The Arabic writing system has horizontal stress,
and that's the origin of the Latin numerals. The thing is, in the "conventional" Latin scheme the "2" and "5" have a mix of h and v stress, the "7" is mostly (sometimes entirely) h, the "4" has this wierd diagonal, and the rest are all v stress. Watduhel?! I'm actually a big fan of variation, but this is just arbitrary (in the same way that so many typographic traditions are).

I've tried to bring a method to the madness, but I'm actually just as unhappy with the "2" as you are. I think/hope that the "5" and "4" work great, and the "7" would work better in a serif face (where it could get a beak -and maybe foot serifs- to fill the void).

The "2" might just need to be wider. Or maybe you're right that the diagonal needs to become straight[er] (and thinner) at the bottom (instead of turning back). Another thing about the "2" is that it has a stiff spine (with a totally straight segment in there), and maybe that ruins it.

BTW, great idea about making the "7"'s diagonal echo the "4"'s more (at the top). Thanks!

hhp

The more I think about this, the more I get confused... :-/

I can build a set of "normal" numerals and make them work 90%, but I won't really be happy until I can justify what I've made. Imitating precedent has its merits, but doing so because of intellectual sloth is just... sucky.

Why give the numerals consistent vertical stress? Does it help low-level readability, or does it just satisfy our Modernist desire for Regularity? It's the former that intrigues me most, but I think in the latter I at least have a safety net: giving the numerals v stress makes them visually harmonious with the alphabetic glyphs. The question is, do they *reduce* readability?...

The thing is, no glyph is pure text or pure display. During my "Alphabet Reform" project I realized that you simply cannot *show* structures: they can only be *explained* (something graphic designers don't seem to like to do, btw). That's why "theoretical" alphabets (like Bayer's "Universal") are always ugly - they have to be seen as a conceptual compromise, not usable things.

So any actual glyphs I make/show inescapably represent parts of an actual font, something that can be used at different sizes, in different ways. Among other things, this brings the conflict between aesthetics and functionality to the fore. Specifically: my "2"'s spine makes it structurally divergent from some other numerals, which makes it more legible; my "1" has that ugly bar because it helps pull it away from the UC "I" and lc "el"; my "3" has a sag in its waist because that makes it more of a 3 (and not because Garamond did it); and many other such machinations.

So your/my ideas for making the "2" more pleasing would make it less legible. Legibility isn't everything, and I sacrifice it myself sometimes (for example, that "8" is not ideal, but the "optimal" structure I have in mind is simply too ugly), but knowing where to strike the balance depends on so many factors! For example, if these numerals were meant for marking addresses on houses, the "1"'s bar would go. But if they were meant for telephone books, I would dig out my ugly "8".

The more versatile you make a face, the less well it will perform in specific applications. Look at Tracy's Adsans. It's ugly, but man, does it ever *work*.

Speaking of other existing fonts, I was looking through the Rookledge Typefinder (again...), and I actually found a font that implements v stress for numerals in a largely pleasing way: Egyptian 505. And who designed (or at least directed) it? Gurtler, one of my very favorite designers! :-)
(BTW, Gill I worship, but his attempts at v-stress numerals I can't stand to look at.)

On the other side of the spectrum, there are also fonts (though very rare) which give horizontal stress to the *alphabetics*! Like Excoffon's Olive, of course, but also Bloemsma's Balance. What's interesting is that they also make *all* the numerals have h stress.

hhp

BTW, I think the problem with the "6"/"9" is that they close in on themselves (which is somewhat unconventional). That gives them a bit of movement (a bit like the "3"'s waist), but I can't decide if that's good or bad. Or maybe the curves are wrong...

hhp

I'm making alternates for 2 and 7 (and maybe 6/9 too), and here's an 8 from a font I did about two years ago. I think it's *structurally* optimal in terms of legibility. But it's too grunge.



Now, when it comes to familiarity, I agree that it's important. But I also think too many designers use Emigre's mantra as an escape from having to look the beast that is Readability in the eye. You can't just make whatever shape that turns you on, if you care about reading comfort: there are realities about our retina, our brain, etc. which make some fonts inherently less readable than others. For one thing, the reason there's a difference between legibility and readability is proof that familiarity cannot be all-important.

This doesn't mean I or anybody else understands -or can ever understand- readability completely. But *that* doesn't mean we should try to understand it more. It might not be very sexy, but you don't marry the sexiest person you know.

I'm a big champion of innovation, but only when it makes things better for users/readers, not to stroke my own ego. There are so many stale, dysfunctional traditions in typography - they have to be fixed. But too many designers think of themselves are "artiste"s, and are too afraid/lazy/dumb to consider the intricacies of immersive reading. These people will never make a real text font.

Lastly, about your "conversion" on the 6/9:
I believe in the separation of conscious observation of shapes versus the quick-and-dirty reality of our subconscious reading mechanism. What does this mean? It means that a layman simply cannot look at a glyph and declare if it's highly readable or not - we cannot hope to understand our subconscious directly. For example, there have been many tests done asking readers to choose the most readable font out of a set, then they measure the actual reading speeds of the various fonts. Invariably, the most readable font is smaller than the one readers think is best. What the readers are consciously choosing actually has to do with *legibility*, not readbaility. And the two have a congruence of only about 50%.

So, how can we find out if a given glyph works or doesn't? 1) Do a lot of good field testing. 2) Simply *think* about it as much as possible (instead of running away saying "it doesn't matter"), using your understanding of how humans read. That's Craft.

hhp

A couple of corrections... :-/

1) Paragraph 3's second sentence should be:
"But *that* doesn't mean we shouldn't try to understand it more."

