\li You can enclose a word in an \refterm{hbox}, thus preventing \TeX\ +from hyphenating it. +\li You can set the value of penalties such as |\hyphenpenalty| +\ctsref\hyphenpenalty. +\endulist +\noindent If a word contains an explicit or discretionary hyphen, +\TeX\ will never break it elsewhere. +\endconcept + + +\conceptindex{input streams} +\concept {input stream} + +See \conceptcit{file}. +\endconcept + +\conceptindex{insertions} +\concept insertion + +\looseness = -1 +An \defterm{insertion} is a vertical list containing material +to be inserted into +a page when \TeX\ has finished building that page.\footnote +{\tighten +\TeX\ itself doesn't +insert the material---it just makes the material available to +the output routine, which is then responsible for transferring +it to the composed page. +^^{output routine//insertions, treatment of} +The only immediate effect of the ^|\insert| command +\ctsref{\insert} is to change \TeX's page break calculations so that it +will leave room on the page for the inserted material. Later, when +\TeX\ actually breaks the page, it divides the inserted material into +two groups: the material that fits on the current page and the material +that doesn't. +^^{page breaks//insertions at} +The material that fits on the page is placed into box registers, +one per insertion, +and the material that doesn't fit is carried over to the next page. +This procedure allows \TeX\ to do such +things as distributing parts of a long footnote over several consecutive +pages.} Examples of such insertions are footnotes and figures. The +\refterm{\plainTeX} commands for +creating insertions are ^|\footnote|, ^|\topinsert|, |\mid!-insert|, +^^|\midinsert| +and ^|\pageinsert|, as well as the primitive ^|\insert| command +itself (\pp\xrefn\footnote--\xrefn{endofinsert}). +\TeX's mechanism for handling insertions is rather complicated; +see \knuth{pages~122--125} for the details. +\endconcept + + +\concept {interline glue} + +\defterm{Interline glue} is the glue that \TeX\ inserts in front of every +\refterm{box} in a \refterm{vertical list} except for the first one. +The interline glue is ordinarily specified so as to maintain a constant +distance between the baselines of the boxes. +Its value is jointly determined by the ^|\baselineskip|, +^|\lineskip|, and ^|\lineskiplimit| parameters \ctsref{\baselineskip}. +\endconcept + + +\conceptindex{items} +\concept item + +The term \defterm{item} is often used to refer to a component of +a horizontal, vertical, or math list, i.e., a list of items that +\TeX\ is building while it is in a horizontal, vertical, or math mode. +\endconcept + + + +\conceptindex{justification} +\concept {justified text} + +\defterm{Justified text} is text that has been typeset so that both +margins are even. Unjustified text, on the other hand, has been typeset +with ``ragged'' margins on one or both sides. +Documents typed on old-fashioned typewriters almost always have +ragged right margins. +Although documents produced by \TeX\ are +justified by default, you can if you wish produce documents (or +sequences of lines) that have ^{ragged right}---or ^{ragged left}---margins. +You can also get \TeX\ to center a sequence of lines, thus making both +margins ragged. +^^{centered text} +You can use the +^|\leftskip|, ^|\rightskip|, and ^|\raggedright| commands +(\pp \xrefn{\leftskip},~\xrefn{\raggedright}) for these purposes. + +When \TeX\ is producing justified text, it usually +needs to stretch or shrink the glue within each line to make the margins +come out even. When \TeX\ is producing unjustified text, on the other +hand, it usually leaves the glue within each line at its natural width. +Many typographers prefer unjustified text because its interword +spacing is more uniform. +\endconcept + + +\conceptindex{kerns} +\concept kern + +^^{spacing//adjusting with kerns} +A \defterm{kern} indicates a change to the normal spacing between +the items of a vertical or horizontal list. +A kern can be either positive or negative. By +putting a positive kern between two items, you push them further apart +by the amount of the kern. By putting a negative kern between two +items, you bring them closer together by the amount of the kern. For +instance, this text: +\csdisplay +11\quad 1\kern1pt 1\quad 1\kern-.75pt 1 +| +produces letter pairs that look like this: +\display{11\quad 1\kern1pt 1\quad 1\kern-.75pt 1} +You can use kerns in vertical mode to adjust the space between +particular pairs of lines. + +A kern of size $d$ is very similar to a \refterm{glue} item that has +size $d$ and no stretch or shrink. Both the kern and the glue insert or +remove space between neighboring items. The essential difference is +that \TeX\ considers two boxes with only kerns between them to be tied +together. That is, \TeX\ won't break a line or a page at a kern unless +the kern is immediately followed by glue. Bear this difference in mind +when you're deciding whether to use a kern or a glue item for a +particular purpose. + +\TeX\ automatically inserts kerns between particular pairs of adjacent +letters, thus adjusting the space between those letters and enhancing +the appearance of your typeset document. +For instance, the Computer Modern $10$-point roman font contains a kern +for the pair `To' that brings the left edge of the `o' under the +`T'. Without the kern, you'd get \hbox{``T{o}p''} rather than ``Top''---% +the difference is slight but noticeable. +The metrics file +(^{\tfmfile}) +for each \refterm{font} specifies the placement and size of the +kerns that \TeX\ automatically inserts when it is setting text in that~font. +\margin{paragraph deleted to save space; most of the material was +already in this section.} + +\endconcept + + +\concept leaders + +You can use \defterm{leaders} to fill a space with copies of +a pattern, e.g., +to put repeated dots between a title and a page number in a table of contents. +A leader is a single copy of the pattern. +The specification of leaders contains three pieces of information: +\olist\compact +\li what a single leader is +\li how much space needs to be filled +\li how the copies of the pattern should be arranged within the space +\endolist + +\bix^^|\leaders| +\bix^^|\cleaders| +\bix^^|\xleaders| +{\tighten +\TeX\ has three commands for specifying leaders: +|\leaders|, \hbox{|\cleaders|}, and |\xleaders| (\xref\leaders). The +\refterm{argument} of each command specifies the leader. +The command must be followed by \refterm{glue}; the size of the glue +specifies how much space is to be filled. The choice of command determines +how the leaders are arranged within the space.} + +Here's an example showing how |\leaders| works: +\csdisplay +\def\dotting{\leaders\hbox to 1em{\hfil.\hfil}\hfil} +\line{The Political Process\dotting 18} +\line{Bail Bonds\dotting 26} +| +Here we've put the leaders and their associated glue into a \refterm{macro} +definition so that we can conveniently +use them in two places. This input produces: + +\vdisplay{\advance\hsize by -\parindent +\def\dotting{\leaders\hbox to 1em{\hfil{.}\hfil}\hfill}% +\line{The Political Process{\dotting}18} +\line{Bail Bonds{\dotting}26} +} + +The \refterm{hbox} following |\leaders| specifies +the leader, namely, an hbox 1\em\ wide containing a dot +at its center. +The space is filled with copies of this box, +effectively filling it +with dots whose centers are 1\em{} apart. +The following |\hfil| (the one at the +end of the macro definition) is glue that +specifies the space to be filled. +In this case it's whatever space is needed to fill out the line. +By choosing |\leaders| rather than |\cleaders| or |\xleaders| we've insured +that the dots on different lines line up with each other. + +In general, the space to be filled acts as a window +on the repeated copies of the leader. +\TeX\ inserts as many copies as possible, but some space is +usually left over---either because of where the leaders fall +within the window or because +the width of the window isn't an exact multiple of the width of the +leader. +The difference among the three commands is in how they arrange the leaders +within the window and how they distribute any leftover space: + +\ulist +\li For |\leaders|, \TeX\ first produces a row of copies of the leader. +It then aligns the start of this row with the left end of the innermost +box $B$ that is to contain the result of the |\leaders| command. +In the two-line example above, $B$ is a box produced by |\line|. +Those leaders that fit entirely in the window are placed into $B$, +and the leftover space at the left and right ends is left empty. +The picture is like this: +\vdisplay{% +\def\dotting{\leaders\hbox to 1em{\hfil{.