Oldstyle tabular figures



  • Thanks, Frank. I couldn't tell if they were mats or castings from the picture! They looked like they were castings in that lead-free metal you use at the MPM. 
  • Distilling some things Craig said: tabular OS nums are easier to make, and easier to read. All-caps looks more glorious than lc too, that doesn't make it a good idea. (Avoid donning those Larson Blinkers™

    boumas do matter.)

    Maxim's wonderful example (I've seen it before...) looks quaint only because of the finish; it's structurally rock solid. BTW note also the wonderful ascending "3" and "5" – the way those numerals should be positioned, today.
  • @Hrant H. Papazian Interesting, know anything more about the "3" and "5" raised positioning? That's atypical of oldstyle figures historically though, correct?
  • It's how the French used to do it. Better. Because the conventional scheme of today sits too low. Set "Porsche 935" and watch the sinkhole.

    I also see an advantage to making the "2" ascend (something Eric Gill did once in a while).
  • Kent LewKent Lew Posts: 905
    Set "Porsche 935" and watch the sinkhole.
    That seems a bit arbitrary. With any scheme of ascending and descending figures, there’s bound to be some combination that sits either “too low” or “too high.”

    It seems akin to concluding that it would be better if we all made y ascend — just set “gypsy” and watch the sinkhole.

  • Kent, indeed it's just an example. But if you look at the conventional OS set, they're clearly sitting much lower; and the risk of distracting lowness is much higher than in the lc.

    https://www.flickr.com/photos/[email protected]/347980084/in/album-72157627102851774/

    But Frode, you're not kidding about Latin not having enough extenders (and any interesting ascenders besides the "f").

  • Chris LozosChris Lozos Posts: 1,402
    It mostly depends on what you are accustomed to. A person who only knows romance Latin languages might be thrown for a loop trying to make Greek lc more "regular."
  • I think that would be so cool. If anyone does go to a German auto manufacturer selling crustier-than-thou old style French figures to "solve all their problems", please publish the proceedings. Meanwhile I can count, as well as performing basic 6th-grade mathematics, and am so proud to show it off.

    The relationship of today's proportions of extenders in the old style figures is perhaps as close as possible to the proportion of extenders found in the latin lowercase. That it works better this way from Maui to Mecca, and has for some time, is a good indicator of how, if one looks.

    While shifting 3-4 figures to ascend, is adorable, and may make them easier to draw tabular, it makes the results better for something... or more likely, it was done because of something else, other than use with latin, greek or cyrillic lowercase, Arabic, Hebrew and others. 

    Being a member of SCOOOFS, I'll leave the math on the Non-Latin extenders to others.

    And thank you for letting me visit.

  • David Berlow said:
    The relationship of today's proportions of extenders in the old style figures is perhaps as close as possible to the proportion of extenders found in the latin lowercase.
    Only when you look at the alphabet itself, instead of actual text; when you factor in letter frequencies (where descenders are far less frequent than ascenders*) number sinkholes abound. Actual letter frequency is also the reason you (yes, you too, thankfully) make the ascenders longer than the descenders, even though the latter cry for far more room to elaborate (versus those sticks up top). Well there is one school (now three, actually) where students seem to be told to make ascenders and descenders exactly equal (Medievalists being Modernist?) but that's another thread.

    * http://themicrofoundry.com/image/s_rome1-4.gif
    & http://themicrofoundry.com/ss_uniglyph2.html

    Some conventions are followed blindly. Well, most actually.

  • Oh, I see. Yes, there is likely to be confusion over the use of frequency, if someone uses it for things it is not really related to. And yes, someone, I am sure, can stretch definitions not often practiced as far as the minds eye can't see as i see shown and written here.

    But back at old style figures..."Don't put anymore complexity into the ascender / cap height area (by moving more OS figures up there)", is much much much more likely to be more useful, if that is what OS fig users over 100s of years of type culture decided was normal. And they not only liked OS figs that way, but from Helvetica to many of Carter's figures and beyond, type designers have knocked a bit off the height, aka the top, of many lining figures, to compensate against what is being somehow suggested, against actual evidence.