2) I forgot to "conclude" the next-to-last paragraph... What I was getting at is that your "coming-around" to the shapes of 6 and 9 is *conscious* (and that has gobs to do with familiarity), but it's unsafe to extrapolate from there and claim that they work OK during actual reading.

hhp

Hello again...

I hope that type designers will study your work, Hrant, as the conventional stress, or rather lack of stress, is aesthetically horrible. When it comes to readability, it seems that you know much more than me. But I think that a consistent stress could benefit readability (although it would reduce legibility by reducing variation) as it calms down and strictens the overall impression (and so the readers mind) and sets the focus on individual variation. It seems to me that readability is much about regularity vs irregularity, and trying to walk the thin line between these two abysses. Your thoughts are indeed interesting, and it

First, Anders, thanks for your encouragement - I happen to need it right now...

My understanding of readability is still shallow. The only real credit I can take (and am proud to take) is that unlike many designers I *care* about readability.

About your suggestion for the "4":
I've always liked open-top "4"s, but they kind of stop working well in darker weights.

As for the advantages of straight versus curve, that's a very interesting line of thought. In terms of legibility of numerals you want each number to be as different as possible (as you implied), so a totally straight-lined "4" is best. But in terms of readability, no glyph should stand out too much in terms of *structure* (even though I'm still unsure about the merits of totally even color). For example, I think the conventional lc "g" (the bicameral one with the closed descender) is too different from the rest of the alphabet: I make my text "g"s with an open descender. In fact that's the best justification for making all the numerals completely v-stressed.

The "6"/"9": I'm starting to think they need to be open, like you said. For one thing, it's better for legibility.

The "1": I thought the bar at the foot was ugly... But now I think the top's ugly too! What to do?

The "7": Putting full serifs at its foot might seem logical to you and me (both aesthetically and functionally), but for some reason very few people seem to do it that way... Another stale convention?

BTW, I'm flattered that you think I should write a book, but for now I'm already in enough trouble for spending too much time on stuff that doesn't make money, so I'll have to refer you to an article I wrote about Alphabet Reform, since it contains the basis of my thoughts about readability. It's in this book:
http://gunnarswanson.com/GDandR/GDandR_TOC.html

Eventually, I plan on writing a short on-line thing.

hhp

What do you guys think of this attempt by Catich?
http://www.themicrofoundry.com/other/catich_nums.gif
(They're intended to match the Trajan style.)

hhp

mmmmm... I like the lapidary nuances. The details on the '2' and the ... well, quite frankly ... I really like this.

I think it shows that giving the numerals vertical stress can totally work.

But the "2"... Why compromise it like that?

hhp

I just realized that the typeface Yellow (by Jürgen Weltin), a winner in the Bukvaraz competition, has "rationalized" numerals. Yay!

hhp

An exercise in vertical stress, eh?

For the most part, I think they work very well. The "2" is grating on me slightly, but I think it's only due to me being used to seeing the thick stroke/thin stroke distribution attuned to more horizontal stress. It does seem a trifle thick at the top of the main stroke, though, compared to the "6" and the bottom of the "9". The thicker tapers of the ends of the "3" work a bit better, due I think to the tighter nature of the curves (that may temper somewhat my criticism of the "2" -- but not completely ). Take another look at the "2", in any event. Perhaps a tighter angle between the main stroke and the base may be an alternate solution.

The gentle curve and taper of the "4" is really elegant, and plays well off of the "7". In fact, I think I'd like to see more of the "4"'s taper and curve incorporated into the "7".

While you said that these figures were intended for text sizes, I can envision them as attractive display numerals as well. They'd also look great cut out of metal. Very nice.

Indeed.... I had to fight my first impulse to ask for a thinning at the top of the curve of the "2", and a thickening of its base stroke. I have been intrigued by your previous challenges to the conventional mixed-stress figure structure, and I think what you are doing here with uniformly vertical stress is equally interesting.

Hmmm... On looking at it a second time, maybe the "2" could benefit from the same curve (if not the taper) that you used in the "4" (and will apply to the "7").

The "5" does indeed work very well, as, I think, does the "3". In fact, I very much like the slight downturn of the middle vertex of the "3".

I keep looking at the "6" and "9" as well, but I can't put my finger on why. Then again, I'm never quite satisfied with those figures in my own fonts either. The curves required in those two figures are deceptively complex, and difficult to keep balanced....

In any event, keep it up. I'm interested to see the continuing results.

Courage, Camille. :D

Your point about balancing aesthetics and legibility is well taken. I'd be curious to see the two different extremes as applied to this design. Dig out that ugly "8", as well as the rest of the discarded alternates, and parade them for us.

I often find when constructing a font that the glyphs that I think aren't working are simply reacting badly with other glyphs that are the real culprits. Also, sometimes when I think a glyph isn't working, it turns out to be simply my discomfort with a design that I'm not used to, but the design itself turns out to work very well, thank you very much.

It's like what all those folks at Emigre keep saying: We read best what we read most. Or something like that. Legibility borrows a great deal from familiarity. The moral (if there is one) may be, without overstating it: don't discard an innovation because it is unfamiliar. I think what you're attempting is worth the investigation, and I think you're getting somewhere with it.

I also think you may be right about the "6" and "9". As I look at them today, they work. And I think that's a splendid example of the point in my previous paragraph.