}\hfil}\hfill}% +\def\pp{The Political Process} +\line{\dotting} +\line{\hphantom\pp\hfil$\Downarrow$\hfil\hphantom{18}} +\vskip 4pt +\setbox0 = \hbox{\pp} +\setbox1 = \hbox{18} +\dimen0 = \hsize \advance\dimen0 by -\wd0 \advance \dimen0 by -\wd1 +\advance\dimen0 by -0.8pt +\hbadness=10000 +\line{\pp + \vrule\vbox{\hrule width \dimen0\vskip 2pt + \hbox to \dimen0{\hfil window\strut\hfil} + \vskip 2pt\hrule width \dimen0}% + \vrule 18} +\line{\hphantom\pp\hfil$\Downarrow$\hfil\hphantom{18}} +\vskip 2pt +\line{\pp{\dotting}18} +} +\vskip\medskipamount +{\tighten +\noindent +This procedure ensures that in the two-line example on the previous page, +the dots in the two lines +are vertically aligned (since the \refterm{reference points:reference point} +of the hboxes produced by |\line| are vertically aligned). +\par} + +\li For |\cleaders|, \TeX\ centers the leaders within the window +by dividing the leftover space between the two ends of the window. +The leftover space is always less than the width of a single leader. + +\li For |\xleaders|, \TeX\ distributes the +leftover space evenly within the window. +In other words, if the leftover space is $w$ and the +leader is repeated $n$ times, +\TeX\ puts space of width $w/(n+1)$ between adjacent leaders and +at the two ends of the leaders. +The effect is usually to spread out the leaders a little bit. +The leftover space for |\xleaders|, like that for |\cleaders|, +is always less than the width of a single leader. +\endulist + +So far we've assumed that the leaders consist of hboxes arranged +horizontally. Two variations are possible: +\olist +\li You can use a +rule instead of an hbox for the leader. +\TeX\ makes the rule as wide as necessary to extend +across the glue (and the three commands are equivalent). +\li You +can produce vertical leaders that run down the page by including them in +a \refterm{vertical list} rather than a \refterm{horizontal list}. In +this case you need vertical glue following the leaders. +\endolist +\noindent +See \knuth{pages~223--225} for the precise rules that \TeX\ uses +in typesetting leaders. +\eix^^|\leaders| +\eix^^|\cleaders| +\eix^^|\xleaders| +\endconcept + + +\conceptindex{ligatures} +\concept ligature + +A \defterm{ligature} is a single character that replaces a +particular sequence of adjacent characters in a typeset document. +For example, the word `|office|' is typeset as \hbox{``office''}, +not \hbox{``of{f}ice''}, by high-quality typesetting systems. +Knowledge of ligatures is built into the +\refterm{fonts:font} that you use, so there's nothing explicit you need do +in order to get \TeX\ to produce them. (You could defeat the ligature +in ``office'', as we did just above, by writing `|of{f}ice|' in your input.) +\TeX\ is also capable of using its ligature mechanism to typeset the +first or last letter of a word differently than the same letter as it would +appear in the middle of a word. +You can defeat this effect (if you ever encounter it) by using the +^|\noboundary| command (\xref\noboundary). + +Sometimes you may need a ligature from a European language. +^^{European languages} +\TeX\ won't +produce these automatically unless you're using a font designed for that +language. A number of these ligatures, e.g., `\AE', are available as +commands (see ``Letters and ligatures for European alphabets'', +\xref{fornlets}). +\endconcept + + +\conceptindex{line breaks} +\concept{line break} + +A \defterm{line break} is a place in your document where \TeX\ ends +a line as it typesets a paragraph. +When \TeX\ processes your document, it collects the contents of each +paragraph in a \refterm{horizontal list}. +When it has collected an entire paragraph, +it analyzes the list to find what it considers to be the best possible +line breaks. \TeX\ associates ``^{demerits}'' with various symptoms of +unattractive line breaks---lines that have too much or too little +space between words, consecutive lines that end in hyphens, and so forth. It +then chooses the line breaks so as to minimize the total number of demerits. +See \knuth{pages~96--101} for a full description of \TeX's line-breaking rules. + +You can control \TeX's choice of line breaks in several ways: +\ulist + +\li You can insert a \refterm{penalty} (\xref{hpenalty}) somewhere in +the horizontal list that \TeX\ builds as it forms a paragraph. +^^{penalties//in horizontal lists} +A +positive penalty discourages \TeX\ from breaking the line there, while a +negative penalty---a bonus, in other words---encourages \TeX\ to break +the line there. A penalty of $10000$ or more prevents a line break, +while a penalty of $-10000$ or less forces a line break. You can get +the same effects with the ^|\break| and +^|\nobreak| commands (\pp\xrefn{hbreak},~\xrefn{hnobreak}). + +\li You can tell \TeX\ to allow a hyphen in a particular place by +inserting a discretionary hyphen +^^{discretionary hyphens} +with the |\-| command \ctsref{\@minus}, or +otherwise control how \TeX\ hyphenates your document \seeconcept +{hyphenation}. +^^|\-//in line breaking| + +\li You can tell \TeX\ to allow a line break after a ^{solidus} (/) between +two words by inserting ^|\slash| \ctsref{\slash} +between them, e.g., `|fur!-longs\slash fortnight|'. + +\li You can tell \TeX\ not to break a line between two particular words by +inserting a ^{tie} (|~|) between those words. +^^|~//in line breaking| + +\li You can adjust the penalties associated with line breaking by +assigning different values to \TeX's line-breaking +\refterm{parameters:parameter}. + +\li You can enclose a word or sequence of words in an \refterm{hbox}, +thus preventing \TeX\ from breaking the line anywhere within the hbox. +^^{hboxes//controlling line breaks} +\endulist + +It's useful to know the places where \TeX\ can break a line: +\ulist +\li at glue, provided that: +\olist +\li the item preceding the glue is one of the following: +a box, a discretionary item (e.g., a discretionary hyphen), +the end of a math formula, +a whatsit, +or vertical material produced by |\mark| or |\vadjust| +or |\insert| +\li the glue is not part of a math formula +\endolist +\noindent +When \TeX\ breaks a line at glue, it makes the break at the left edge +of the glue space and forgets about the rest of the glue. +\li at a kern that's immediately followed by glue, +provided that this kern isn't within a math formula +\li at the end of a math formula that's immediately followed by glue +\li at a penalty, even one within a math formula +\li at a discretionary break +\endulist +When \TeX\ breaks a line, it discards any +sequence of glue, kerns, and penalty items that follows the break point. +If such a sequence is followed by the beginning of a math formula, it +also discards any kern produced by the beginning of the formula. +\endconcept + + +\conceptindex{lists} +\concept list + +A \defterm{list} is a sequence of \refterm{items:item} +(\refterm{boxes:box}, \refterm{glue}, \refterm{kerns:kern}, etc.) +that comprise the contents of an hbox, a vbox, +or a math formula. +See \conceptcit{horizontal list}, \conceptcit{vertical list}. + +\endconcept + + +\concept {log file} + +See \conceptcit{file}. +\endconcept + + +\conceptindex{macros} +\concept macro + +{% Use a brace here so that definitions of explanatory macros remain local. +% The closing brace is at the end of the concept. +A \defterm{macro} is a definition that gives a name to a pattern of +\TeX\ input text.\footnote{More precisely, the definition gives a name +to a sequence of tokens.} The name can be either a \refterm{control +sequence} or an \refterm{active character}. The pattern is called the +``replacement text''. The primary command for defining macros is the +|\def| control sequence. + +\def\arctheta{\cos \theta + i \sin \theta} +As a simple example, suppose that you have a document in which +the sequence `$\cos \theta + i \sin \theta$' occurs many times. +Instead of writing it out each time, you can define a macro for it: +\csdisplay +\def\arctheta{\cos \theta + i \sin \theta} +| +Now whenever you need this sequence, you can just ``call'' the macro +by writing `|\arctheta|' +and you'll get it. For example, `|$e^{\arctheta}$|' will give you +`$e^{\arctheta}$'. + +\bix^^{macros//parameters of} +But the real power of macros lies in the fact that a macro can have +parameters. When you call a macro that has parameters, you provide +arguments that are substituted for those parameters. For example, suppose +you write: +\pix\indexchar # +\def\arc#1{\cos #1 + i \sin #1} +\csdisplay +\def\arc#1{\cos #1 + i \sin #1} +| + +The notation |#1| \xrdef{@msharp} indicates the first parameter +of the macro, which in this case has only one parameter. You now can +produce a similar form, such as `$\arc{2t}$', with the macro call `|\arc +{2t}|'. + +More generally, a macro can have up to nine parameters, which you +indicate as `|#1|', `|#2|', etc\null. in the macro definition. \TeX\ +provides two kinds of parameters: delimited parameters and undelimited +parameters. Briefly, a delimited parameter has an \refterm{argument} +that's delimited, or ended, by a +specified sequence of tokens (the delimiter), while an undelimited +parameter has an argument that doesn't need a delimiter to end it. +First we'll explain how macros work when they have only undelimited +parameters, and then we'll explain how they work when +they have delimited parameters. + +^^{parameters//undelimited} +If a macro has only undelimited parameters, those parameters must appear +one after another in the macro definition \emph{with nothing between +them or between the last parameter and the left brace in front of the +replacement text}. +A call on such a macro consists of the macro name followed by +the arguments of the call, one for each parameter. Each argument is +either: + +\ulist \compact +\li a single \refterm{token} other than a left or right brace, or + +\li a sequence of tokens enclosed between a left brace and +a matching right brace.\footnote{The +argument can have nested pairs of braces within it, and each of these +pairs can indicate either a \refterm{group} or a further macro +argument.} +\endulist + +When \TeX\ encounters a macro, it expands the macro in its gullet +\seeconcept{\anatomy} +by substituting each argument for the corresponding +parameter in the replacement text. The resulting text may contain other macro +calls. When \TeX\ encounters such an embedded macro call, it expands +that call immediately without looking at what follows the +call.