It looks like you're new here. Sign in or register to get started.
André G. Isaak said:
I’ve always been a bit skeptical of the etaoni... Letter frequencies in English vary considerably depending on whether types or tokens are considered as well as on the type of writing (e.g. academic writing will contain a much higher percentage of latinate forms). I suspect h and f (two letters with ascenders) have much higher frequencies than the above would indicate given that they occur in many common function words (the, this, of, for etc.)
By doing what you propose, "Oldstyle Numbers" and "Lining Figures" would be changed when you use them in the design software programs.
I was wondering if there was a way of doing this only by OpenType coding.
Besides which, it's not within the power of the font, or its designer, to take a feature that is off by default (per the OpenType spec), and force an app to turn it on in the app UI. What you can do is control what the default glyphs are.
(You can also put glyphs into GSUB features that are on by default or even required, such as 'rlig' 'rclt' or 'rvrn' ... but I don’t see any way in which such features would be relevant/helpful here. None of them would turn on 'onum' for you.)
Now I understand perfectly, thank you so much for your patience!
Because not all users will have access to, are comfortable with, or know how to use the alternate figures palette—and documents may require tabular setting.
An image previously provided by @Joe Elwell on TD:
BTW, I also see an advantage to making the "2" ascend (something Eric Gill did once in a while).
@André G. Isaak I just rediscovered this elaboration:
The center of gravity of virtually any Latin-script languages is much higher than it seems when you only look at the collection of letters, versus factoring in frequency. The Legros & Grant "uniglyph" (as I call it) is quite revealing, as I believe is my lateral-scaling visualization.
BTW the numeral forms are essentially the same, it's a question of how frequently clusters of numerals will distract the eye by sitting too low; and considering capital letters, it seems pretty impossible for them to ever really sit too high.
Personally, I don’t think this is a matter of general principle at all, but one of style. Since numerals are combined in arbitrary combinations that may include no ascending or descending elements, all ascending elements, or all descending elements, or any mixture, I think it is silly to worry about centre-of-gravity. It is perfectly possible for a text set in Didot types to include a number like 979—all descending—, while there will be a greater likelihood of numbers having multiple ascending numerals in this style, thereby undermining the whole point of ranging numerals, which is to avoid the all-caps like impact of massed lining numerals. It is also worth noting that while the Latin script involves more ascending than descending letters, and more frequently in text, all but one of those ascenders are simple straight stems, while the descender space is where the g lives, and having the 3 5 and 9 descend locates these shapes in the same space as the only descending letter bowls.
It’s been almost a decade since I last released a typeface that had conventional ranging numerals based around the proportions of lowercase x-height. So I hope I can avoid being accused of mere conventionalism in thinking Hrant overstates the case for the 18th Century French model and its descendants. They’re interesting, evocative of particular styles of typography, and—like the more common ranging numerals—look better in some combinations than in others. Which leads to another idea, which would be a font in which the ranging behaviour of the numerals varied contextually based on their sequence in a number, both in terms of balancing the variation in ascending and descending shapes, and in avoiding massed groups of either.
Calling it "French" is a shorthand, but a solid one because (as far as I know) it's where it started, and hasn't really been adopted elsewhere. And I think it ended in France only because of WW2. Calling the conventional one "Anglo" is more of a stretch, but I think it's better than "non-French". :-)
In a text face (where OS nums make sense) style is sillier than functionality, and readers can suffer from limitations imposed in favor of style. No reader will consciously notice the vertical position of OS numerals... but they can still suffer from the –unnecessarily– frequent disruption in the usage of the vertical space. And by the same logic, also not silly: making the descenders shorter than the ascenders... very much in spite of the "g" and "y". Like in Castoro. :->
Even within the realm of style, treating styles as ghettos limits Culture. It's entirely possible to make a staid workhorse face but be inspired by the numerals in an old French Didot.
But most of all, this is not about the superiority of the old French scheme specifically, but the superiority of keeping one's mind open and seeing the harmfulness of certain conventions. There is no Design without such challenges. So, rethink the vertical alignment of the OS numerals, from scratch, even though most type designers will cry foul at the violation of their preconceptions. Because readers will benefit.
The bottom line (pardon the pun) is that the center-of-gravity of languages that use the Latin script is generally above the middle of the x-height. Let's design accordingly, instead of blindly following convention.