\chapter{More things you should know} %\addtocontents{toc}{sleeve} \section{Changing font} Once desktop publishing arrived\footnote{Desktop publishing arrived long before Paul Brainerd of Aldus invented the term and Apple created the bandwagon. Xerox \textsc{Parc} had been doing all that about 1978 with the Bravo system -- even to the extent of on-demand immediate laser printing. \TeX\ users throughout the known world had been publishing via their desktop terminals, and the \textsc{Unix} world had also been using the \texttt{nroff}\slash \texttt{troff} family and its pre-processors. The key to success is marketing.} people expected to change font\index{fonts} at the drop of a hat. Brought up on typewriters, with no typographic knowledge at all, they suddenly acquired a whole new vocabulary of `Palatino', `Bookman', `Zapf Chancery', and strongly held opinions about kerning, letterspacing, serifs, tracking and other arcane typographical subjects (rational views may be found in~\cite{RM80} and~\cite{PL92}). \section{Some history (again)} When Knuth designed \TeX, he also designed fonts to go with it. Or rather, he used an existing typeface, Monotype's Modern 8A, and produced what he termed `Computer Modern'. He had some help, but basically this `family' is the suite of fonts with which \TeX\ (and \LaTeX) were initially tuned. That is not to say you cannot use others, just that some work \emph{may} be needed before it will be a success. The Computer Modern family comprises about 75 `different' fonts. Most families comprise three or four different fonts -- medium, bold, italic and so on. Now, 75 fonts does not mean that they are all strikingly and immediately different: Knuth took the notion of a `design size' rather seriously. What this means is that the font is designed to be displayed and read at a specific size. Thus we have several versions of Computer Modern\index{Computer Modern} Roman, designed to be read at 5, 6, 7, 8, 9, 10, 12, 17 and 25 points~\cite{DEK-cm}. These are genuinely different: the proportions change subtly as we change size. Contemporary digital fonts tend to be stored as `outline' information, where to arrive at a particular size, we `merely' magnify the `prototype' outline. This will not change the outline, just the size. Adobe encoded `hints' into their \PS\ `Type~1' outlines, which help preserve the subtle changes. MicroSoft's `TrueType' fonts contain similar hints -- but distinct enough to avoid legal action. However, these hints have as much to do with accommodating digital fonts to the various resolutions at which they may be displayed as to modifying them for the point size at which they are displayed. Returning to the Computer Modern fonts, they are usually stored as `rather compressed' bit maps, for a given resolution. Thus a laser printer would typically require resolutions of 300\,dpi or 600\,dpi, and a phototypesetter perhaps 1270\,dpi. Outline fonts clearly have an advantage here, since you would only need one font to be stored, The rasterization of an outline takes place at print time, in the printer's own processor. At the time, Knuth's solution seemed a good one. It certainly assured quality, but it does require a lot of storage space for all those bit maps. But disk space is cheap. And if you can store the bit maps down on your laser printer or phototypesetter, the data transmission times can be reduced. If you are working on a network, you may have to store the bit maps on a server, where you then have the time consuming business of shipping them down to the printer. If we delve a little more into the process, we discover that Knuth also created a `language' to describe fonts -- \MF~\cite{DEK86}. The descriptions of Computer Modern are encoded in \MF\ (see, for example~\cite{DEK-cm}). Some implementations of \LaTeX\ will include \MF, and you may find that when you need particular fonts, they are generated on the fly. Typically, you will take many years to require all the Computer Modern fonts at all the sizes possible, and you can afford to accumulate them over time, \MF\ generating them as required. Since some of these fonts may only ever be required in rather exceptional and infrequent circumstances, a common strategy is to delete the less well used ones after use. Given the current