\footnote{In computer science terminology, the expansion is ``depth +first'' rather than ``breadth first''. Note that you can modify the +order of expansion with commands such as |\expandafter|.} When \TeX's +gullet gets to a \refterm{primitive} \refterm{command} that +cannot be further expanded, \TeX\ passes that command to \TeX's stomach. +The order of expansion is sometimes critical, so in order to help +you understand it we'll give you an example of \TeX\ at work. + +Suppose you provide \TeX\ with the following input: +\csdisplay +\def\a#1#2{\b#2#1\kern 2pt #1} +\def\b{bb} +\def\c{\char49 cc} +\def\d{dd} +\a\c{e\d} % Call on \a. +| +Then the argument corresponding to |#1| is |\c|, +and the argument corresponding to |#2| is |e\d|. +\TeX\ expands the macro call in the following steps: + +{\vskip\abovedisplayskip\obeylines % ugly +|\b e\d\c\kern 2pt \c| +|bbe\d\c\kern 2pt \c| +|\d\c\kern 2pt \c|\quad(`|b|', `|b|', `|e|' sent to stomach) +|dd\c\kern 2pt \c| +|\c\kern 2pt \c|\quad(`|d|', `|d|' sent to stomach) +|\char49 cc\kern 2pt \c| +|\c|\quad(`|\char|', `|4|', `|9|', `|c|', `|c|', % +`|\kern|', `|2|', `|p|', `|t|' sent to stomach) +|\char49 cc| +(`|\char49|', `|c|', `|c|' sent to stomach) +\vskip\belowdisplayskip} + +\noindent Note that the letters `|b|', `|c|', `|d|', and `|e|' and the +control sequences `|\kern|' and `|\char|' are all primitive +commands that cannot be expanded further. + +\bix^^{parameters//delimited} +A macro can also have ``delimited parameters'', which can be mixed with +the undelimited ones in any combination. The idea of a delimited +parameter is that \TeX\ finds the corresponding argument by looking for +a certain sequence of tokens that marks the end of the argument---the +delimiter. That is, when \TeX\ is looking for such an argument, it +takes the argument to be all the tokens from \TeX's current position up +to but not including the delimiter. + +You indicate a delimited parameter by writing `|#|$n$' ($n$ +must be between $0$ +and $9$) followed by one or more tokens that act as the delimiter. The +delimiter extends up to the next `|#|' or `|{|'---which makes sense +since `|#|' starts another parameter and `|{|' starts the replacement text. + +The delimiter can't be `|#|' or `|{|', so you can tell a delimited +parameter from an undelimited one by looking at what comes after it. + +If the character after the parameter is `|#|' or `|{|', you've got an +undelimited parameter; otherwise you've got a delimited one. Note +the difference in arguments for the two kinds of parameters---an +undelimited parameter is matched either by a single token or by +a sequence of tokens enclosed in braces, while a +delimited parameter is matched by any number of tokens, even zero. + +An example of a macro that uses two delimited parameters is: +\def\diet#1 #2.{On #1 we eat #2!} +\csdisplay +\def\diet#1 #2.{On #1 we eat #2!!} +| +Here the first parameter is delimited by a single space +and the second parameter is delimited by a period. If you write: +\csdisplay +\diet Tuesday turnips. +| +you'll get the text ``\diet Tuesday turnips.''. +But if the delimiting tokens are enclosed in a group, \TeX\ doesn't consider +them as delimiting. So if you write: +\csdisplay +\diet {Sunday mornings} pancakes. +| +you'll get the text `\diet {Sunday mornings} pancakes.' +even though there's a space between `|Sunday|' and `|morning|'. +When you use a space as a delimiter, +an end-of-line character ordinarily also delimits the argument +since \TeX\ converts the end-of-line to a space before the macro +mechanism ever sees it. +\eix^^{parameters//delimited} +\eix^^{macros//parameters of} + +Once in a while you might need to define a macro that has `|#|' as a +meaningful character within it. +You're most likely to need to do this when you're defining a macro +that in turn defines a second macro. +What then do you do about +the parameters of the second macro to avoid getting \TeX\ confused? +The answer is that you write +two `|#|'s for every one that you want +when the first macro is expanded. For example, suppose you +write the macro definition: +\def\first#1{\def\second##1{#1/##1}} +\csdisplay +\def\first#1{\def\second##1{#1/##1}} +| +Then the call `|\first{One}|' defines `|\second|' as: +\csdisplay +\def\second#1{One/#1} +| +and the subsequent call `|\second{Two}|' produces the text +\def\second#1{One/#1}% +`\second {Two}'. + +A number of commands provide additional ways of defining macros +(see pp.~\xrefn{mac1}--\xrefn{mac2}). +For the complete rules pertaining to macros, see \knuth{Chapter~20}. +}% close brace at the start of the `macro' concept. +\endconcept + + +\concept magnification + +When \TeX\ typesets your document, it multiplies all dimensions +by a +\refterm{magnification} factor $f/1000$, +where $f$ is the value of the ^|\mag| parameter \ctsref\mag. +Since the default value of |\mag| is $1000$, the normal case is that +your document is typeset just as specified. +Increasing the magnification is often useful when you're typesetting a document +that will later be photoreduced. + +You can also apply magnification to a single \refterm{font} so as to get +a smaller or larger version of that font than its ``^{design size}''. You +need to provide the device driver with a ^{shape file} +\seeconcept{font} for +each magnification of a font that you're using---% +unless the fonts are built into your printer and your device driver +knows about them. +When you're defining a font with +the |\font| command \ctsref{\font}, you can specify a magnification with +the word `|scaled|'. For example: + +\csdisplay +\font\largerbold = cmbx10 scaled 2000 +| +defines `|\largerbold|' as a font that is +twice as big as |cmbx10| (Computer Modern +Bold Extended $10$-point) and has the character shapes +uniformly enlarged by a factor of~$2$. + +Many computer centers find it convenient to provide fonts scaled by a ratio +of $1.2$, corresponding to magnification values of $1200$, $1440$, etc. +\TeX\ has special names for these values: +^^|\magstep| +`|\magstep1|' for $1200$, +`|\magstep2|' for $1440$, and so forth up to `|\magstep5|'. The special +value `^|\magstephalf|' corresponds to magnification by $\sqrt{1.2}$, which +is visually halfway between `|\magstep0|' (no magnification) and +`|\magstep1|'. For example: + +\csdisplay +\font\bigbold = cmbx10 scaled \magstephalf +| + +You can specify a \refterm{dimension} as it will be +measured in the final document independent of magnification by putting +`^|true|' in front of the unit. For instance, `|\kern 8 true pt|' +produces a kern of $8$ points whatever the magnification. + +\endconcept + +\concept margins + +The \refterm{margins} +of a page define a rectangle that normally +contains the printed matter on the page. +You can get \TeX\ to print material outside of this rectangle, +but only by taking some explicit action that moves the material there. +\TeX\ considers headers and footers to lie outside the margins. + +The rectangle is defined in terms of its upper-left corner, its width, and +its depth. The location of the upper-left corner is defined by +the ^|\hoffset| +and ^|\voffset| parameters +\ctsref\voffset. The default is to place that corner one inch from the top +and one inch from the left side of the page, corresponding to a value of +zero for both |\hoffset| and |\voffset|.% +\footnote{This seems to us to be an odd convention. +It would have been more natural to have the $(0,0)$ +point for |\hoffset| and |\voffset| be at the upper-left corner of the +paper and to have set their default values to one inch.} +The width of the rectangle is given by ^|\hsize| and the depth by ^|\vsize|. + +The implications of these conventions are: +\ulist\compact +\li The left margin is given by |\hoffset|\tplus|1in|. +\li The right margin is given by the width of the paper minus + |\hoffset|\tplus|1in|\tplus|\hsize|. +\li The top margin is given by |\voffset|\tplus|1in|. +\li The bottom margin is given by the length of the paper minus + |\voff!-set|\tplus|1in|\tplus|\vsize|. +\endulist +From this information you can see what parameters you need to +change in order to change the margins. + +Any changes that you make to |\hoffset|, |\voffset|, or |\vsize| become +effective the next time \TeX\ starts a page. In other words, if you change +them within a page, the change will affect only the \emph{following} pages. +If you change |\hsize|, the change will become effective immediately. +\endconcept + + +\conceptindex{marks} +\concept mark + +A \defterm{mark} is an item that you can insert into a +horizontal, vertical, or math list and later recover from within your +output routine. Marks are useful for purposes such as +keeping track of topics to appear in page headers. +Each mark has a list of tokens---the ``^{mark text}''---associated with it. +The ^|\mark| command \ctsref{\mark} expects such a token list as its argument, +and appends an item containing that token list (after +expansion) to whatever list \TeX\ is +currently building. The ^|\topmark|, ^|\firstmark|, and ^|\botmark| commands +\ctsref{\topmark} can be used to retrieve various marks on a page. +These commands are most often used in page headers and footers. +^^{footers//marks used in} +^^{headers//marks used in} + +\margin{This example of {\tt\\mark} replaces the previous explanatory +paragraph.