X-height and extender changes across weight

Jay Langly

It's been explained to me that the x-height of a typeface increases as a type gets heavier to recover some more vertical room and to prevent the type looking 'squat', also there it helps to balance the image optically when setting different weights next to each other. Correct me if I'm wrong.

I've observed that this doesn't happen also for the ascender and descenders, why is that?

Michael Vokits

Frode Bo Helland said:

There was a thread initated by the ever curious @Nina Stössinger on Typophile, if I remember correctly.

Still offline, alas.

Nina Stössinger

I can’t remember the exact title of that thread, haven’t been able to dig anything up . What I did find locally was this image I made for that thread (December 5, 2009, in case that helps for finding anything), to show that some text faces do indeed shift extender lengths as well; compare Adobe’s Caslon and Garamond:


I get the impression this sort of treatment differs a lot from design to design, and from designer to designer; in the end it comes down to balancing it in a way that looks right. It’s a tradeoff: Making the weights look more like they are the same size in text will mean that they don’t line up when placed next to one another, which I guess can be undesirable in a face that’s more for display use.

In my understanding, the lowercase is usually most in need of enlarging because (1) it gets denser more quickly when the shape gets bolder, having a lot of horizontal elements and (2) it contains a lot of closed shapes, and the size of the counters is crucial in how large we perceive a shape to be. Most ascenders and descenders in Latin just being vertical sticks, making them bold within the available space isn’t as problematic, and their influence on the perceived size of a shape maybe also isn’t as big.

The above suggests that this problem is more pressing the lower the contrast is. As an aside, in my upcoming reversed-contrast text face I’ve had to significantly recompute the vertical proportions between weight masters. As the weight sits mostly on the horizontals, dark weights required a lot more vertical space – mostly in the lowercase; but leaving the ascender line in place would have shortened the relative length of the ascender so much that the ascender serifs didn’t fit in very well anymore. So I had to raise the ascender line a bit too. FWIW. There are interesting parallels to Arabic there, I was reminded of a talk I saw by Kristyan Sarkis where IIRC he mentioned having to adjust vertical metrics between weights rather significantly for Arabic because since the weight sits on the horizontals, the shape also grows in that direction when made bold.

Deleted Account

I've observed that I want weight to effect sans extenders more than serif. The serif has, well, serifs, and serifed extenders don't optically change height very much.

On the other hand, it's not terribly useful or visually appealing for a heavy sans' square ascender to exceed the cap ht, for text or display use, while it is often invaluable, optically, to have a thin sans' square ascender, in the same family, exceed the cap ht, sometimes the cap overshoot, and sometimes, if it is a small enough master, thin enough weight, and wide enough body, the ascender seems to want to go on up forever. 

I think this is because we are trying to make upper and lowercase heights that appear consistent across weights. And that means extenders and under/overshoot must compensate for size, weight and width, at least. And... what Nina said about reverse stress in Latin, and normal stress in Arabic, becomes super-stressful to vertical metrics in many CJK designs just beyond bold.

Kent Lew

Here’s a Wayback of the original Typophile thread.

http://web.archive.org/web/20100103041624/http://typophile.com/node/65057

Simon Cozens

So, http://typophile-archive.simon-cozens.org/65057 , then.

Kent Lew

Simon — You changed the URL for your reconstruction without notice. Otherwise, I might have looked there. Is this a stable address now?

attar

Too bad profile pictures are stripped out, it helps following the course of who says what.

Simon Cozens

Sorry, Jay, I've totally derailed this thread. :-( Maybe we could put Typophile Archive stuff in another thread, and please file bugs, feature requests and so on at https://github.com/typophile/typophile.github.io/issues/new

Deleted Account

The one great thing about typophile, I guess, is that it didn't get interrupted by people looking for threads on typophile.

Mark Simonson

I think there actually were, because the built-in search engine was broken.

Deleted Account

...then, you're saying there was nothing good about typophile?;)

Mark Simonson

Let's see... Endless spam, endless type id requests, broken search, Flash-based image upload, poor layout for small screens, responsiveness grinding to a crawl frequently... Other than that, it was great.

Nick Shinn

Remarkable discussions on the Build pages, very educational.

Deleted Account

In any case, back on this thread, in the other thread, Nina said she was making the x-height of the bold, 4 units taller than the x-height of the regular. Is that right? 

Nick Shinn

There is also the question of what to do with alignment zones.
If the Bold has a higher x-height than the Regular, then should the alignment zones be standardized throughout the family, based on the Bold?

That’s what I did here (Regular outline pasted into Bold font).

I also recall that in making the Bold I increased the weight—e.g. Make Parallel Path—of the Regular by an amount that conveniently increased its x-height. (That was a separate process than increasing the thickness of the vertical stems.)

Nina Stössinger

What is the reason to standardize alignment zones if the overshoots differ?

Jackson Showalter-Cavanaugh

Isn't this what the Postscript Family Blues are for? My notes say those zones are used if the difference is < 1px.