My guess is that sans wins because many sites are using a sans web font for headlines and Georgia for copy. And because sans seems to be more popular for corporate identities right now.
In general, sans-serifs have been more popular for a very long time. (See also: http://www.myfonts.com/bestsellers/). Websites in particular seem to love the safe sans (webfont) headlines paired with system font (usually serif) body. Nothing unsurprising here.
Having been the one who compiled the list, I concur with a lot of the above comments.
Mind you, when I look at the next 10 or 20 most popular typefaces on WebINK, there are plenty of serif faces cropping up—though still less than half of that next group.
This amazing difference is on one site, a site that has not a single serif typeface designed for the web, all having being repurposed from print... Maybe coincidence, but I know another web type site, where the popularity of serif vs sans in about even, due in part to the inclusion of good serif text faces for the web, being popular enough to be in the top ten.
Fonts.com just released some new serif fonts just for this purpose recently, and it should be interesting to see how their font popularity changes. But the site I'm referring to is webtype.com/fonts/ then sort by popularity, where the order you find now started to form late last year.
What and where are the new web serifs on fonts.com? Are they the same ones MT developed for e-readers? I tried finding them on Fonts.com but Monotype is still incapable of promoting anything made by a living member of their type team.
Completely agree with David that there are many "just converted" web fonts and not enough tailored designs.
In general, converted serif fonts will not look good for body text on the web, unless you use then bigger than 20px. And web designers does not like to set the body text so big (just in case the webfonts fail to load and the fallback font is shown)
Tailored designs will do look good when set at standard sizes (13 to 16px range), and they can become pretty popular pretty fast. For example Even's Merriweather and Sol's Bitter. Both are ranking pretty high on the GWF stats, having about 53 million views per week, each one.
Just "converting" is fast... while tailoring require extra work. So I applaud anyone taking the time and doing the extra work to tailor new webfonts.
The e-text fonts are a good step in the right direction, however to me, they looks like a halfway optimization. Why? Because when set at 16px the standard MT Baskerville will produce a 7 pixels x-height. eText Baskerville will increase it only by 1, producing a timid 8 pixel x-height. While most tailored web fonts from the top foundries (Webtype, H&FJ, Typoteque, etc..) will produce a 9 pixels x-height, that seems to be the "de facto" standard.
Of course there more to it than making the fonts look bigger, but of all the optimizations (spacing, contrast, hinting, etc) that seems to be the more noticeable one. Or the one that makes the bigger difference.
Libre Baskerville (our tailored web version of ATF Baskerville) will also produce a 9 pixels x-height. And it's becoming popular pretty fast. The stats shows that it was requested 16 million times just last week, at it's increasing by 36% each week. Being among some of the fastest growing fonts on the directory.
In conclusion, I believe that there are still room for Serif fonts to become popular on the web, but they will need to be tailored for this new medium.
Because of the successful adoption of Libre Baskerville, we also have been working on Libre Caslon Text and Display, tailoring a Caslon for body web text, and another one for headers.
I apologise beforehand for this post: In the link mentioned, body copy is in sans-serif, the heading in serif. Does that not subvert the message just so neatly?
Because when set at 16px the standard MT Baskerville will produce a 7 pixels x-height.
Presuming that you are deliberately distinguishing here between px and pixels, surely the accuracy of this statement depends on the resolution, since px is no longer a device unit?
It could be, but if that is the case it is really old. The execution of several of the RE serif faces followed the general proportions of 6-7 point ATF and Linotype serif text faces, which had enlarged lower cases, not just x height. I.e. they are really wide and openly spaced as well. There is a picture of one of these on our mini site, fontbureau.com/readingedge.
The two most unbullshitic things about masters for small sizes is that they appear larger small and appear confused large, and second, when in the presence of low resolution, they gather more pixels for any given pixel per em than your average master.