} +Here is a simplified example. +Suppose you define a section heading macro as follows: +\csdisplay +\def\section#1{\medskip{\bf#1}\smallskip\mark{#1}} +% #1 is the name of the section +| +^^|\mark| +This macro, when called, will produce a section heading in boldface and +will also record the name of the section as a mark. +You can now define the header for each printed page +as follows: +\csdisplay +\headline = {\ifodd\pageno \hfil\botmark\quad\folio + \else \folio\quad\firstmark\hfil \fi} +| +Each even (left-hand) page will now have the page number followed by the +name of the first section on that page, while each odd (right-hand) page +will have the page number followed by the name of the last section on +that page. Special cases, e.g., no sections starting on a page, will +generally come out correctly because of how ^|\firstmark| +and ^|\botmark| work. + +When you split a page using the |\vsplit| command \ctsref{\vsplit} you can +retrieve the mark texts of the first and last marks of the split-off +portion with the ^|\splitfirstmark| and ^|\splitbotmark| commands +\ctsref{\splitfirstmark}. + +See \knuth{pages~258--260} for a more precise explanation of how +to create and retrieve marks. +\endconcept + + +\concept math mode + +{\tighten +A \defterm{math mode} is a \refterm{mode} that \TeX\ is in when it is +building a math formula. \TeX\ has two different math modes: ^{text +math mode} for building a formula to be embedded within a line of text, +and ^{display math mode} for building a formula to appear on a line by +itself. You indicate text math mode by enclosing the formula in +|$|'s, and display math mode by enclosing the formula in +|$$|'s. +%\TeX\ will accept most \refterm{commands:command} in +%math mode. If it encounters a command in math mode that doesn't make +%sense in a formula, it will complain. +An important property of both +math modes is that \emph{input spaces don't count}. See +\knuth{pages~290--293} for details on how \TeX\ responds to different +commands in math mode. +\par} + +\endconcept + + +\conceptindex{mathcodes} +\concept mathcode + +A \defterm{mathcode} is a number that \TeX\ uses to identify and +describe a math character, +^^{math characters//described by mathcodes} +i.e., a character that has a +particular role in a math formula. A mathcode conveys three pieces of +information about a character: its \refterm{font} position, its +\refterm{family}, and its \refterm{class}. +Each of the $256$ possible +input characters has a mathcode, which is defined by the \TeX\ program +but can be changed. + +^^{family//as part of mathcode} +\TeX\ has sixteen families of fonts, numbered $0$--$15$. Each +family contains three fonts: one for \refterm{text size}, one for +\refterm{script size}, and one for \refterm{scriptscript size}. \TeX\ +chooses the size of a particular character, and therefore its font, +according to the context. The class of a character specifies its role +in a formula (see \knuth{page~154}). For example, the equals sign `|=|' +is in class $3$ (Relation). \TeX\ uses its knowledge of character +classes when it is deciding how much space to put between different +components of a math formula. + +The best way to understand what mathcodes are all about is to see how +\TeX\ uses them. So we'll show you what \TeX\ does with a +character token $t$ of \refterm{category code}~11 or~12 in a math +formula: + +\olist\compact +\li It looks up the character's mathcode. +\li It determines a family $f$ from the mathcode. +\li It determines the size $s$ from the context. +\li It selects a font $F$ by picking the font for size $s$ in family $f$. +\li It determines a character number $n$ from the mathcode. +\li It selects as the character $c$ to be typeset the character +at position $n$ of font $F$. +\li It adjusts the spacing around $c$ according to the class of $t$ and +the surrounding context. +\li It typesets the character $c$. +\endolist + +The context dependence in +items (3) and (7) implies that \TeX\ cannot typeset a math character +until it has seen the entire formula containing the +math character. For example, in the formula +`|$a\over b$|', \TeX\ doesn't know what size the `|a|' should be until it +has seen the |\over|. + +{\tighten +The mathcode of a character is encoded according to the formula $4096c ++ 256f + n$, where $c$ is the class of the character, $f$ is its +\refterm{family}, and $n$ is its \refterm{\ascii\ character} code within +the family. You can change \TeX's interpretation of an input character +in math mode by assigning a value to the ^|\mathcode| +table entry \ctsref{\mathcode} +for that character. The character must have a +\refterm{category code} of $11$ (letter) or $12$ (other) for \TeX\ to +look at its |\mathcode|. +}\par + +^^{family//variable} +You can define a mathematical character to have a ``variable'' family by +giving it a class of $7$. Whenever \TeX\ encounters that character in a +math formula, it takes the family of the character to be the current +value of the |\fam| parameter \ctsref{\fam}. A variable family enables +you to specify the font of ordinary text in a math formula. For +instance, if the roman characters are in family $0$, the assignment +|\fam = 0| +will cause ordinary text in a math formula to be set in roman type +rather than in something else like math italic type. If the value of +|\fam| is not in the range from $0$ to $15$, \TeX\ takes the value to be +$0$, thus making classes $0$ and $7$ equivalent. +\TeX\ sets |\fam| to $-1$ whenever it enters math mode. +\endconcept + + +\conceptindex{mathematical units} +\concept{mathematical unit} + +A \defterm{mathematical unit}, denoted by `|mu|', is a unit of distance +that is used to specify \refterm{glue} in math formulas. See +\conceptcit{muglue}. +\endconcept + + +\conceptindex{modes} +\concept mode + +When \TeX\ is processing your input in its stomach \seeconcept{\anatomy}, +it is in one of six \defterm{modes}: +\ulist\compact +\li ^{ordinary horizontal mode} (assembling a paragraph) +\li ^{restricted horizontal mode} (assembling an \refterm{hbox}) +\li ^{ordinary vertical mode} (assembling a page) +\li ^{internal vertical mode} (assembling a \refterm{vbox}) +\li ^{text math mode} (assembling a formula that appears in text) +\li ^{display math mode} +(assembling a formula that appears on a line by~itself) +^^{horizontal mode}^^{vertical mode}^^{math mode} +\endulist +The mode describes the kind of entity that \TeX\ is putting together. + +Because you can embed one kind of entity within another, e.g., a vbox +within a math formula, \TeX\ keeps track not just of one mode but of a +whole list of modes (what computer scientists call a ``stack''). +Suppose that \TeX\ is in mode $M$ and encounters something that +puts it into a new mode \Mprimeperiod. When it finishes its work in +mode \Mprimecomma, it resumes what it was doing in mode \Mperiod. + +\endconcept + + +\concept muglue + +\defterm{Muglue} is a kind of \refterm{glue} that you can use only in math +formulas. It is measured in ^|mu| (\refterm{mathematical +units:mathematical unit}). +^^{mathematical units}^^{glue//mathematical} +One |mu| is equal to \frac1/{18} em, where +the size of an em is taken from \refterm{family} 2 of the math fonts. +\TeX\ automatically adjusts the size of muglue according to the context. +For instance, a glue size of |2mu| is normally smaller within a +subscript than it is within ordinary text. +You must use the ^|\mskip| command to produce muglue. +For example, `|\mskip 4mu plus 5mu|' produces mathematical glue with +natural space of four |mu| and \refterm{stretch} of five |mu|. + +\endconcept + + +\conceptindex{numbers} +\concept number + +In \TeX, a \defterm{number} is a positive or negative integer. +You can write a number in \TeX\ in four different ways: +\olist\compact +\li as an ordinary decimal integer, e.g., |52| +\li as an octal number, e.g., |'14| ^^{octal numbers} +\li as a hexadecimal number, e.g., |"FF0| ^^{hexadecimal numbers} +\li as the code for an \refterm{\ascii\ character}, e.g., |`)| +or |`\)| +\endolist +\noindent +Any of these forms can be preceded by `|+|' or `|-|'. + +An octal number can have only the digits |0|--|7|. +^^{octal numbers} +A hexadecimal number can have digits |0|--|9| and +|A|--|F|, representing +values from $0$ to $15$. +^^{hexadecimal numbers} +You can't, alas, use lowercase letters when you write a hexadecimal number. +If you need an explanation of octal and hexadecimal numbers, +you'll find one on \knuth{pages~43--44}. + +A decimal, octal, or hexadecimal number +ends at the first character that can't be part of the number. +Thus a decimal number ends when \TeX\ sees, say, a letter, even though a +letter between `|A|' and `|F|' would not end a hexadecimal number. +You can end a number with one or more spaces and +\TeX\ will ordinarily ignore them.\footnote{ +When you're defining a macro that ends in a number, you should always +put a space after that number; otherwise \TeX\ may later combine that +number with something else.