processor speed commonly available, this is quite a sound approach. But it does depend on you having a suitably configured system where all these various components can inter-communicate. Equally, Postscript and Truetype versions of the Computer Modern fonts are now generally available, both commercially and within the public domain. For many years \TeX\ and \LaTeX\ gained a reputation of working with only Computer Modern and unfortunately, some people, especially publishers, did not like Computer Modern, usually preferring Times. The reputation was unfounded, but it stemmed from two main origins: firstly, it did take some time and effort (and sometime access to proprietary information) to utilise other fonts; and secondly, other fonts did not necessarily have the full range of characters which were readily available for Computer Modern. Eventually the problem was solved quite comprehensively by Frank Mittelbach and Rainer Sch{\"o}pf with the introduction of the `New Font Selection Scheme'~\cite{FM89} or \textsf{NFSS}. Although we will look at this later, if you are anxious to eschew the use of good old Computer Modern for your particular favourite, consult Chapter~7 of the \textsl{Companion}. \section{What fonts do we have?} We won't actually use all the fonts\index{fonts} here, but just look at the ones we get by default with \LaTeX.\footnote{These notes were originally written with the intention that the font used would be Computer Modern. The house style of this august organ is Baskerville. If the \textsf{NFSS} transformations have been done correctly, all should be well.} Depending how you count, there are eleven types or `styles' of font available, divided into three independent groups: the `family' group is Roman, Sans Serif and Typewriter: the `series' group comprises Medium and Bold: while the `shape' group is Upright, Italic, Slanted and Small Capitals. That leaves one orphan, Normal. Normal is the style used in the `body' of the text. But even then that only comes to ten. The eleventh is an `emphasis' style. The results of using emphasis is to change the appearance in some way. Note that we do not use underlining at all -- that is a typing convention for emphasis which is \underline{never}\footnote{hardly ever} used in typesetting. We are in the Gutenberg tradition, not the Sholes tradition. Table~\ref{tfonts} shows what these fonts look like on the page. If you look closely you will note that there appears to be no difference between Upright, Medium, Roman and Normal. At this stage, this is true; later we shall see the distinctions in more detail. \begin{question} Sholes who? By now, very few people have direct experience of having used a typewriter. For those who do, what other leftovers from the days of the typewriter afflict us? Many word processing packages defer to typewriting practise and may provide hints and clues. Is it necessary to ask who Gutenberg was? or Caxton? or the cutely named Wynkyn de Worde? or even Baskerville? \end{question} \begin{table*} \begin{center} \begin{tabular}{|llp{3in}|} \hline Style & instruction & example \\ \hline Roman &\verb|\textrm| &\textrm{The quick brown fox comes to the aid of the Hamburgerfons} \\Upright &\verb|\textup| &\textup{The quick brown fox comes to the aid of the Hamburgerfons} \\ Sans Serif &\verb|\textsf| &\textsf{The quick brown fox comes to the aid of the Hamburgerfons}\\ Typewriter &\verb|\texttt| &\texttt{The quick brown fox comes to the aid of the Hamburgerfons}\\ Medium &\verb|\textmd| &\textmd{The quick brown fox comes to the aid of the Hamburgerfons} \\ Bold & \verb|\textbf|&\textbf{The quick brown fox comes to the aid of the Hamburgerfons}\\ Upright & \verb|\textup|&\textup{The quick brown fox comes to the aid of the Hamburgerfons}\\ Italic & \verb|\textit|&\textit{The quick brown fox comes to the aid of the Hamburgerfons}\\ Slanted &\verb|\textsl| &\textsl{The quick brown fox comes to the aid of the Hamburgerfons}\\ Small Capitals & \verb|\textsc|&\textsc{The quick brown fox comes to the aid of the Hamburgerfons}\\ Normal &\verb|\textnormal| &\textnormal{The quick brown fox comes to the aid of the Hamburgerfons}\\ \hline \end{tabular} \end{center} \caption{font styles\index{fonts} readily available}\label{tfonts} \end{table*} To change the type `style', we use the simple, fairly mnemonic instruction given in Table~\ref{tfonts}. \LaTeX\ has a very powerful technique available which makes life easy -- grouping. Grouping is a notion central to the whole existence of \LaTeX. We have already met some sorts of grouping, where we have a \verb|\begin| and \verb|\end|. There is a simpler grouping\index{grouping} -- \verb|{| and \verb|}|. If we, for example say \begin{verbatim} The song really \emph{is} \textit{A-Sitting On A Gate}: and the tune's my own invention. \end{verbatim} the parts enclosed in braces are the parts which will be emphasised (by \verb!