A Rant About Fonts

Fonts are interesting. Fonts are fun. Fonts are artistic. Font design has a long and interesting history. The choice of a font depends on many factors; a font suitable for use in an advertisement to suggest a whimsical juvenility is not suitable for use in an engineering working group meeting with a client. There is no such thing as a univerally “good” font or “bad” font.

Some fonts, though, have some characteristics which make them crappy for almost all purposes, and some of these are being used all over the place.

Fonts for Everyday Reading

Usually^[1], text is used for plain old reading. For this purpose, readability is a key consideration in choice of font. Recognition of characters and words should be rapid, smooth, and automatic, so that the reader can pay attention to the content and meaning of the text. The neural processes involved in this are (obviously) a bit complicated, and there are some interesting side effects^[2] of these processes that make font selection challenging.

There are a few simple rules which should be followed. One of these is so obvious that it seems like it shouldn't be necessary to even mention: the reader should be able to distinguish different letters. Different letters should look different.

One would think this is so obvious that anyone violating it would be loudly mocked, even if they were your boss. Yet very large organizations and corporations choose such fonts as part of their official “look-and-feel” (branding, style), and very large software corporations who create widely used authoring^[3] and presentation software choose such fonts as the default. I find this baffling (and aggravating when forced to follow the aforementioned corporate branding rules).

The offending (and offensive) fonts to which I refer are those of the sans-serif variety. The most commonly used of these is probably Arial; others include (in alphabetical order) Futura, Geneva, Arial's parent Helvetica, and Lucida Grande. (There is at least one exception to this, Trebuchet, which I see sometimes on the internet; while technically sans-serif, it does include an important adjustment to one character so that it satisfies the rule of distinguishability.)

Why, you might ask, is this so important, given all the interpolation and error-correction done by our neural processes? There are (at least) three reasons: interpolation is not always perfect; learners of the language (children and non-native speakers) will have difficulty; and sometimes the indistinguishability of characters interrupts the flow of reading, thereby distracting from content (this may be related to the first reason). Here is an example^[4]:

• There are several techniques which can be used to clarify a complicated point, including:
  I.   Using more words;
  II.  Mentioning analogies;
  III. Illustrating by example.

 I'm pretty sure I'm not the only one who coughed, choked, or perhaps even threw up a little when I hit the third point. “1: using more words; 2: mentioning analogies; 3: 3-ustrating… umm… huh?” Even if you didn't even hickup on that, remember that those less familiar with the language will have difficulties.

I suppose, having listed three ways to clarify a point, I should use all three, so here's an analogy. Imagine baking a cake, and the recipe says “A cup of white crystal stuff; A quarter teaspoon of white crystal stuff”. If you've had enough experience in the kitchen, you'll probably recognize that the first is sugar and the second is salt. This may not be so obvious to the beginner, and the consequences of an error will almost certainly make you choke and cough.

I have seen claims made that Arial has been found to be the best font to use during presentations. This makes me wonder if those making the claim have attended any actual presentations which involved the use of polysyllabic words; alternatively, perhaps they think that comprehension is unimportant and content should be hidden (I have encountered people who think this).

Three obvious sets of similar characters are {Il1} (uppercase i, lowercase L, and the digit one), {O0} (uppercase “oh” and the digit zero), and {S5} (uppercase “ess” and the digit five). The first of these is the most important for everyday text; all three are important in technical work, including writing code and scientific work where abbreviations are used. It is unlikely that you will encounter “O0” or “0O” in everyday text. (Trebuchet includes a little “hook” at the base of the lowercase “ell” to distinguish it from uppercase “i”: {Il1}.)

There are lots of other rules or guidelines that should be followed to enhance readability, and which are frequently violated. These involve issues like spacing or kerning^[5], relative character size, consistency of line thickness and curvature, and so on. It would please me, though, to see the offensive fonts I previously mentioned banished to unlabelled drawers in the dusty basements of typeface museums^[6].

Some History and stuff

Back in the good old days, people used paper and pen, the latter of which used ink, which is a liquid. The tip of the pen was a thin blade which would be dipped in ink, and then placed on the paper and dragged across, thereby leaving a line of ink. The width of the line could be modulated by changing the angle of the pen. The problem with this is that the ink does not immediately begin a regular flow upon first contact of pen and paper. When you put the thin blade on the paper and start dragging it across, you get an uneven start to the line, depending on how much ink there is on the pen, how much pressure is applied, etc.

However, if you initially move the blade in its thin direction (like you would be slicing the paper, if the blade was sharper), the gaps are immediately filled and the regular ink flow begins, at which point you can then move in a direction transverse to the blade, leaving a nice wide line. This initial thin stroke seems to be the historical origin of the serif (see, e.g., [D], pp. 22-24); later refinements were motivated by more æsthetic considerations, including the development of much more elaborate or decorative fonts.

Then along came mechanical innovations like moveable type and the printing press, improvements in ink, the ball-point pen, and other stuff that I don't know enough about, and the serif became a strictly æsthetic option. At some point, simplicity became fashionable, leading to fonts such as Geneva and Helvetica and, because of copyright considerations, Helvetica's spawn Arial.

So, with respect to the mechanical production of type, serifs have become irrelevant (except for those who still practice the old pen techniques, including calligraphers) — so choice of serif becomes a question of readability.

A few years ago, a major software company introduced six fonts whose names start with the letter “C” and made them the default for their authorship, presentation, and electronic communication software. I probably shouldn't comment on this, other than noting that the word “crayon” also starts with the letter “C”, and that is not the only relation between those fonts and a crayon. (Perhaps I've said too much.)

Notes
1. ^ Based on informal obvservations of raw letter count when you look around or browse the internet.
2. ^ There are lots of fun little exercises and examples (and actual studies) that show how much our brain interpolates when reading; for example, it has been said that the ends of the word and the middle of the word play slightly different roles in word recognition, at least for short words. As I recall, these studies have yielded somewhat inconsistent results, so rather than look for citations, I'm being lazy and basing some of this blog post on anecdotes.
3. ^ I use the word loosely; such software frequently presents more of a hindrance than an aid to actual authorship. The term “word processor” is often used but, in my opinion, is a misnomer.
4. ^ I made up this example, it is not a quote from anywhere. It assumes your browser can display Arial, or another sans-serif font.
5. ^ Look closely at this example, previously encountered in this blog post: “stuff”. Specifically, look at how the closing quote character overlaps with the preceding letter. Correcting this involves kerning, i.e., adjusting the position of a character depending on what characters are nearby. One could try to correct this example using   which results in this: “stuff ” but that space is too big. TeX and METAFONT do a good job of automatically taking care of all this (assuming that the font designer paid attention!).
6. ^ Such as the Hamilton Wood Type & Printing Museum. I learned about it from this film that I saw on PBS.
7. ^ The correct spelling is “Mediæval”, but if you don't have the “æ” ligature, either “Medieval” or “Mediaeval” is acceptable; I prefer the latter. (It's pronounced “Med-eee-Eye-vul”, not “mid-Evil”.)

Bibliography
• [D] Mark Drogin, “Medieval^[7] Calligraphy: Its history and Technique”, Dover reprint, 1989.