I think it will be helpful to properly understand the origins of the tall ascenders in many typefaces. It is not that the ascenders are made tall relative to the cap height, but that the cap height and ascender height are independently determined relative to the x-height. This means that virtually any relationship of cap to ascender height is possible so long as each, independently, looks good relative to the x-height. The tendency in the humanist manuscript hands on which the first roman and italic types were based was that extenders tended to be quite long relative to the x-height, while caps tended not to be overly large, presumably so as not to dominate the page. These proportions were translated into type, and persisted in serif book hands for a long time. In 18th Century England this begins to change, and Caslon's and Baskerville's caps are larger relative to the x-height, and hence the difference between the caps and the ascenders is less.

Don't know if this will help, but an essay by Martin Majoor might be interesting for you:

http://www.typotheque.com/articles/my_type_design_philosophy

Note that the first (modern) sans was envisioned as a display face, where the relationships of the ascenders and descenders to the caps usually fill a different role.

The shortening of ascenders has another modern component. Revivals of classical faces such as Bembo (and even Baskerville), first in photocomp, then PostScript, used shorter ascenders. Whether this was a stylistic decision or simply made considering the marketing of fonts for uses other than book composition, I do not know. It would have been great fun to have been a fly on the wall where those decisions were made -- which likely wasn't the drawing room.

Also, remember that with photocomp > PostScript, the size of the metal font used for the one-master-for-all sizes was a consideration. I believe the 8-point master was selected for Bembo, at least as offered by Linotype for their L-202 typesetting machine.

The smaller masters usually had a design where the character contrast worked better with offset printing from photographic repro, as opposed to repro pulled from metal type on a proofing press. But the smaller masters also usually had a larger x-height, which is essentially equivalent to shorter ascenders.

* * *

If you follow the Majoor essay on down, note his reasons for drawing Seria, which he considered a classically proportioned font, better for bookwork. For whatever reason, Seria hasn't caught on -- at least in the States -- which I find a bit sad.

John, or others, may know more more details, these remarks are essentially only my observations, and I offer them based on how I think marketing decisions for the sale of fonts have affected the extender lengths, and also by the increased use of type for signage and advertising. I'd add that at our shop -- bookwork only -- and where the license permits, we go into the fonts and restore the ascender/descender proportions of the 10-12 point metal fonts.

FWIW

I believe the 8-point master was selected for Bembo, at least as offered by Linotype for their L-202 typesetting machine.

Strange … my Linotype catalogue from 1984 shows a rather skinny Bembo. Apart from the ascenders and descenders being rather short, it does not look like an 8 point design at all. The Monophoto / PostScript version has also always been rather skinny (as is Jan van Krimpen’s Spectrum for Monophoto / PostScript). Recently, Bembo book has been re-introduced. Apparently that design was based upon a master for smaller sizes. The ascenders and descenders are somewhat longer that the roman, so having Bembo book for text and Bembo roman for display sizes makes Bembo roman look rather funny. As if in its underpants …
I’d highly recommend to read Tim Ahrens’ fantastic book on optical scaling for an overview on what has been done for smaller sizes (both historically and recently).

More recently, Monotype released quite a promising range of eText typefaces. As far as I can judge from the screen, they would also serve as a ‘caption’ size, as Adobe and others call it. They are primarily marketed as Web or e-Reading fonts, available as Webfonts only though, but I suppose that many of these fonts would also do better in small print than their regular companions.

Strange … my Linotype catalogue from 1984 shows a rather skinny Bembo. Apart from the ascenders and descenders being rather short, it does not look like an 8 point design at all.

Yes, I think -- but only that -- that in the times of the Linotype 202, 505, & maybe the earlier V-I-P, the various font offerings differed. Perhaps what was used/offered even varied by country.

I distinctly remember having to check master sizes -- for example, a designer friend of mine was briefly taken with Vendome, and wanted to use it for a text. The only "Vendome" available for the L-202 was based on an 18-point master -- at least, in the States. We tried it, once.

The Bembo we had for the L-202 was listed in the catalog, as an 8-point master. It was "skinny" not in width, but in weight. Sometimes Linotype offered choice of master sizes for a font, but these were usually between a size clearly for text, and a size clearly for display. Don't remember, but Bembo may have been an exception, with an 8, 12, and 18 point master size available. If so, the other masters were even thinner in weight.

For anyone studying the the proportions of type, esp. as they occur with metal and hot-metal revivals in photocomp & PostScript, best to go and check the old specimens & catalogs.

Looping this back to the question of the thread, in the Linotype hot metal days, you could order long descenders for some fonts. As best I remember, this was always where the hot metal was itself a revival of an earlier foundry font. Odd in a way, because the ascender size wouldn't change. Whether this was a matter of commerce or aesthetics, I don't know. Probably a bit of both.