\emph!), or turned into italics (by \verb!\textit!). The remainder of the passage will use whatever happens to be the font already in use. The word \verb|is| is really `emphasised'\index{emphasis}, while the \verb|A-Sitting On A Gate| part is a title (of sorts -- see Table~\ref{talice}). We may wish all such titles to be represented in a particular font or style, to assist readers distinguish this sort of information. It is a good idea to review the reasoning behind changing font: presumably it is to provide the reader with visual clues to the structure of the contents of the document. Thus headings are usually in a different font from the text, and different levels of heading usually have slightly different characteristics: quoted text is often presented in a different font, and so too is emphasised material. Given that human perception seems to work best with $7\pm2$ discrete items at a time, you can see that you probably do not want too many different fonts at a time. \begin{table*} \begin{center} \small \begin{tabular}{|ll|} \hline what the name of the song is called& Haddocks' Eyes \\ what the name is & The Aged Aged Man\\ what the song is called & Ways and Means\\ what the song really {\em is} & A-sitting On A Gate\\ the tune & I give thee all, I can no more\\ \hline \end{tabular} \end{center} \caption{Carrolliginian confusion -- totally irrelevant here}\label{talice} \end{table*} At last, this brings in the use of emphasis: \LaTeX\ uses the instruction \verb|\emph|\index{emph@\texttt{emph}} to denote emphasis. Let's look again at the last example to see that \begin{verbatim} The song really \emph{is} \textit{A-Sitting On A Gate}: and the tune's my own invention. \end{verbatim} gives \begin{quote} The song really \emph{is} \textit{A-Sitting On A Gate}: and the tune's my own invention. \end{quote} Here we have distinguished the two elements `emphasised text' and `italicized text'. In this circumstance, it is hardly visible, except in the marked up text. But let's change things a bit: \begin{verbatim} \textit{The song really \emph{is} \textit{A-Sitting On A Gate}: and the tune's my own invention.} \end{verbatim} This time we find that the emphasised text is in the roman font: \begin{quote} \textit{The song really \emph{is} \textit{A-Sitting On A Gate}: and the tune's my own invention.} \end{quote} \LaTeX\ is smart enough to do that, but unfortunately is not smart enough to change the title information. After all, you have specified that it is italicised. The fact that you then repeat the italicisation instruction changes nothing, although you may have wished that the second \verb+\textit+ instruction changes to some different font so that that part looks different. Nevertheless, this can be a useful feature. Do not think of `emphasis'\index{emphasis} as simply `italics', because that is not what it means. It is a function which may be implemented in a variety of ways, depending on the circumstances. \begin{question} Take some of the earlier text material and change fonts, either over the whole document, or just selected parts. Which fonts do you find easier to read? Are the fonts which are easier to read on the screen also the fonts which are easier to read on paper? \end{question} \section{Bigger or smaller} You will have observed that fonts are available in more than one size. \LaTeX\ has a series of instructions which allow you to change the size of any font, very easily. The instructions used are given in Table~\ref{tsize}. Their exact behaviour depends on the options you have set up. The order of the sequence remains constant, but sometimes two adjacent `sizes' may use the same sized font (without telling