As for the Pabloian 7, 8 or 9 px x ht, I would hesitate to add an absolute l.c. pixel ht to a font spec for small masters to avoid forcing all small masters into a mold. 7, an odd number, can sometimes prove more readable than 8, if 9 is just too big for some kind of text style, but we will see as more people do them.
And of course, we will see how these effect web font popularity lists as they each have their own offerings of specialty text faces... I assume...
It is low-res, yes, but is it necessarily small...
I designed RE small-first, low-res second. It's tricky but fun.
John:
small size-specific type scaled to display size
You said it better than me, but ya, is that not the reason there are 1,000's of display faces for every text face? It's much easier to decide what the text might look like, but when it scales up, people start to argue, and before you know it this year's whole top ten is derived from a single ancient text face, or something.
> a site that has not a single serif typeface designed for the web
That is not quite literally true. We have a few of those, as well as a bunch more that were designed for both screen and print, and superhinted for screen use.
That said, David Berlow’s approach to original typefaces for screen use has produced fabulous results.
I wonder when that chart was last changed, Daniel. It’s pretty out-of-date. Webtype currently carries fonts from 10 foundries, for instance, and Typekit now offers desktop use.
Comments
Mind you, when I look at the next 10 or 20 most popular typefaces on WebINK, there are plenty of serif faces cropping up—though still less than half of that next group.
This is not to say sans is not useful.
In general, converted serif fonts will not look good for body text on the web, unless you use then bigger than 20px. And web designers does not like to set the body text so big (just in case the webfonts fail to load and the fallback font is shown)
Tailored designs will do look good when set at standard sizes (13 to 16px range), and they can become pretty popular pretty fast. For example Even's Merriweather and Sol's Bitter. Both are ranking pretty high on the GWF stats, having about 53 million views per week, each one.
Just "converting" is fast... while tailoring require extra work. So I applaud anyone taking the time and doing the extra work to tailor new webfonts.
The e-text fonts are a good step in the right direction, however to me, they looks like a halfway optimization. Why?
Because when set at 16px the standard MT Baskerville will produce a 7 pixels x-height.
eText Baskerville will increase it only by 1, producing a timid 8 pixel x-height.
While most tailored web fonts from the top foundries (Webtype, H&FJ, Typoteque, etc..) will produce a 9 pixels x-height, that seems to be the "de facto" standard.
Of course there more to it than making the fonts look bigger, but of all the optimizations (spacing, contrast, hinting, etc) that seems to be the more noticeable one. Or the one that makes the bigger difference.
Libre Baskerville (our tailored web version of ATF Baskerville) will also produce a 9 pixels x-height. And it's becoming popular pretty fast. The stats shows that it was requested 16 million times just last week, at it's increasing by 36% each week. Being among some of the fastest growing fonts on the directory.
In conclusion, I believe that there are still room for Serif fonts to become popular on the web, but they will need to be tailored for this new medium.
Because of the successful adoption of Libre Baskerville, we also have been working on Libre Caslon Text and Display, tailoring a Caslon for body web text, and another one for headers.
The two most unbullshitic things about masters for small sizes is that they appear larger small and appear confused large, and second, when in the presence of low resolution, they gather more pixels for any given pixel per em than your average master.
As for the Pabloian 7, 8 or 9 px x ht, I would hesitate to add an absolute l.c. pixel ht to a font spec for small masters to avoid forcing all small masters into a mold. 7, an odd number, can sometimes prove more readable than 8, if 9 is just too big for some kind of text style, but we will see as more people do them.
And of course, we will see how these effect web font popularity lists as they each have their own offerings of specialty text faces... I assume...
And will look like a horse if used big
John: You said it better than me, but ya, is that not the reason there are 1,000's of display faces for every text face? It's much easier to decide what the text might look like, but when it scales up, people start to argue, and before you know it this year's whole top ten is derived from a single ancient text face, or something.
That is not quite literally true. We have a few of those, as well as a bunch more that were designed for both screen and print, and superhinted for screen use.
That said, David Berlow’s approach to original typefaces for screen use has produced fabulous results.