This puzzles me too, along with which numbers are supposed ascend or descend.

Of most OSF that I've seen, they tend to have the short numbers slightly taller than the lowercase x-height, which sometimes is equal to the small-caps height, but those may not be available for comparison.

trust your eyes. i know that sounds glib, but that is really the only hard-and-fast rule in many instances when it comes to typeface design. i believe Jason's observances are accurate as a general rule, but again: rely on your own sense of perception.

I once assumed that the osf would align on x-height and made them so. After printing, I realized they looked a bit too small. As Paul just said optical is optimal. There is no substitute for a trained human eye in type design.

ChrisL

'optical is optimal'
i like that :^)

Chris, I know you said they looked too small but do you know why, i.e. what were they small against? Was it because you're used to the small caps being very slightly bigger? I am trying to get to a good reason rather than your being conditioned over time by being exposed to common usage. "They just looked wrong" somehow doesn't cut it for me: I'm way too inquisitive. Plus how something looks is very subjective. This is a bigger discussion - Is the way something looks all there is to type design?

Nick,
Think of it like overshoot. The human eye/mind percieves the "O" as being off the baseline compared to an "L" for example. You can delve into what it is about human sight which makes this true but nevertheless, it is. You may even spend months figuring out a complex formula to tell you how much overshoot is correct--if you like that sort of thing. The trouble is, you can do it by eye much quicker so why bother? As I've said for years, the eye is the final arbitor.
I am not saying that "how it looks is all there is" but if you are looking for some absolute measure to define perfect placement or proportion, you are about to spend a lot longer looking for a formula than you would just designing type. You have to ask yourself which interests you the most, the pursuit of a definitive infallible answer, or designing a library of type.

"how something looks is very subjective"
To some degree but not really as mystical as you might think--there is a range of acceptable tolerance that most people fall within (again, think overshoot). There is optical correction involved in all joins and stem widths vs horizontals. You can create a statistical measure of "how much" is right and use the mean. Then create a ratio based on this for every weight, slant, degree of width, contrast and style or you can train your eye to see. It is up to you. You must realize that your audience does not use your formula, they just quickly use their eyes. You can jump up and down and stamp your feet if a user disagrees with your measure and point to your formula as proof. This will not change the users impression one bit. Renners drawings for Futura were originally quite geometric but had to be optically corrected quite a bit (some by him and some by the foundry) to look correct to humans.

ChrisL

PS: "why, i.e. what were they small against?"

the lowercase letters.

Thanks Chris, you are, of course, a gentleman and a scholar.

I won't waste my time (or anyone else's) on formulae believe me, nevertheless I want to know why some fonts go for OSF that are lc x-ht and some that are sc X-ht. There must be some rationale behind both. They can't both 'look right' (can they?) Does a rationally good reason ever supercede (or at least subdue) the eye as the "final arbiter"?

No prizes for guessing that I view type design as a science more so that some who view it as an art. Maybe (I hope) my fonts will function better than someone else's which may look nicer (although looking nice is a very subjective criterion - functionality is less subjective). Though for me function and aesthetics are very closely related.

I think you may misunderstand what working by eye does. It is not all about what looks better aesthetically. It is also about what works better to a reader. The use of science or human senses to create type are both attempting to produce functional type. They are not mutually exclusive. Think of the eye as a measuring device loaded with an algorithm that seeks to measure form in the most functional way for other human eyes.

ChrisL

The eye is critical, but some knowledge can serve as a great springboard.
Like knowledge about extender distribution in typical text.

I've concluded that for running text the best system is usually where the "x-height" of the numerals is quite high; these are typically called "hybrid" numerals, and they were used on occasion as far back as the 18th century IIRC; and some designers (like De Groot) do it regularly today. I also firmly believe that the 3 and 5 (and maybe even the 2) should ascend, but that's a harder sell.

hhp

Hrant,
Are what you call Hybrids lining figures reduced down to small cap height or a tad taller?

somewhereChrisL

Well, I think that numerals come before smallcaps, but it can often
often make sense to make the two the same height (which depends on
lc x-height, fullcap width, etc.).

hhp

BTW Nick, there's a lot of stuff about hybrid numerals in Typophile.
Here's one of my favorite threads: http://typophile.com/node/2253
And a nice related one too: http://typophile.com/node/15443

hhp

Chris, I guess I'm asking, why is it that the OSF look small against lowercase? They are not really related to the small caps (are they?)

Hrant,
>>>it can often often make sense to make the two the same height
Why does that make any sense? Is it so that the OSF work well with the small caps. I know of designers who are now releasing 'lining small cap figures' if that makes sense and not using the OSF with small caps at all since they are arguably unrelated forms.

I love logical, that's all.

> why is it that the OSF look small against lowercase?

Probably because their extenders are more modest - they curl and taper.

Although not "related" to smallcaps, the reason I think it can make sense to have the numeral ascenders* line up with the smallcaps is that, all else being equal, it's useful to line things up. Think of something like "MI6" for example. If they're coming close, snap them together (unless you're making a grunge font I guess).

* Or, in a lining system, the top of the numerals.

hhp

I prefer my small caps and old style figures to be a bit taller than the x-height mainly because it makes them easier to distinguish from the lowercase letters in text. The idea for using small caps and old style figures in text is so that numbers and acronyms don't disrupt the flow and even color of the text, but I think they do still need to be somewhat distinct from the lowercase.