    Figures are surprisingly complex for just 10 things. The Latin zero and one e.g. are two of humankind's oldest signs. They are not originally of the alphabet, but come down from Much older signs. One, never changed much, and thus, is different in a lot of scripts, but zero flipped its meaning, from everything to nothing, so it is pretty much the same in most scripts.;p Once you figure that out, and ask yourself why Latin has left the o, O and 0, and i, I and 1 for us to wrestle with, you find a multiple stream of perceptive processes going on that's hard enough to explain to natives.
  • What's been considered normal for a while is important; so is always re-thinking what normal should be, lest we become production monkeys. It would be interesting to research why the French gave up on their superior scheme – I highly doubt it was because their type designers suddenly decided the Anglo-Saxon scheme worked better; it probably had to do with French type design going into a coma after one of the World Wars, waking up with damaged memory, and emulating their saviors.

    Readers can handle much more complexity than we give them credit for (the Chinese aren't exactly illiterate) and the region in Latin above the x-line is pretty sparse.

    Number sinkholes are quite possibly the easiest thing to fix in what's considered normal in type design; for one thing readers wouldn't bat an eyelash. The fact that simple logic doesn't seem to get very far here & now is not a good sign for our culture.

    Latin hasn't "left us" with confusable characters; they are in constant –if typically slow– change, and in fact evidence than conscious design decisions are lacking in writing system development, which is a shame. Also, type designers are best qualified to correct that, and not at the superficial level of glyph instances, but in the deepest structures.
  • It's how the French used to do it. Better. Because the conventional scheme of today sits too low. Set "Porsche 935" and watch the sinkhole.

    That’s why PORSCHE’s nomenclature is always set in all caps.

  • All caps, yes...Roman started out that way. And when Romans said "all caps" they really meant it, but then the introduction of thewordspace, holes from a lowercase, hole-ridden figures instead of Roman numerals, what is to be done!?

    Let's start with LA. If it could be renamed Tos Angeles, we have kerning, TA is better. Next, neither F or f before ? works, so no questions should end in words or acronyms with F or f. WTF? and wtf? see you later, excused after one more go.

    As it stands, this instant, one can "go Roman", like Porche or get to no-holes Latin with Roman numerals too. And if you ask me, I like to call it white space. We... deal with the issue of white space in great complexity, with great care, both in Latin, and as the Latin, and English concepts of white space meets the concepts of white spaces in other languages and scripts.

    Those concepts, are different to say the least, and not unobviously of great concern to many. Now that all the white space is virtual, and all the black is getting loose via subs, pos and vars, we have a lot of opportunities to control it, including where the old style figures make what white space, in what script, writing what language in which style, of what font, at what size, and color, assuming it doesn't move and need to be adjusted for frames per second, I guess.

    Hrant can you please explain what the illustrations at the end of your links, I refer to above, mean for old-style figures?
  • I've become the biggest fan of the underdog white in recent years (because we read notan, not the black) but I'm also a fan of challenging one thing at a time, and in the context of this thread proposing a reform of the entire Latin vertical space seemed a bit imperious, so I figured to here expose people to a much more modest illogic. It's hard to convince people that the Latin three-level house needs remodeling; if I can just help people help readers by simply raising two (well, three) numerals, that's better than nothing.

    Really, if we're serious about the white, we should start by stopping the reckless spread of the black-painting school of type design.

    My illustrations? I thought it was clear: they refute the quote at the top of that post.


    Jürgen, people still write about Porsche independently and use any typesetting/fonts they like... We have to make text look right in spite of them.

    Coincidentally I remember a discussion on Typophile where somebody asked whether it's worth making an at sign that goes with the UC. David opined that nobody sets email addresses in all-caps. I had to point out that Mercedes does (or at least, did). A riveting story, I know...
  • Ray LarabieRay Larabie Posts: 1,193
    edited June 2016
    Does anyone remember seeing typefaces that have old-style numerals which look like shorter lining numerals? I'm pretty sure I've seen this style before but I can't recall where.

    The lining numerals are conventional, designed at cap height or closer to cap height and weighted to harmonize with the capital letters.