you). \begin{table*} \begin{center} \begin{tabular}{|ll|} \hline size&style\\ \hline \verb|\tiny| & \tiny Aa\\ \verb|\scriptsize| & \scriptsize Aa\\ \verb|\footnotesize| & \footnotesize Aa\\ \verb|\small| & \small Aa\\ \verb|\normalsize| & \normalsize Aa\\ \verb|\large| & \large Aa\\ \verb|\Large| & \Large Aa\\ \verb|\LARGE| & \LARGE Aa\\ \verb|\huge| & \huge Aa\\ \verb|\Huge| & \Huge Aa\\ \hline \end{tabular} \end{center} \caption{the sizes\index{size} available}\label{tsize} \end{table*} The immediate question is `are all these sizes available for all the fonts?' \begin{question} Are all these sizes available for all the fonts? Remember to try the different size options as well as the different fonts. \end{question} The way that these size-changing instructions are used is quite different from the font-changing instructions. A font changing instruction looks like \verb+\textrm{text}+, while to change size you say \verb+{\small text}+. This difference is quite crucial. The size changing merely indicates a new condition which is operational until the end of the group in which it occurs. We could, for example, say \begin{verbatim} \Huge Come, \huge I'll \LARGE take \Large no \large denial, \normalsize we must \small have a \tiny trial \end{verbatim} and each change in size would take place when the instruction occurs in the text: \begin{quote} \Huge Come, \huge I'll \LARGE take \Large no \large denial, \normalsize we must \small have a \tiny trial \end{quote} We might be advised to enclose the whole thing in braces, lest any succeeding text was also placed in the \verb+\tiny+ size of font. If the size changes are not grouped, separating them out, they remain in effect. This different style of usage harks back to the older form of \LaTeX\index{latex!latex2.09@\protect{\LaTeX\,2.09}}. It should not cause too much confusion, since you would not normally want to change size in your text. Any such size changes should be restricted to special conditions, like section headings or new environments, which you would not be writing as you go along, but would have been defined separately. Later we may see how to do these. \section{Accumulation} It may come as a pleasant surprise to realise that it is possible to combine the font changing and the size changing mechanisms, and that something sensible happens. For example, \begin{verbatim} a \textsf{\Large mouses's} \textit{\tiny tail} \end{verbatim} will result in `a \textsf{\large mouses's} \textit{\tiny tail}', which is presumably the effect we wished. Sometimes fonts are not available in all sizes: had you written \begin{verbatim} a \textsl{\large mouses's} \textsc{\tiny tail} \end{verbatim} you may find there is no \verb|\tiny| small capitals font available. \LaTeX\ will substitute something else -- and does so without telling you. To some extent this lack of a suitable font is implementation dependent, since it can be possible to create or obtain the `correct' font for use in this situation. In the current conditions it would yield: \begin{quote} a \textsl{mouse's} {\tiny \textsc{tail}} \end{quote} Do note that these size changing instructions are absolute, not relative. Saying something like \begin{verbatim} a {\small mouses's {\small tail}} \end{verbatim} will not make the tail even smaller. In passing, note that if there is any chance of the text appearing on more than one line, it would be best to end a paragraph (by inserting a blank line) before the closing \verb+}+ or else you may get small text on the existing line spacing. Or if you were using \verb|\large|, the even worse case of large text on the existing line spacing, with ungainly and uneven line spacing as \LaTeX\ moves things around to fit. But you really ought not to be changing size as you go along like this. The examples here are hardly the stuff of the normal article, report or book. In the previous example, we would not be able to obtain a slanted small capital font by \begin{verbatim} a \textsl{mouse's {\tiny \textsc{tail}}} \end{verbatim} But if we return to our description of fonts in three groups, a family, a series and a shape, we will find that these three are truly indpendent. That is to say I can ask for a bold sans serif \textbf{\textsf{mouse's tail}} by \begin{verbatim} \textbf{\textsf{mouse's tail}} \end{verbatim} or for a