On the Typography of Flight-Deck Documentation (1992) [pdf] | Hacker News Reader

On the Typography of Flight-Deck Documentation (1992) [pdf]

ti.arc.nasa.gov

138 pointsdanso27 comments

As a cautionary note, a few of the claims in this document don’t stand up well in light of more recent research. If you’re interested in the technical aspects of typography and how humans really read text, you might like to try these more up-to-date starting points:

Alex Poole’s “Which Are More Legible: Serif or Sans Serif Typefaces?”:

http://alexpoole.info/blog/which-are-more-legible-serif-or-s...

Kevin Larson’s “The Science of Word Recognition”:

http://www.microsoft.com/typography/ctfonts/wordrecognition....

In particular, if you’re about to comment on how serifs or sans-serifs are better for X, or to explain how we read by following the serifs along a line or by recognising word shapes, I strongly encourage you to study the above first.

There are unfortunately a lot of urban legends in typography that simply aren’t supported by the evidence. Some of them came about because of studies that have since been shown to be flawed in methodology or even based on outright falsified data. We help no-one by perpetuating the myths.

Very interesting, and many thanks. One of the main takeaway from this should be (at least, it was for me):

It is of course possible that serifs or the lack of them have an effect on legibility, but it is very likely that they are so peripheral to the reading process that this effect is not even worth measuring (Lund, 1999)

While there is a footnote or two saying sans is better generally in the article, reading the full text actually argues quite a bit for each side. For example, legibility problems cited with sans are: " The sources of similarity between the characters of modern sans-serif fonts are: 1. The standardized or modular appearance of the letters (“P,” “R”). 2. The effect of mirror images between the upper and lower part of the character (“E,” “B,” “D”). 3. The use of equal radius for different letters (“G,” “O,” “C”) (Craig, 1980; Cheetham and Grimbly, 1964). "

I really dislike how sans-serif fonts conflate the lowercase "L" and the uppercase "i". This is a big problem in mail clients where someone can create email addresses that look identical to established addresses (e.g. Iarry@google.com).

I'm surprised that the sans-serif is here claimed to be more readable "always". In low resolution circumstances and where single letters are to be recognized (e.g. abbreviations common in short technical instructions) I believe it's so. But I also know that if you'd give me book printed in sans-serif, I wouldn't want to read it: whenever I try, I find it very stressful. I don't have the explanation why though.

Taste?

If it's taste, it's at least a taste shared by (almost) all typesetters.

That would be called "convention".

It's a similar reason for the advantage of lowercase over uppercase: the overall shape of the letter is easier to recognize if it's "bumpier," as in serif fonts.

The serifs create visual guidance for your eye as it moves down the line.

^This. As I understand it, serifs form a horizontal channel that guide the reader through long passages of text (ie. in books, blog posts). For long lists printed vertically, sans works better because there are no serifs to interrupt the eye path. This is why bullets work well - they guide the eye vertically down the page.

Induce a little stress first - say, put yourself in a cockpit with a few hundred lives to protect - then try a reading experiment involving both styles. I think you'll find that in high-stress environments, the less noise the better.

What is noise vs. signal is the issue at stake here. The pro-serif people will argue that the serifs provide extra features on the glyphs, so they might provide more signal. Nevertheless my intuition says you're right.

One possibility is that, for books, the combination of low font size and bad printing techniques made serif types more readable.

For example, a P or an F without a serif may lose a stem or an arm (because the ink didn't spread uniformly) . In a sans-serif font you would have the P or F without the lower part, but by adding the serif you have a place where more ink can get, reinforcing the letter format.

See something typed on a typewriter (with a bad tape) or an old book and you'll see what I mean.

This is a good read. I ran into an interesting meta-problem on page 12 (numbered page 9 on the page itself).

What do you call the diagram labeling?

The single world label pointing at a feature with a line could be a 'callout' perhaps.

What is the word-in/around-arrow pointing to two vertical lines? A range indicator? It looks like: |<--label-->|

I've also seen a variation where the word points to a line covering the range with bracketed ends, somewhat like: label -[ feature

It's a similar problem to "what do I call that operator?" when you don't know what it is, like ?: being "the ternary" operator (or more precisely the conditional operator)

The best description I could find was "dimension line," but all the examples have numeric measurements rather than descriptive text.

In technical drawing. "Dimension line" is correct. More specifically a "dimension line indicating a feature".

A single character (usually in a "bubble" or circle) would be a callout. One or more words in combination with an extension-line and an arrow-head would be a "note".

On a related note, does anyone know of any studies of monospace vs. proportional fonts on reading speed/comprehension? I can find a number of examples for reading on older, low-DPI computer screens, but not much for high DPI screens or high-resolution print.

You make an interesting point. Small observation: Traditionally, even the monospaced fonts from the 70s-90s on terminals and older machines, were of the kerned variety (serif-like appearance).

If you kern a monospace font it's no longer monospace (fixed width), so that seems unlikely. Maybe you mean something else?

Yes, I wanted to say "stroked", as in:

>> even the monospaced fonts from the 70s-90s on terminals and older machines, were of the stroked variety (serif-like appearance)

Good catch.

I am also one of those (few?) here, that are very partial to serif over Sans-serif. My browser uses Serif by default, and all my documents use Serif for body (and sans for headings if people want heading to look different)

I too find serif fonts more readable, without necessarily being able to give a "scientific", "why" it is so. As another commenter mentioned the "l" and uppercase "I" is one frequent area of confusion with sans. And maybe serifs with their kernings give a certain visual guidance as the eyes bounce off the words...

Having said that I've definitely given "Times New Roman" the boot. I've never found if aesthetically pleasing to the various other font names I used to see mentioned in all those books I read.. Now with proliferation of FOSS fonts like Linux Libertine, PT Serif and so many other more eminently readable fonts, Times New Roman is a distant memory to me :)

It is a shame that many new OReilly books are adopting Sans fonts for their paper version as well. I for one am very displeased with this as there are times when I would like to buy a paper version for my bookshelf. Fine, do it for the ebook version, but not for the paper one, please... :(

A standard I have to refer to often is ISO 128 / AS 1100.101 (1992). Technical Drawing - General Principles. Which has a lot to say on the issue of Letters and Numerals in part 4.1, I've added this as it's mildly similar, since the readability of technical drawings is also critical, it was widely read in a technical industry and that it was last revised in the same year as the OP document.

It should be remembered that this standard was last revised in 1992; As a consequence, computer generated text or "Machine-made characters" don't really get a look in, although degradation through reproduction (for micro-film) is noted as an area of concern. It favours sans-serif fonts that are "uniform and capable of being produced at reasonable speed by hand" and that "Upper-Case letters should be used. Lower-case letters shall be used for conventional signs and symbols normally requiring such characters, e.g. mm, kg, kPa."

The recommended stencil characters are outlined as Gothic and ISO 3098/1 Type B (aka "Font 27"). The latter being preferred by every engineering firm and operator I have come across in their published standards.

I found this very useful and have been looking specifically on documentation on checklists and useability of certain fonts!

It's nice to see the author play with fonts to illustrate his points (3.2, 3.3, 3.5 are most obvious).

As someone who had to use checklists many times in the air, I appreciate how much research have gone into making checklists legible.

Don't use checklists that have been photocopied 10 times.

Reply on news.ycombinator.com