    The old-style numerals are the same form as the lining numerals but closer to the x-height and weighted to harmonize with lowercase letters. Like small-cap lining numerals but assigned as old-style.
  • Chris LozosChris Lozos Posts: 1,402
    edited June 2016
    I always make a small caps height set of lining figures if the font has small caps.  I do not assign them as oldstyle though.
  • Ray, shorter lining numerals are typically called "3/4" numerals; OS numerals with bodies notably taller than x-height (my most favorite scheme BTW) are often called "hybrid", although on Typophile at least that was never fully settled. As far as I remember both come from 18th century Scotland. Matthew Carter has been an occasional contemporary practitioner of hybrid numerals.
  • Are small caps figures a pretty rare use case?
  • Ray LarabieRay Larabie Posts: 1,193
    Right, ¾ numerals but assigned to onum. Does that ever happen? I think the point of the user choosing the old-style numerals feature is to make numerals friendly with lowercase. In my opinion, up/down old-style numerals suit historical typefaces but look ludicrous in ultramodern/techno styles. I'm think about abandoning the up/down style except in typefaces with a historical pastiche. Just using the ¾ style, designed to look good with lowercase but without the "look at me!" bounce...and assigned to the unfortunately named old-style feature.
  • I agree OS should mean text-friendly, and/but wouldn't co-opt onum for ¾ lining numerals. However making hybrid numerals (not so bouncy) and assigning onum to them, seems great. Just make sure to raise the 2, 3 and 5.  :->
  • Ray LarabieRay Larabie Posts: 1,193
    Once you get to that point, why pop the 6 and 8 up?
  • It would be great to start with a clean slate, do some heavy-duty analysis and figure out a robust OS numeral alignment scheme. But to me just pulling things up a bit from the too-low convention, leaning on the French/Gill precedents (most people are suckers for precedent, might as well leverage that :-) is a decent plan B.
  • Craig EliasonCraig Eliason Posts: 1,255
    As I'd draw them, oldstyle three and five descend because their bottoms look more like descenders than their tops look like ascenders. Just as with all the other oldstyle figures, the more "rigid" or "primary" portion of the figure goes in the x-height space. In other words, the vertical positioning derives from the shape/structure of the figure in a sensible way.  

    If you want to raise the three to sit on the baseline, you probably have to curl the bottom up to make it firmer (the figure is now "sitting" instead of "hanging"). You see that in the picture Maxim posted. But then you're increasing the odds of misreading it as an eight. Same with the five, which becomes more six-like. 

    It ain't broke.
  • Ray LarabieRay Larabie Posts: 1,193
    It is broke in terms of ultramodern faces. It looks stupid.
  • Chris LozosChris Lozos Posts: 1,402
    The point of individuals designing type is that solutions can and should vary and the users get choices. There is a difference between looking stupid and not looking the way another designer might do it.
  • Craig, those are good points; let me think about them.

    Ray, I think you misunderstood Craig: he means the conventional contemporary OS num alignment scheme isn't broken, while (I think) you're saying OS nums themselves are broken (at least in the kinds of designs you favor).

    One thing I'd say here is that looking smart/stupid is the purview of display type, while acting smart/stupid is the purview of text type. Things in a text type can look stupid (like a vertically cramped binocular "g") but can nonetheless be great design compromises.
  • Craig: "It ain't broke."

    That is not likely to matter for these kinds of things. ;)

    Hrant: "My illustrations? I thought it was clear: they refute the quote at the top of that post."

       My quote was that the extender relations in OS figs were correct glyph by glyph if you'd like. Have now covered 0123568 and 9. 7 perhaps? You responded that I have to look at frequency, and the whole alphabet, I explained why not. You posted not whole alphabets, and nothing to do with visualization of actual type, or actual white space, or OS figs?

       Bring it home. How do your illustrations of not-alphabets, relate to decoding the mysteries of old style figures?

    Hrant: "Really, if we're serious about the white, we should start by stopping the reckless spread of the black-painting school of type design."

       Oh, we Latin designers... are serious. But might OS figure notan is reckless terminology misapplied to Latin glyphs, despite a beautiful thing when taken literally for anything it does apply to. 

  • The illustrations are data visualizations, and a great way to "see" an entire language, as opposed to its pretty-painting atoms. In real text the Latin lowercase sits much higher than one would think by looking at the letterwise vertical distribution; in contrast the average vertical position of numeral strings in the contemporary OS convention sits much lower. I'm a huge fan of variation (hence my interest in what Craig wrote) and feel we're actually too conservative there; but it has to not distract the reader while reading a previous line. A good comparison is the trickiness of the binocular "g": you might see one that looks viable on its own, only to discover that it creates a distracting texture when doubled (common in English) breaking immersive reading.

    Most type designers are indeed quite serious. But each about something different. Which is fine, great even. As long we're open about our differences concerning expressing versus helping, and selling versus conceiving.
  • Chris LozosChris Lozos Posts: 1,402
    We can make tools for our users as an option when they need them. As long as we do not demand that they are used unilaterally, but as a conscious choice, we are doing them a service.
Sign In or Register to comment.