typewriter medium small caps \texttt{\textmd\textsc{{mouse's tail}}} by \begin{verbatim} \texttt{\textmd\textsc{{mouse's tail}}} \end{verbatim} This is an example that fails in the current font set up. If you examine the log file closely enough, you will find that you have been told, and also told what substitute has been chosen. This is a topic we'll look at in more detail later. In principle, you may combine any family with any series with any shape and expect to see something sensible. \begin{question} Try some of these combinations. How would you make the mouse's tail {\small progressively \scriptsize smaller}? \end{question} \section{Accenting the positive} Now it is time for something even more frivolous. One of the nice features of \LaTeX\ (although of marginal real use in English), is its excellent support of diacritical\index{diacriticals} marks and foreign letters. Naturally there is an ulterior motive for introducing these now. First the special letters. \LaTeX\ recognises instructions for the diphthongs\index{diphthongs} \OE, \AE, \oe\ and \ae\ (commonly used in Latin and in some Scandinavian languages, among others). It also recognises the German `\ss' (ess-zet) symbol. It will handle the \AA, \aa, \O\ and \o\ of some Scandinavian languages. And lastly, it copes with the Polish suppressed-L, \L\ and \l. How do we get these into our text? Follow the \verb+\+ by a special instruction as shown in Table~\ref{tnatchar}. \begin{table*} \begin{center} \begin{tabular}{|cl|} \hline instruction&explanation\\ \hline \verb|\ss| & gives the German \ss \\ \verb|\OE| & gives the \OE\ diphthong \\ \verb|\oe| & gives the \oe\ diphthong \\ \verb|\AE| & gives the \AE\ diphthong \\ \verb|\ae| & gives the \ae\ diphthong \\ \verb|\O| & gives the letter \O\ \\ \verb|\o| & gives the letter \o\ \\ \verb|\AA| & gives the letter \AA\ \\ \verb|\aa| & gives the letter \aa\ \\ \verb|\L| & gives the letter \L \\ \verb|\l| & gives the letter \l\ \\ \hline \verb|?`| & gives the symbol ?` \\ \verb|!`| & gives the symbol !` \\ \hline \end{tabular} \end{center} \caption{the so-called `national'\index{national characters} characters}\label{tnatchar} \end{table*} How do you use these new instructions? The `recommended' way for beginning \LaTeX-users is to enclose them in braces\index{braces}. This never fails. Thus to write \OE dipus in \LaTeX, you actually write \verb|{\OE}dipus|. This helps to distinguish \verb|{\OE}dipus| from \verb|\OEdipus|, which \LaTeX\ would assume was a new (probably unknown) instruction. There is another, briefer, way, which you will encounter frequently: leave a blank space between the instruction and the rest of the word \begin{verbatim} \OE dipus \end{verbatim} Note that in this form any extra spaces between \verb|\OE| and \verb|dipus| will be ignored, as far as creating the output is concerned. Since all extra spaces are ignored, leaving a few extra, or even writing \begin{verbatim} \OE dipus \end{verbatim} will not leave a space between the diphthong and the rest of the word which follows when the passage is set. In other words, a line break (a `carriage control') is just another space to \LaTeX. This gobbling up of extra blanks\index{space} is a normal feature of \LaTeX. If we group\index{grouping} the instruction, there is no need to follow it by a space, and in fact \verb|{\OE} dipus| will give the result `\OE{} dipus', which is not what was required. \begin{question} In what ways are the following different? \begin{verbatim} {\OE }dipus, {\OE} dipus, \OE {dipus}, \O{E}dipus \end{verbatim} And what other `useful' possibilities are there? \end{question} What implication does this shorthand method have for instructions which come at the end of words? Consider trying to write: \begin{verbatim} the Schlo\ss of the Rhine valley \end{verbatim} The word `Schlo\ss' would appear as we require, but the spaces which follow would be ignored, and the next word would begin immediately after the \ss. This is generally not what we want. In order to solve this problem, there are a variety of solutions. We could have used \verb+{\ss}+ (the grouped instruction), as recommended, or we can use a new instruction, \verb*|\ |, that is, the backslash followed by a space, which introduces a `command space'\index{space} (or a `control space' or