Font overload in academic documents

Question

I'm a researcher in cryptography and regularly read and write acedemic papers which are, to say the least, not objects of great beauty. Recently I've noticed that the guideline not to use more than two or maybe three fonts in a document is regularly flouted and wonder how much this has to do with it.

Here's an example of mine (content is of course irrelevant):

The done thing seems to be to have different fonts for different categories of things:

• Serif for body text, same in bold for headings, italics for emphasis.
• Sans-Serif or monospace for algorithm names (the "KeyGen" etc. in the example).
• Italics (but in a slightly different font I think) for inline math.
• sans-serif, bold or small caps for names of security notions (small caps in the example above).
• "blackboard bold", "calligraphic", "fraktur" etc. (the different options available in LaTeX) for different classes of things like collections of algorithms, participants in a protocol.
• and so on ...

My questions are:

• The above example looks a bit chaotic to me. Is it mainly the number of fonts, or am I missing any other key design principle (like not using fonts that go well together or bad spacing)?
• Is there any point from a design perspective in using different fonts in a techinical or academic document to denote different categories of things or would it be better to just to use the same font for all?
• The argument for "one font per category" seems to be something like making it easier for the reader to recognise at a glancce what kind of thing I'm referring to - is there any justification for this? Or a better way to achieve the same?
• I've read this GD topic, any more pointers to how I can make my academic papers look better?

Answer 1

As you identify, there are a number of issues and they all stem from your implementation of (or possibly how you're using) TeX.

For a bald list, I cite

• the use of a Scotch Roman face
• poor letter spacing
• lines too closely-spaced
• poor mixing of fonts (sans and script)

TeX's Scotch Roman face is old-fashioned and fussy. All those serifs! It's this fussiness which needs a greater leading. It's not helped by poor spacing between letters: some are a great deal more cramped than others. Look at the word ciphertext. The small caps are not real small caps: they are simply reduced in size keeping the proportions the same. Small-caps don't do that. The sans and script fonts are a poor mix, because they do not have the same optical density as the main font. It's fine using sans for the algorithm names, but they don't have to stand out in that context.

Here's a reset example of your text. The main font is Calluna, and the software I used can't cope with fi or ffi ligatures, and I had a limited range of fonts available for the symbols. Nor does it do small-caps natively: I set "small-caps" and then increased their width to 115% so that they looked like real small caps (it keeps the main strokes the same width). The end result is an even distribution of ink.

Leading is increased as that helps with the mixing of symbols and retaining readability. The italics are well-designed and don't alter the overall blackness of their text. The sans words are set in Calluna Sans Light. This is derived from Calluna so the letter-shapes are very similar (and the same size!) and using the Light variant again keeps the optical density even so they don't stick out. This helps your head-words to be headings.

There would certainly be scope to add more space between paragraphs as others have suggested, but if this is a list of small items it might break it up too much. Allowing more light into the page with increased leading generally would be sufficient for a list; larger, "real" paragraphs might benefit from an extra half-line of whitespace.

Answer 2

The principle of different fonts for different categories of elements is not incorrect, however those fonts need to coexist in a more or less harmonic way.

This is my very personal opinion: While I don't think the individual letters are a problem (you do need two types of Gs, for example, and these are both serif and look like they belong to the same extended family), I do find the KeyGen sans-serif a bit strange. It serves a good purpose: It differentiates those words from the rest, but it does so in an unnecessarily 'exaggerated' way. I wouldn't normally combine serifs and sans-serifs in the same line/paragraph. While both styles are arguably equally good for legibility (see this question), mixing them can be more... problematic, and this is why I think this happens:

While legibility is about easily recognizing individual letters or words, readability has more to do with the optimum arrangement and layout of whole bodies of text. Your fonts are legible, but are your paragraphs readable? You have a particularly complex example, so it would be quite difficult to resign some variability, but do you need to have so many fonts coexisting? I would ask: Is all of your information equally important, so you need to give each element a distinctive look? The answer might be yes, I have no idea what your definitions say, but I can imagine you can't loose the following: Different single letters, uppercase acronyms, italic definitions (maybe?) and distinguishable algorithms.

That's why I would focus on the readability. Readability is an interaction of several conditions, so you just have to consider other elements such as line length, line spacing, size of fonts, width of font, familiarity with the font, paragraph size, etc. Unless you have space constrains, you don't need to pack everything so tightly. Instead of trying to find fonts that look better together (except for that sans-serif!), I'd try to work on the whitespace, especially the space between paragraphs.

Answer 3

It's difficult to say really.

Many tech docs require special fonts to convey special symbols. If a font is needed for an equation on page XX that same font should be used for equations throughout. I suspect your sample suffers from this to a degree. It's unfortunate that the equation font seems to also be an italicized font. If an equation typeface without italics were used, it would go a long way to improving readability to me.

The italics could also be used specifically for names. Genome names are always italicized. Although in your sample it appears as though italics are used for names as well as for emphasis. That creates some confusion to me. Setting the italics at 1pt larger would help. This is often the case for italics. They tend to visually appear smaller than surrounding text (even if the font is the same). By setting italics 1pt larger, you circumvent that visual difference. And using some other form of emphasis alteration may be beneficial. However, other weights of the typeface may not be available. You could use underline, but, well... I just hate those and avoid them if at all possible.

The different typefaces for "KeyGen" etc. would appear to be in order to emphasize those items without using italics or boldface fonts. That does make some sense.

I personally would not indent the first line. If anything I'd hang the first line rather than indent it.

And I agree with user568458's comment, space between paragraphs would help a great deal.

Answer 4

following a style guide for documents like this will work wonders here. i'm actually really surprised no one's mention the Chicago Manual of Style, or another yet
http://en.wikipedia.org/wiki/The_Chicago_Manual_of_Style