} + +The fourth form above specifies a number as the +\minref{\ascii} code for a character. +^^{characters//\ascii\ codes for} +\TeX\ ignores spaces after this form of number also. +You can write a number in this form either as |`|$c$ or as |`\|$c$. +The second form, though longer, has the advantage that you can use it +with \emph{any} character, even `|\|', `|%|', or `|^^M|'. +It does have one rather technical disadvantage: when \TeX\ is expanding +a token sequence for a command such as |\edef| or |\write|, +^^|\edef//expansion of {\tt\\'\it c} in| +^^|\write//expansion of {\tt\\'\it c} in| +occurrences of `|\|$c$' within numbers will also be expanded if they can be. +That's rarely the effect you want. + +The following are all valid representations of the decimal number +$78$: +\csdisplay +78 +078 "4E '116 `N `\N +| + + +You can't use a number in text by itself since a number isn't +a command. +However, you can insert the decimal form of a number in text +by putting a ^|\number| command (\xref\number) in front of it +or the roman numeral form by putting a ^|\romannumeral| command +in front of it. + +You can also use ^{decimal constant}s, i.e., numbers with a fractional part, +for specifying dimensions \seeconcept{dimension}. +A decimal constant has a ^{decimal point}, which +can be the first character of the constant. +You can use a comma instead of a period to represent the decimal point. +A decimal constant can be preceded by a plus or minus sign. +Thus `|.5in|', +`|-3.22pt|', and `|+1,5\baselineskip|' are valid dimensions. +You can't, however, use decimal constants +in any context \emph{other} than as the ``factor'' part of a dimension, +i.e., its multiplier. + + +\endconcept + + +\concept{ordinary mode} + +An \defterm{ordinary mode} is a \refterm{mode} that \TeX\ is in when it is +assembling a paragraph into lines or assembling lines +into a page. See \conceptcit{horizontal mode}, \conceptcit{vertical mode}. +\endconcept + + +\concept outer + +\bix^^{macros//outer} +An \defterm{outer} macro is one that you can't use in certain contexts +where \TeX\ is processing tokens at high speed. +The purpose of making a command outer is to enable \TeX\ to catch +errors before it's gone too far. +When you define a macro, you can make it outer with the +^|\outer| command \ctsref\outer. + +You cannot use an outer macro in any of the following contexts: +\ulist\compact +\li within an argument to a macro +\li in the parameter text or replacement text of a definition +\li in the preamble to an alignment +\li in the unexecuted part of a conditional test +\endulist +\noindent +An outer context is a context in which you can use an outer macro, +i.e., it's any context other than the ones just listed. + +For example, the following input would be a forbidden use of an +outer macro: +\csdisplay +\leftline{\proclaim Assertion 2. That which is not inner + is outer.} +| +The |\proclaim| macro (\xref{\@proclaim}) is defined in \plainTeX\ +to be outer, but it's being used here in a macro argument to |\leftline|. +\eix^^{macros//outer} + +\endconcept + + +\concept {output routine} + +When \TeX\ has accumulated +at least enough material to fill up a page, it chooses a breakpoint +and places the material before the breakpoint in |\box255|. It then +calls the +current \defterm{output routine}, which processes the material and eventually +sends it to the \dvifile. +^^{\dvifile//material from output routine} +The output routine can perform further +processing, such as inserting headers, footers, and footnotes. +\refterm{\PlainTeX:\plainTeX} provides +a default output routine that inserts a centered page number +at the bottom of each page. +By providing a different output routine you can achieve such +effects as double-column output. +You can think of the output routine as having a single responsibility: +disposing of the material in |\box255| one way or another. + +The current output routine is defined by the value of ^|\output| +\ctsref{\output}, which is a list of \refterm{tokens:token}. When \TeX\ +is ready to produce a page, it just expands the token list. + +You can make some simple changes to the actions of the \plainTeX\ +output routine without actually modifying it. For example, by assigning +a list of \refterm{tokens:token} to |\headline| or +|\footline| \ctsref{\footline} you can have \TeX\ produce a different +header or footer than it ordinarily would. + +The output routine is also +responsible for collecting any \refterm{insertions:insertion}; +combining those insertions and any +``decorations'' such as headers and +footers with the main contents of the page and packaging all +of this material in a box; and +eventually sending that box to the \dvifile\ +^^{\dvifile//material from output routine} +with the ^|\shipout| +command \ctsref{\shipout}. +Although this is what an output routine most often does, +a special-purpose output routine might behave differently. +\endconcept + + +\conceptindex{output streams} +\concept {output stream} + +\margin{This concept was out of order.} +See \conceptcit{file}. +\endconcept + + +\conceptindex{pages} +\concept page + +\TeX\ processes a document by assembling \defterm{pages} one at a time +and passing them to the output routine. +As it proceeds through your document, \TeX\ maintains a list of lines +and other +items to be placed on the page. (The lines are actually hboxes.) +This list is called the ``^{main vertical list}''. +Periodically \TeX\ goes through a process called ``exercising the +^{page builder}''. +The items added to the main vertical list between exercises of the +page builder are called ``^{recent contributions}''. + +The page builder first examines the main vertical list to see if it's +necessary to ship out a page yet, either because the items on the main +vertical list won't all fit on the page or because of an explicit item, +such as |\eject| \ctsref\eject, that tells \TeX\ to end the page. +If it's not necessary to ship out a page, then the page builder is done +for the time being. + +Otherwise the page builder analyzes the main vertical +list to find what it considers to be the best possible page break. +It associates penalties with various kinds of unattractive page +breaks---a break that would leave an +isolated line at the top or bottom of a page, a break just before a +math display, and so forth. It then +chooses the least costly page break, +where the cost of a break is increased by any penalty associated with that +break and by the badness of the page that would result +(see \knuth{page~111} for the cost formula). If it finds several +equally costly page breaks, it chooses the last one. + +{\tighten +Once the page builder has chosen a page break, +it places the items on the list that are before that break +into ^|\box255| and leaves the remaining ones for the next page. +It then calls the output routine. |\box255| acts as a mailbox, with the +page builder as the sender and the output routine as the receiver. +Ordinarily the output routine processes |\box255|, adds +other items, such as insertions, headers, and footers, to the page, and +ships out the page to the \dvifile\ +^^{\dvifile//material from output routine} +with a |\shipout| command. +(Specialized output routines may behave differently.) +From \TeX's standpoint, it doesn't matter whether or not the output +routine ships out a page; +the only +responsibility of the output routine is to process |\box255| one way or +another. +\par} + +{\tighten +It's important to realize that the best place to break a page isn't +necessarily the last possible place to break the page. +Penalties and other considerations may cause the page break +to come earlier. +Furthermore, \TeX\ appends items to the main vertical list in batches, +not just singly. +The lines of a paragraph are an example of such a batch. +For these reasons the page builder usually has items left over when it +breaks a page. +These leftover items then form the beginning of the main vertical list +for the next page (possibly in the middle of a batch). +Because items are carried over from one page to another, +you can't assume that as \TeX\ is processing +input, the current page number accurately reflects the page on which the +corresponding output will appear. See \knuth{pages~110--114} for a full +description of \TeX's page-breaking rules. +\par} + +\endconcept + + +\conceptindex{page breaks} +\concept{page break} + +A \defterm{page break} is a place in your document where \TeX\ ends a +page and (except at the end of the document) starts a new one. +See \conceptcit{page} for the process that \TeX\ goes through in choosing +a page break. + +You can control \TeX's choice of page breaks in several ways: +\ulist +\li You can insert a penalty (\xref{vpenalty}) +^^{penalties//in vertical lists} +between two items in the main vertical list. A positive +penalty discourages \TeX\ from breaking the page +there, while a negative penalty---a bonus, in other words---% +encourages \TeX\ to break the page there. A penalty of $10000$ +or more prevents a +page break, while a penalty of $-10000$ or less forces a page break. +You can get the same effects with the |\break| and +|\nobreak| commands \ctsref{vbreak}. + +\li You can adjust the penalties associated with page breaking +by assigning different values to \TeX's page-breaking +\refterm{parameters:parameter}. + +\li You can enclose a sequence of paragraphs +or other items in the main vertical list within a \refterm{vbox}, +thus preventing \TeX\ from breaking the page anywhere within the sequence. +\endulist + +Once \TeX\ has chosen a page break, it places the portion of the main vertical +list that precedes the break into |\box255|. +It then calls the current \refterm{output routine} +to process |\box255| and eventually ships its contents to the \dvifile. +^^{\dvifile//material from output routine} +The output routine must +also handle \refterm{insertions}, such as footnotes, that \TeX\ has accumulated +while processing the page. + +It's useful to know the places where \TeX\ can break a page: +\ulist +\li At glue, provided that the item preceding the glue is +a box, a whatsit, a mark, or an insertion. +When \TeX\ breaks a page at glue, it makes the break at the top +of the glue space and forgets about the rest of the glue. +\li At a kern that's immediately followed by glue. +\li At a penalty, possibly between the lines of a paragraph. +\endulist +When \TeX\ breaks a page, it discards any +sequence of glue, kerns, and penalty items that follows the break point. + + +\endconcept + + +\concept page builder + +See \conceptcit{page}. +\endconcept + + +\concept {page layout} + +\bix^^{margins} +\bix^^{headers} +\bix^^{footers} +When you're designing a document, you need to decide on its +\defterm{page layout}: the page size, +the margins on all four sides, the headers and footers, if any, +that appear at the top and bottom of the page, +and the amount of space between the body of the text and the headers or +footers. \TeX\ has defaults for all of these. It assumes an $8 \frac1/2$-% +by-$11$-inch page with margins of approximately one inch +on all four sides, no header, +and a footer consisting of a centered page number. + +The margins are determined jointly by the four parameters +^|\hoffset|, ^|\voffset|, ^|\hsize|, and ^|\vsize| (see +``margins'', \xrefpg{margins}, +for advice on how to adjust them). +\eix^^{margins} +The header normally consists of a single line that appears at the top of each +page, within the top margin area. You can set it by assigning +a \refterm{token} list to the ^|\headline| parameter (\xref{\headline}). +Similarly, +the footer normally consists of a single line that appears at the bottom +of each +page, within the bottom margin area. You can set it by assigning +a \refterm{token} list to the ^|\footline| parameter (\xref{\footline}). +For example, the input: +\csdisplay +\headline = {Baby's First Document\dotfill Page\folio} +\footline = {\hfil} +| +produces a header line like this on each page: +\vdisplay{ +\dimen0 = \hsize +\advance \dimen0 by -\parindent +\hbox to \dimen0{Baby's First Document\dotfill Page 19}} +\noindent +and no footer line. + +You can use marks to place the current topic of a section +of text into the header or footer. +^^{marks//with headers or footers} +See \conceptcit{mark} for an explanation of how to do this. +\eix^^{headers} +\eix^^{footers} +\endconcept + +\conceptindex{paragraphs} +\concept paragraph + +Intuitively, a \defterm{paragraph} is a sequence of input lines that's +ended by a blank line, by a ^|\par| command \ctsref{\@par}, +^^|\par//ending a paragraph with| +or by an intrinsically vertical command, such as |\vskip|. +More precisely, a paragraph is a sequence of commands that \TeX\ processes +in ordinary horizontal mode. +When \TeX\ has collected an entire paragraph, it forms it into a sequence of +lines by choosing line breaks \seeconcept{line break}. +The result is a list of hboxes with glue, interline penalties, +and interspersed vertical material between them. +Each hbox is a single line, and the glue is the interline glue. + +\eject +\TeX\ starts a paragraph when it's in a vertical mode +and encounters an inherently horizontal command. +In particular, it's in a vertical mode when it's just finished a paragraph, +so the horizontal material on the line after a blank input line starts the +next paragraph in a natural way. +There are many kinds of inherently horizontal commands, but the most common +kind is an ordinary character, e.g., a letter. + +\looseness = -1 +The ^|\indent| and ^|\noindent| commands +(\pp\xrefn{\indent},~\xrefn{\noindent}) +are also inherently horizontal commands that tell +\TeX\ either to indent or not to indent the beginning of a paragraph. +Any other horizontal command in vertical +mode causes \TeX\ to do an implicit |\indent|. +Once \TeX\ has started a paragraph, it's in ordinary horizontal mode. +It first obeys any commands that are in ^|\everypar|. +It then proceeds to collect items for the paragraph until it gets a signal +that the paragraph is ended. +At the end of the paragraph it +resets the paragraph shape parameters ^|\parshape|, |\hang!-indent|, +^^|\hangindent| +and ^|\looseness|. + +\TeX\ ordinarily translates a blank line into |\par|. +It also +inserts a |\par| into the input whenever it's in horizontal mode and +sees an intrinsically vertical command. +So ultimately the thing that ends a paragraph is always a |\par| command. + +When \TeX\ receives a |\par| command, it first +fills out\footnote{% +More precisely, it executes the commands: +\csdisplay +\unskip \penalty10000 \hskip\parfillskip +| +thus appending items for these commands +to the end of the current horizontal list.} +the paragraph it's working on. +It then breaks the paragraph into lines, +adds the resulting list of items to the enclosing vertical list, +and exercises the page builder +(in the case where the enclosing vertical list is the main vertical list). +If the paragraph was ended by an intrinsically vertical command, +\TeX\ then executes that command. + +\endconcept + +\conceptindex{parameters} +\concept parameter + +The term \defterm{parameter} has two different meanings---it can refer +either to a \TeX\ parameter or to a macro parameter. + +A \TeX\ parameter is a \refterm{control sequence} that names +a value. +The value of a parameter can be a \refterm{number}, a \refterm{dimension}, +an amount of \refterm{glue} or muglue, or a \refterm{token list}. +For example, the ^|\parindent| parameter +specifies the distance that \TeX\ skips at the start of an +indented paragraph. + +You can use the control sequence for a parameter either to retrieve the value +of the parameter or to set that value. \TeX\ interprets the control sequence +as a request for a value if it appears in a context where a value is expected, +and as an \refterm{assignment} otherwise. +^^{assignments} +For example: +\csdisplay +\hskip\parindent +| +produces horizontal \refterm{glue} whose natural size is given by |\parindent|, +while: +\csdisplay +\parindent = 2pc % (or \parindent 2pc) +| +sets |\parindent| to a length of two picas. The assignment: +\csdisplay +\parindent = 1.5\parindent +| +uses |\parindent| in both ways. Its effect is to multiply the value of +|\parindent| by $1.5$. + +You can think of a parameter as a built-in \refterm{register}. +^^{registers//parameters as} +You'll find a complete list of the \TeX\ parameters on \knuth{pages~272--275}. + +A \refterm{macro} parameter is a placeholder for text that is to be +plugged into the definition of a macro. See \conceptcit{macro} +for more information about this kind of parameter. + +\endconcept + + +\conceptindex{penalties} +\concept penalty + +A \defterm{penalty} is an item that you can include in a +horizontal, vertical, or math list +in order to discourage \TeX\ from breaking the list +at that point or encourage \TeX\ to break the list there. +^^{horizontal lists//penalties in} +^^{vertical lists//penalties in} +A positive penalty indicates a bad break point, while +a negative penalty indicates a good break point. +Breaking an +ordinary horizontal +list produces a \refterm{line break}, while breaking an +ordinary vertical list produces +a \refterm{page break}. +(A penalty has no effect in restricted horizontal or +internal vertical \refterm{mode}.) + +You can use the +|\penalty| command (\pp\xrefn{hpenalty},~\xrefn{vpenalty}) +to insert a penalty explicitly. +A penalty of $10000$ or more prevents a break, while a penalty of +$-10000$ or less forces a break. +\endconcept + + +\concept{\plainTeX} + +\defterm{\PlainTeX} is the form of \TeX\ described in this +book and in \texbook. \PlainTeX\ is part of the standard \TeX\ +system, so documents that use only the facilities of \plainTeX\ can +usually be transferred from one installation to another without +difficulty. + +\PlainTeX\ consists of the \refterm{primitive} commands together with a +large collection of macros and +other definitions. These additional definitions are given in +\knuth{Appendix~B}. They should also be in the file |plain.tex| +somewhere in your computer system. +\endconcept + + +\concept{primitive} + +A \defterm{primitive} \refterm{command} is one whose definition is built +into the \TeX\ computer program. In contrast, a command that is not +primitive is defined by a \refterm{macro} or some other form of +definition written in \TeX\ itself. The commands in \refterm{\plainTeX} +consist of the primitive commands together with +other commands defined in terms of the +primitive ones. +\endconcept + + +\concept {reference point} + +^^{baselines} +^^{boxes//reference point of} +The \defterm{reference point} of a \refterm{box} is the point where the +left edge of the box intersects its \refterm{baseline}. When \TeX\ is +processing a \refterm{horizontal:horizontal list} or \refterm{vertical +list}, it uses the reference points of the boxes in the list to line up +those boxes horizontally or vertically \seeconcept{box}. +\endconcept + +\conceptindex{registers} +\concept register + +A \defterm{register} is a named location for storing a value. +It is much like a variable in a programming language. +\TeX\ has five kinds of registers, as shown in the following table: + +\vdisplay{\tabskip 10pt\halign{\tt #\hfil &#\hfil\cr +{\it Register type}&{\it Contents}\cr +box&a \refterm{box} \idxref{box registers}\cr +count&a \refterm{number} \idxref{count registers}\cr +dimen&a \refterm{dimension} \idxref{dimension registers}\cr +muskip&\refterm{muglue} \idxref{muglue registers}\cr +skip&\refterm{glue} \idxref{glue registers}\cr +toks&a \refterm{token} list\idxref{token registers}\cr}} + +The registers of each type are numbered from $0$ to $255$. +You can access register $n$ of category $c$ by using the form `|\|$cn$', +e.g., |\muskip192|. +You can use a register +anywhere that information of the appropriate type is called for. For +instance, you can use |\count12| +in any context calling for a number or |\skip0| +in any context calling for glue. + +You put information into a register by \refterm{assigning:assignment} +something to it: + +\csdisplay +\setbox3 = \hbox{lagomorphs are not mesomorphs} +\count255 = -1 +| +The first assignment constructs an hbox and assigns it to +box register~$3$. +You can +subsequently use `|\box3|' wherever a box is called for, and you will +get just that hbox.