even a `hard space'). Thus what we probably wanted was \begin{verbatim} the Schlo\ss\ of the Rhine valley \end{verbatim} In this approach, the `hard space' lets \LaTeX\ know where the instruction ends. If you look at this more closely, you will realise that there are other ways to signify the end of a instruction. One of them was illustrated with the \verb|{\OE}dipus| sequence. There the \verb|}| was able to indicate the end of the instruction. Actually, \verb|\OE{}dipus| would have had the same effect. The sequence \verb|{}| looks odd, and seems to mean nothing, but from time to time, even nothing has its uses. In the alternative shown with the \verb*|Schlo\ss\ | sequence, the occurrence of a \verb|\| `obviously' begins a new instruction, and therefore indicates the end of the previous one. There is an advantage to using the shorthand, which is not readily apparent: the kerning information is used. Once we group the instructions, the kerns\index{kern} between characters are ignored. While it is unlikely that you will notice this in many situations (especially if you are not familiar with French or German typography), it is perhaps an encouragement to use the shorthand. There is always some pleasure to gained from the feeling that you are somehow doing things `right'. In Table~\ref{tnatchar} the two `inverted' symbols were included as a sort of national character. These are a little odd, and have been known to cause confusion from time to time. If you are careless where you type spaces and confuse your opening and closing quotes, you could end up with something like this dreadfully forced example: \begin{verbatim} ....oh no!`he said quite emphatically.... \end{verbatim} These two national characters need not be enclosed in braces, since they are not standard \LaTeX\ instructions, in the sense that they do not begin with the backslash. They work rather differently. On the other hand, feel free to put braces in if you want. Nothing untoward will occur. \LaTeX\ also has lots of diacriticals\index{diacriticals} (Lamport refers to them as accents\index{accents|see{diacriticals}}, but some are not). The list is given in Table~\ref{tdiacrit}. \begin{table*} \begin{center} \begin{tabular}{|cll|} \hline \verb|\`| & grave & \verb|\`{e}| gives \`e \\ \verb|\'| & acute or aigu & \verb|\'{e}| gives \'e \\ \verb|\^| & circumflex or hat & \verb|\^{o}| gives \^o \\ \verb|\v| & inverted circumflex (h\'a\v cek accent) & \verb|\v{c}| gives \v c \\ \verb|\u| & breve & \verb|\u{o}| gives \u o \\ \verb|\=| & macron, long vowel &\verb|\={u}| gives \= u \\ \verb|\"| & umlaut or dieresis & \verb|\"{u}| gives \" u \\ \verb|\H| & Hungarian umlaut & \verb|\H{o}| gives \H o \\ \verb|\~| & tilde & \verb|\~{n}| gives \~{n}\\ \verb|\.| & dot accent & \verb|\.{y}| gives \.y \\ \verb|\t| & tie & \verb|\t{oo}| gives \t{oo}\\ \verb|\c| & cedilla & \verb|\c{c}| gives \c c \\ \verb|\d| & dot under & \verb|\d{d}| gives \d d \\ \verb|\b| & bar under & \verb|\b{a}| gives \b a \\ %\verb|\r| & circle over (ring) & \verb|\r{u}| gives \r{u} \\ \hline \multicolumn{3}{l}{and although not diacriticals, we should mention}\\ \hline \verb|\i| & dotless i & \verb|\i| gives \i\\ \verb|\j| & dotless j & \verb|\j| gives \j \\ \hline \end{tabular} \end{center} \caption{the diacriticals\index{diacriticals}}\label{tdiacrit} \end{table*} By and large, the instructions are fairly logically named. In order to get accents over i and j, you really ought to take the dot off first. \LaTeX\ supports a dotless i and j, provided by \verb|\i| and \verb|\j|. These allow you to do things like \^{\i} (from \verb|\^{\i}|), or even \t\i\j, (from \verb:\t{\i\j}:), should you ever find a reason to do so. The general rule with all these sequences is -- accent first, then letter. At first this sounds counter intuitive, after all, we say `e-acute', or `o-circumflex'. None of these accents are really `fundamental' to \LaTeX, in the sense that they are all created in much the same way, a way that is accessible. If we knew enough we could even create our own diacriticals. For example, Polish has a \r{u} character, where the `ring' can be defined as a diacritical symbol. \begin{question} In order to demonstrate your skills with these fancy fripperies, try setting some of the following: \begin{quote} \upshape Hs\"an Tsang, Tath\=agata, \'S\=akyamuni, Vai\'sravana; Pi-ma-w\^en, Li Y\"uan-chi, \=Ananda K\' a\'syapa, Manju\'sr\=\i; V\"ain\"am\"oinen, \"Aij\"o, V\"olusp\'a; Anne Bront\"e, Honor\'e de Balzac, Fran\c cois Rabelais; na\"\i ve, r\'egime, fa\c cade, man\oe uvre, encyclop\ae dia; \AE gean Sea, Ch\^ateau d'If, Gda\'nsk, \"O\H o\c z, N\^\i mes. \end{quote} \begin{quotation} \upshape Zde se v\v semo\v zn\v e sna\v z\'\i\ m\v e p\v reluvit, abych je\v st\v e n\v ekolik m\v es\'\i a napsal je\v st\v e jednu oper\.