Comment by sublinear
12 hours ago
I find serifs distracting and "noisy" regardless of text length. Their modern use restricted to headings, wordmarks, etc. makes more sense to me.
We ended up with sans-serif for body text on the web because it's more legibile on low-res screens. I still can't believe that didn't settle the argument right there and then decades ago. If something is easier to read on those crappy old screens, it's easier to read everywhere.
Not all "sans-serifs" are similar, i.e. they cannot be put in a single class opposed to serifs.
The sans-serifs designed before WWII have evolved towards a maximum simplification of the letter forms, eliminating not only the serifs, but also the contrast between thin lines and thick lines and also some other details of the shapes that differentiate the letters. The extreme simplification was achieved in some of the "geometric" sans-serifs, where most of the letter shapes were simplified into circles and straight lines. The motivation for shape simplification was that in the beginning the sans serifs were intended for advertising, i.e. for printing on cheap paper with cheap devices, which could not reproduce fine details. More over such texts were intended to be readable from great distances, from where fine details cannot be perceived.
After WWII, some sans-serifs have just reproduced with negligible changes some of the earlier sans-serifs, e.g. Helvetica and the like.
Most of these sans-serifs, like Arial, are extremely bad for use in computers, because they have a lot of ambiguous letters and digits. That mattered less for English texts, which have high redundancy, allowing the guessing of the intended letters, but in computers there are a lot of abbreviations, keywords, identifiers, expressions and other kinds of character strings with low redundancy, where it may be difficult or impossible to recognize the intended letters or digits.
However, there are also several kinds of very different sans-serifs among those designed after WWII, which attempt to remove the disadvantages of the classic sans-serifs.
One kind is the sans-serifs that have only the goal to remove all ambiguities in letters, digits and other symbols, like most of the monospace typefaces intended for programming.
Another kind, besides removing ambiguity also reintroduces some of the letter shapes from serifs, which have been simplified in the classic sans-serifs, e.g. double-storey "g" and "a", old-style digits, distinct shapes for the italic variant of a font, etc. These slightly more complex shapes increase the distinctness of the letters and digits, making them more easily recognizable. An example of such a typeface is FF Meta.
Finally, there is another kind of sans-serifs that also reintroduces the contrast between thin lines and thick lines. Moreover, among these there are typefaces, e.g. Optima or Palatino Sans, which (on displays with high enough resolution) produce an optical effect similar to serifs, which is achieved not with serifs but by replacing the straight side edges of the lines with slightly concave edges.
In my opinion these kinds of sans-serifs that are intermediate between classic serifs and classic sans-serifs are superior to both classic types, especially for computer use.
Despite being less affected by low screen resolutions, the classic sans-serifs like Helvetica, Arial and many others should never be used for anything, because of their ambiguous glyph shapes.
Serifs have non-ambiguous glyph shapes, but they are heavily distorted at low-screen resolutions or when seen from a distance, so they are also not recommended in many applications.
The good choice is in most cases to use one of the sans-serifs with non-ambiguous glyphs, or for a more beautiful text one of the typefaces inspired by FF Meta or one of those inspired by Optima. For example, my default typeface is Palatino Sans.
Any study about the legibility of serif and sans-serif typefaces that does not include a separate category for non-classic sans serifs is deeply flawed and its results are meaningless.
> Not all "sans-serifs" are similar
True, sure.
> i.e. they cannot be put in a single class opposed to serifs.
Not sure I understand you. Are you saying all serif typefaces are similar? There's old style, didone, slab serif. These differ more than the various sans-serif typefaces, which are really all quite similar ;)
The differences between various kinds of serif typefaces, which consist mostly in which is the width ratio between thin lines and thick lines and between serifs and thin lines, and in some small details of how the lines are terminated are much smaller than the difference between classic sans-serifs and some of the post-WWII sans-serifs, which use different shapes for some of the characters.
The differences between various kinds of serifs are hard to see on a low-resolution computer screen. The didone typefaces cannot be rendered correctly on most computer displays at body text sizes, where they are distorted by greatly reducing the contrast between thin lines and thick lines, which can be seen on printed paper.
At small sizes or from a distance one cannot distinguish the various kinds of serif typefaces, except those with slab serifs, where the serifs remain conspicuous. However, the slab serifs frequently look rather identical with a classic sans serif, except that big serifs are attached to the glyphs, whose only advantage it that they may remove some of the ambiguities, which is one of the reasons why many typefaces for programmers use sans-serif shapes for most characters, but slab-serif shapes for a few of them. (The other reason is specific for monospace typefaces, where slab serifs on the narrow characters ensure a more uniform average color of the text, i.e. a less variable ratio between white areas and black areas.)
On the other hand, the differences in character shapes between something like FF Meta or Palatino Sans or JetBrains Mono vs. Helvetica or Arial are extremely obvious and they matter a lot for legibility.
I agree with you that besides classic serifs from before the Napoleonic wars, classic sans-serifs from before WWII and non-classic sans-serifs, in a legibility study one could include a 4th category, slab serif typefaces, which are distinctive enough from older serif typefaces.
However slab serifs do not have any supporters who claim that they are good for anything else than their original purpose, which was for advertising, the same as for sans-serifs. So nothing is lost by not adding them to a legibility study. On the other hand, not considering non-classic sans-serifs is a methodological mistake, because they are frequently more legible than either classic sans-serifs or classic serifs, so omitting them is like organizing a competition only for known losers.