\footnote{But note carefully: using a box register +also empties it so that its contents become void. The other kinds of +registers don't behave that way. You can use the |\copy| command +\ctsref{\copy} to retrieve the contents of a box register without +emptying it.} +The second assignment assigns $-1$ to count register~$255$. + +A register of a given type, e.g., a glue register, behaves just like +a parameter of that type. +^^{parameters//like registers} +You retrieve its value or assign to it +just as you would with a \refterm{parameter}. +Some \TeX\ parameters, e.g., |\pageno|, +are implemented as registers, in fact. + +\PlainTeX\ +uses many registers for its own purposes, so you should not just +pick an arbitrary +register number when you need a register. Instead you should ask +\TeX\ to reserve a register by using one of the commands +^|\newbox|, ^|\newcount|, ^|\newdimen|, ^|\newmuskip|, ^|\newskip|, +or ^|\newtoks| +\ctsref{\@newbox}. These commands are outer, so you can't +use them in a macro definition. +If you could, +you'd use up a register every time the macro was called and probably run out +of registers before long. + +Nonetheless you can with some caution use any register temporarily +within a \refterm{group}, even one that \TeX\ is using for something +else. +After \TeX\ finishes executing the commands in a group, +it restores the contents of every register +to what they were before it started executing the group. +When you use an explicitly numbered register inside a group, +you must be sure that the register isn't modified by any +\refterm{macro} +that you might call within the group. +Be especially careful +about using arbitrary registers in a group that calls macros +that you didn't write yourself. + +{\tighten +\TeX\ reserves certain registers for special purposes: |\count0| through +|\count9| for page numbering information and +^^{page numbering} +^|\box255| for the contents +of a page just before it is offered to the \refterm{output routine}. +Registers |\dimen0|--|\dimen9|, |\skip0|--|\skip9|, +|\muskip0|--|\mu!-skip9|, |\box0|--|\box9|, +and the |255| registers other than |\box255| +are generally available as ``scratch'' registers. +Thus \plainTeX\ provides only one scratch register, |\count255|, for +counts. +See \knuth{pages~122 and 346} for conventions to follow +in choosing register numbers. +\par} + +You can examine the contents of registers during a \TeX\ run with the +^|\showthe| command \ctsref\showthe, e.g., with `|\showthe\dimen0|'. +\endconcept + + +\concept{restricted mode} + +A \defterm{restricted mode} is a \refterm{mode} that \TeX\ is in when it is +assembling an \refterm{hbox} or a \refterm{vbox}. +We follow \texbook\ in using the term ``internal vertical mode'' +for what you might expect to be ``restricted vertical mode''. +See \conceptcit{horizontal +mode} and \conceptcit{vertical mode}.\endconcept + + +\conceptindex{rules} +\concept rule + +A \defterm{rule} is a solid black rectangle. +A rule, like a \refterm{box}, +has \refterm{width}, \refterm{height}, and \refterm{depth}. +The vertical dimension of the rectangle +is the sum of its height and its depth. +An ordinary horizontal or vertical straight line is a special case of a rule. + +\bix^^{horizontal rules} +\bix^^{vertical rules} +\bix^^|\hrule| +\bix^^|\vrule| +A rule can be either horizontal or vertical. The distinction between a +horizontal rule and a vertical one has to do with how you produce the +rule, since a vertical rule can be short and fat (and therefore look +like a horizontal line), while a horizontal rule can be tall and skinny +(and therefore look like a vertical line). \TeX's notion of a rule is +more general than that of typographers, who think of a rule as a line +and would not usually call a square black box a rule. + +You can produce a horizontal rule using the +|\hrule| command and a vertical rule using +the |\vrule| command \ctsref{\vrule}. +For example, the control sequence |\hrule| by itself +produces a thin rule that runs across the page, like this: + +{\offinterlineskip +\nobreak\medskip +\hrule +\medskip} + +The command `|\vrule height .25in|' produces a vertical rule +that runs $.25$~inches down the page like this: +\nobreak\vskip \abovedisplayskip +\leftline{\vrule height .25in} +\vskip \belowdisplayskip + +There are two differences between horizontal rules and vertical rules: +\olist +\li For a horizontal rule, \TeX\ defaults the width to the width of the +smallest \refterm{box} or \refterm{alignment} that encloses it. For a +vertical rule, \TeX\ defaults the height and depth in the same way. (The +default is the size that you get if you don't give a size explicitly for that +dimension.) + +^^{horizontal lists//rule in} +^^{vertical lists//rule in} +\li +{\tighten +A horizontal rule is an inherently vertical item that cannot participate in +a \refterm{horizontal list}, +while a vertical +rule is an inherently horizontal item +that cannot participate in a \refterm{vertical list}. This behavior +may seem strange at first but there is good reason for it: +a horizontal rule ordinarily runs visually from left +to right and thus separates items in a vertical list, +while a vertical rule ordinarily runs visually from top to bottom +and thus separates items in a horizontal list. +%(Look at the rules that are shown above.) +\par} +\endolist + +{\tighten +If you construct a rule with three explicit dimensions, it will look the +same whether you make it a horizontal rule or a vertical rule. +For example, the command `|\vrule height1pt depth2pt width3in|' produces this +horizontal-looking rule: +\par} + +{\offinterlineskip +\nobreak\medskip\nobreak\vskip3pt +\leftline{\vrule height1pt depth2pt width3in} +\medskip} + +You'll find a precise statement of \TeX's treatment of rules on +\knuth{pages~221--222}. +\eix^^{horizontal rules} +\eix^^{vertical rules} +\eix^^|\hrule| +\eix^^|\vrule| +\endconcept + + +\concept {script size} + +\defterm{Script size} describes one of the three related +\refterm{fonts:font} in a family. +^^{family//script size in} +Script size is smaller than \refterm{text size} but larger than +\refterm{scriptscript size}. \TeX\ uses script size for subscripts and +superscripts, as well as for the numerators and denominators of +fractions in text. +\endconcept + + +\concept {scriptscript size} + +\defterm{Scriptscript size} describes the smallest of the three related +\refterm{fonts:font} in a family. +^^{family//scriptscript size in} +\TeX\ uses \refterm{scriptscript size} for second-order subscripts, +superscripts, numerators, and denominators. For example, \TeX\ will use +scriptscript size for a subscript on a subscript or for a superscript on +a scriptsize numerator. +\endconcept + +\concept shrink + +See \conceptcit{glue}. +\endconcept + +\concept space + +You can cause \TeX\ to put \defterm{space} between two items in +several~ways: + +\olist +^^{end of line} +\li You can write something that \TeX\ treats as a space +\refterm{token}: one or more blank characters, the end of a line (the +end-of-line character acts like a space), or any \refterm{command} that +expands into a space token. \TeX\ generally treats several consecutive +spaces as equivalent to a single one, including the case where the +spaces include a single end-of-line. (An empty line +indicates the end of a paragraph; it +causes \TeX\ to generate a |\par| token.) +^^|\par//from empty line| +\TeX\ adjusts the size of +this kind of space to suit the length required by the context. + +^^{glue//creating space with} +\li You can write a skip command that produces the glue +you specify in the command. The glue can +\refterm{stretch} or \refterm{shrink}, +producing more or less space. You can have vertical glue as +well as horizontal glue. Glue disappears whenever it is next to a +line or page break. + +^^{kerns//creating space with} +\li You can write a \refterm{kern}. A kern produces a fixed amount of +space that does not stretch or shrink and does not disappear at a line +or page break (unless it is immediately followed by glue). The most +common use of a kern is to establish a fixed spatial relationship +between two adjacent \refterm{boxes}. +\endolist + +Glue and kerns can have negative values. Negative glue or a negative kern +between adjacent items brings those items closer together. +\endconcept + + +\concept stretch + +See \conceptcit{glue}. +\endconcept + + +\conceptindex{struts} +\concept strut + +{\tighten +A \defterm{strut} is an invisible \refterm{box} +^^{boxes//invisible} +whose width is zero and whose height and depth are slightly more than +those of a ``normal'' +line of type in the context. Struts are useful for obtaining +uniform vertical spacing when \TeX's +usual line spacing is disabled, e.g., within a math formula +or within a horizontal alignment where you've specified ^|\offinterlineskip|. +Because a strut is taller and deeper than everything else on its line, +it determines the height and depth of the line. +You can produce a strut with +the ^|\strut| command \ctsref{\strut} or the ^|\mathstrut| command +\ctsref\mathstrut. +You can use |\strut| anywhere, but you can only use |\mathstrut| when +\TeX\ is in math \refterm{mode}. A strut in \plainTeX\ has height 8.5\pt\ and +depth 3.5\pt, while a math strut has the height and depth of a left +parenthesis in the current \refterm{style} (so it's smaller for +subscripts and superscripts). +\par} + +Here's an example showing how you might use a strut: +\csdisplay +\vbox{\hsize = 3in \raggedright + \strut Here is the first of two paragraphs that we're + setting in a much narrower line length.