{u}. Hay\i r! \.I\c s \"oyle de\u gil. B\"uy\"u\u g\"u k\"u\c c\"u\u g\"une takilmay\i\ pek severdi. Ce f\^ut d'ores et d\'ej\`a une id\'ee d\'eg\'en\'er\'ee et ambig\"ue. \end{quotation} \end{question} If you were normally writing in a language which made use of diacritical marks or national characters, you would find this all rather tedious. You would likely have a keyboard which had characters like (for example) \'e, \aa, \ss, and so on. Having to insert the special \LaTeX\ sequences would, at the very least, be error-prone. It is possible to tailor \LaTeX\ to take account of this, so that when you type the single character \'e, \LaTeX\ interprets it correctly. Later we'll look at some \LaTeX\ `packages' which simplify this task. There should be added benefits, since it will ensure that the kerning information is also handled correctly; and the hyphenation\index{hyphen}\index{hyphenation} should also be correct for the language you are using. The default `language', which \LaTeX\ assumes it is using is American English. Not only does this mean that other languages are hyphenated in a rather haphazard way, it also results in words with diacriticals generated through the instructions of Table~\ref{tdiacrit} being hyphenated even less well (no hyphenation after the first diacritic). This is a deliberate gloss which does not quite tell the truth. We'll revisit this topic later. \section{Hy-phen-a-tion} \LaTeX\ chooses hyphenation points by a hybrid method. It uses an algorithmic technique which is supplemented by a small dictionary of `exceptions'. The algorithm it uses can be taught to recognise the appropriate hyphenation points for other languages (see Appendix~H of~\cite{DEK84}). A large number of alternatives have been collected or developed by Johannes Braams~\cite{JB91} and are available as a `package' to \LaTeX\ users. If you really have to hyphenate German, or Esperanto, or even English now, refer to Chapter~9 of the \textsl{Companion}. You can hyphenate words yourself as they occur, by inserting a special command which indicates a \emph{potential} or \emph{discretionary}. For example, in \verb|Tyr\-rhenian| the \verb|\-| indicates a discretionary hyphen. Note that this instruction is not followed by a space. Declaring each potential hyphenation is tedious, and one alternative is to declare the potential hyphenations as: \begin{verbatim} \hyphenation{Tyr-rhenian manu-script manu-scripts} \end{verbatim} That is, simply a list of hyphenated words, separated by spaces. This has a global effect, since what happens here is that these words are added (temporarily) to \LaTeX's `exception dictionary'. As noted above, \TeX/\LaTeX{} hyphenates by algorithm, but there is a small dictionary of exceptions. The best place for this hyphenation instruction is in the preamble. Note that \LaTeX\ will not realize that \verb|manu-scripts| is merely a regularly formed plural of \verb|manuscript|. Similarly, this mechanism can know nothing about any other regularly formed inflections. On the other hand, the standard hyphenation can cope with many inflections. Explicitly declaring hyphenation points will not help words which already contain hyphens. While \mbox{`pricking'} by itself can be hyphenated to \mbox{`prick-ing'}, `\mbox{pin-pricking'} is not hyphenated to \mbox{`pin-prick-ing}'. If we were to say \begin{verbatim} \hyphenation{pin-prick-ing} \end{verbatim} we could easily end up with \mbox{`pinpricking'}. In this case we would have to use \verb|pin-prick\-ing| throughout the document. \begin{question} One way to see the effect of hyphenation is to reduce the number of words per line. The easiest way you know so far is to use a large font size, or to use the \verb+twocolumn+ option. Choose some text and do so. \end{question}