\strut} +\vbox{\hsize = 3in \raggedright + \strut Here is the second of two paragraphs that we're + setting in a much narrower line length.\strut} +| +This input yields: +\display{\vbox{ +\vbox{\hsize = 3in \raggedright +\strut Here is the first of two paragraphs that we're setting +in a much narrower line length.\strut} +\vbox{\hsize = 3in \raggedright +\strut Here is the second of two paragraphs that we're setting +in a much narrower line length.\strut} +}} +\noindent +Without the struts the \refterm{vboxes:vbox} would be too close +together. Similarly, in the formula: +\csdisplay +$\overline{x\mathstrut} \otimes \overline{t\mathstrut}$ +| +the math struts cause both bars to be set at the same height even +though the `$x$' and the `$t$' have different heights: +\display{ +$\overline{x\mathstrut} \otimes \overline{t\mathstrut}$ +} +\vskip -\belowdisplayskip +\endconcept +\nobreak + +\conceptindex{styles} +\concept {style} + +Material in a math formula is set in one of eight \defterm{styles}, +depending on the context. Knowing about styles can be useful if you want to +set part of a formula in a different size of type than the one +that \TeX\ has chosen according to its usual rules. + +\eject +The four primary styles are: + +\vdisplay{% +\halign{\refterm{# style}\hfil&\hskip .25in(for #)\hfil\cr +display&formulas displayed on a line by themselves\cr +text&formulas embedded in ordinary text\cr +script&superscripts and subscripts\cr +scriptscript&superscripts on superscripts, etc.\cr +}} + +The other four styles are so-called ^{cramped variants}. In these +variants superscripts aren't raised as high as usual, and so the formula +needs less vertical space than it otherwise would. See +\knuth{pages~140--141} for the details of how \TeX\ selects the style. + +\TeX\ chooses a size of type according to the style: + +\ulist\compact +^^{display style}^^{text style} +\li Display style and text style are set in \refterm{text size}, like +`$\rm this$'. + +^^{script style} +\li Script style is set in \refterm{script size}, like `$\scriptstyle +\rm this$'. + +^^{scriptscript style} +\li Scriptscript style is set in \refterm{scriptscript size}, like +`$\scriptscriptstyle \rm this$'. +\endulist + +See \conceptcit{family} for more information about these three sizes. + +\TeX\ doesn't have a ``scriptscriptscript'' style because such a style +would usually have to be set in a size of type too small to read. \TeX\ +therefore sets third-order subscripts, superscripts, etc., using the +scriptscript style. + +Once in a while you may find that \TeX\ has set a formula in a different style +than the one you'd prefer. You can override \TeX's choice with the +^|\textstyle|, ^|\displaystyle|, ^|\scriptstyle|, and ^|\scriptscriptstyle| +commands \ctsref{\textstyle}. +\endconcept + +\concept {\TeXMeX} + +(a)~A variant of \TeX\ used for mathematical typesetting in +Central American countries. +(b)~A spicy cuisine favored by the \TeX\-ni\-cians of ^{El Paso}. +\endconcept + + +\concept {text math} + +We use the term \defterm{text math} to refer to a math formula set within a +line of text, i.e., enclosed in |$|'s. +\ttidxref{$} +\TeX\ sets text math in text math \refterm{mode}. +\endconcept + + +\concept {text size} + +\defterm{Text size} describes the largest of the three related +\refterm{fonts:font} in a \refterm{family}. +^^{family//text size in} +\TeX\ uses text size for ordinary symbols appearing in +\refterm{math mode}. +\endconcept + + +\conceptindex{tokens} +\concept token + +A \defterm{token} is either a single character tagged with a +\refterm{category code}, or a \refterm{control sequence}. \TeX\ reads +characters from a file using its eyes \seeconcept{\anatomy} +and groups the characters into tokens using its mouth. When a token +reaches \TeX's stomach, \TeX\ interprets it as a \refterm{command} +unless it's part of an argument of a preceding command. +\endconcept + + +\conceptindex{units of measure} +\concept {unit of measure} + +See \conceptcit{dimension}. +\endconcept + +\conceptindex{vboxes} +\concept vbox + +^^{vertical lists//vboxes formed from} +A \defterm{vbox} (vertical box) is a \refterm{box} that \TeX\ constructs +by placing the items of a \refterm{vertical list} one after another, top +to bottom. A vbox, taken as a unit, is neither inherently horizontal +nor inherently vertical, i.e., it can appear as an item of either a +vertical list or a \refterm{horizontal list}. You can construct a vbox +with the ^|\vbox| or the ^|\vtop| command +\ctsref{\vtop}. The difference is that for |\vbox|, +the \refterm{reference point} +of the constructed vbox is derived from that of the last (and +usually bottommost) constituent list item, but for |\vtop|, it's that of the +first (and usually topmost) constituent list item. +\endconcept + + +\conceptindex{vertical lists} +\concept{vertical list} + +A \defterm{vertical list} is a list of items +that \TeX\ has produced while it is +in one of its vertical modes, i.e., +assembling either a +\refterm{vbox} or a page. See ``vertical mode'' below. + +\endconcept + + +\concept{vertical mode} + +^^{vboxes//vertical mode for} When \TeX\ is assembling either a +\refterm{vbox} or the main vertical list from which pages are derived, +it is in one of two \defterm{vertical modes}: ^{ordinary +vertical mode} for assembling the main vertical list, and ^{internal +vertical mode} for assembling vboxes. Whenever \TeX\ is in a vertical mode +its stomach \seeconcept{\anatomy} is constructing a \refterm{vertical +list} of items (boxes, glue, penalties, etc.). +\TeX\ typesets the items in the list +one below another, top to bottom. + +A vertical list can't contain any items produced by +inherently horizontal commands, e.g., +^^{vertical lists//can't contain horizontal commands} +|\hskip| or an ordinary (nonspace) character. +\footnote{\TeX\ \emph{ignores} any space characters +that it encounters while it's in a vertical mode.} + +\ulist +\li If \TeX\ is assembling a vertical list in ordinary vertical mode and +encounters an inherently horizontal command, it switches to ordinary +\refterm{horizontal mode}. +\li If \TeX\ is assembling a vertical list in internal vertical mode and +encounters an inherently horizontal command, it complains. +\endulist + +Two commands that you might at first think are inherently vertical are +in fact inherently horizontal: |\valign| \ctsref{\valign} and |\vrule| +\ctsref{\vrule}. +^^|\valign//inherently horizontal| +^^|\vrule//inherently horizontal| +See \knuth{page~283} for a list of the +inherently horizontal commands. + +It's particularly important to be aware that \TeX\ considers an ordinary +character other than a space character to be inherently horizontal. If +\TeX\ suddenly starts a new paragraph when you weren't expecting it, +a likely cause is a +character that \TeX\ encountered while in vertical mode. +You can convince \TeX\ +not to treat that character as inherently horizontal by enclosing it in +an \refterm{hbox} since the |\hbox| command, despite its name, is +neither inherently horizontal nor inherently vertical. +\endconcept + + +\concept whatsit + +A \defterm{whatsit} is an item of information that +tells \TeX\ to carry out some action +that doesn't fit into its ordinary scheme of things. +A whatsit can appear in a horizontal or vertical list, just like a box +or a glue item. +\TeX\ typesets a whatsit +as a \refterm{box} having zero width, height, and depth---in other +words, a box that contains nothing and occupies no space. + +Three sorts of whatsits are built into \TeX: +\ulist +\li The |\openout|, |\closeout|, and |\write| commands +(\p\xrefn{\openout}) +% (2nd) removed \xref to \write, since it's on the same page +produce a whatsit for operating on an output file. +^^|\openout//whatsit produced by| +^^|\write//whatsit produced by| +^^|\closeout//whatsit produced by| +\TeX\ postpones the operation until it next ships out a page +to the {\dvifile} +^^{\dvifile//whatsits in} +(unless the operation is preceded by ^|\immediate|). +\TeX\ uses a whatsit for these commands because they don't have anything +to do with what it's typesetting when it encounters them. +\li The ^|\special| command \ctsref{\special} tells \TeX\ to +insert certain text directly into the \dvifile. +As with the |\write| command, \TeX\ postpones the insertion +until it next ships out a page to the {\dvifile}. +^^{\dvifile//material inserted by \b\tt\\special\e} +A typical use of |\special| would be to +name a graphics file that the device driver should incorporate into +your final output. +\li When you change languages with the ^|\language| or ^|\setlanguage| +commands \ctsref{\language}, +\TeX\ inserts a whatsit that instructs it to use a +certain set of hyphenation rules later on when it's breaking a paragraph +into lines. +\endulist +\noindent +A particular implementation of \TeX\ may provide additional whatsits. +\endconcept + + +\concept width + +^^{boxes//width of} +The \defterm{width} of a \refterm{box} is the amount of horizontal space +that it occupies, i.e., the distance from its left edge to its +right edge. The typeset material in a box can be wider than the box itself. +\endconcept + +\endconcepts + +\endchapter +\byebye |