Some things:
- As has been said the main reason to make the ascenders and caps match or not is whether you're shooting for a display or text font, respectively. This is because when people are consciously looking at something they generally like things to line up, but for immersive reading that makes the caps jump out.
- Sans fonts are generally more for display than text, so they make the heights match more often.
- The larger the x-height, the more reason for them to match.
- If they're coming close enough, just make them snap; somebody could be using your text font for display (you're not there to stop him) and the slight difference will look like you messed up.
- You can of course make the caps narrower (but not much lighter) to compensate for too much height. But be careful not to make them look like lc.
- For writing systems you don't have nativity in, check with a native concerning acceptable proportions. In Armenian for example caps can look narrower than in Latin without being freaky (although I myself prefer them low and wide nonetheless).
- Sometimes caps are made unusually large because the designer wants the font to be more usable for scanning/searching a document (as opposed to reading it fluidly) since proper names stand out more. Think of a history reference book.
- Sometimes caps are made unusually large because you want all-caps setting to stand out more. There are some fonts where the caps are also/alternatively made unusually dark. Credit where due: I learned this last bit from David Berlow on Typophile, a while back (but I can't find the thread).

hhp

However, for Futura, Gill Sans and Helvetica, this is not the case.

In Futura, the ascenders are taller than the caps. Maybe you were thinking of some other face.

There are many versions of Futura - maybe some of them make them match?

BTW are there any fonts where the caps are ascender-height (or just really tall) but the smallcaps are suitable for running text?

hhp

To address the OP:
The dialectic you propose between old style serif typefaces and 20th century sans serifs is way off the mark.

“Jenson”, not Bembo, may be considered the beginning of typographic time (pace blackletter), and Jenson’s ascenders were cap height.
The pertinent questions to ask:
*Did Bembo make the caps smaller or the ascenders taller, vis-a-vis the prevailing Jenson style?
*Why did he do that?

Cap height ascenders reappeared in the 18th century (e.g. with Baskerville) and may be considered modern.
Again, why?
It’s interesting to note that 18th century modernist (aka neoclassical) page layouts had substantial leading between lines, whereas previously the style was to set solid.
A before-and-after comparison of 18th century book page layouts may be helpful.

Except more leading actually favors a difference between cap and ascender heights.

hhp

in the Linotype hot metal days, you could order long descenders for some fonts. As best I remember, this was always where the hot metal was itself a revival of an earlier foundry font. Odd in a way, because the ascender size wouldn't change. Whether this was a matter of commerce or aesthetics, I don't know.

Neither commerce nor aesthetics, per se.

For metal Linotype, the vertical location of the baseline within the em was standardized and absolutely fixed for each nominal size. A standardized alignment was a practical necessity to enable matrix mixing inline. I believe that Linotype’s alignment closely followed ATF’s “American Standard Lining” alignment from the turn of the 20th century — it was a widely adopted standard. But that alignment happens to have given short shrift to the allotment for descenders.

(Somewhere deep in the bowels of Typophile there should be a previous, long thread about Standard Lining — I don’t know if it’s still intact or discoverable.)

Later, to accommodate evolving tastes and desires (primarily concerning book work), provisions were made to outfit some Linotype faces with longer descenders. Note that for these fonts, the baseline alignment remains fixed by the standard and these descenders actually protrude beyond the nominal em — and so these fonts *must* be cast on slugs with at least 1 pt leading.

Charles — The provision of long descenders as an option was not tied to the existence of an old foundry exemplar. Many designs original to Linotype were outfitted with both short and long descenders — Electra, Caledonia, Fairfield, Eldorado, Primer, probably others.

With certain faces, the long descenders were considered the defaults and the “available-upon-request” option was for short descenders in order to allow the font to be set solid.

(Somewhere deep in the bowels of Typophile there should be a previous, long thread about Standard Lining — I don’t know if it’s still intact or discoverable.)

http://typophile.com/node/64943 ?

Craig, thanks for digging that up! I was looking for that image I posted there.

BTW if I'm not mistaken the super-short descenders originally came from blackletter.

hhp

@Hrant: … more leading actually favors a difference between cap and ascender heights.

Not if the desired effect is to create lines of text that are neat and discrete parallel layout elements, in which case ascender overshoots are messy and need to be cleaned up.

(Whether this has any bearing on readability is another matter.)

Except then one would expect a larger x-height (because that makes lines of text more like lines) but Baskerville for example is pretty modest in x-height. I do think however that lining things up is pretty Neoclassical (but irrespective of leading).

hhp

Craig — That thread has some good discussion about Standard Lining, but it’s not actually the one I was thinking of. I believe there was another extended discussion ca. 2003 or 2004, in which I posted a diagram similar to the one Hrant posted in your linked thread (same system, different illustration source). But my memory may be playing tricks on me. A lot of good stuff from that era got misplaced, unthreaded, or truncated after one of the site redesigns.

But my memory may be playing tricks on me.

No, I think I remember the one you're talking about too.

Me too... although increasingly I remember pleasant things that didn't actually happen.

hhp

Monotype offered a version of Times Roman with long descenders, but it seems to have only been briefly used, and only in the United Kingdom. And if I'm not mistaken, Goudy Catalogue is Goudy Old Style with restored descenders.

@quadibloc:

As I remember, Linotype Times was available in the States with long descenders. There were some photocomp fonts -- A-M, for one -- that offered Times with long descenders as well. Whether by agreement or theft, the A-M Times was a Monotype version.

There was also a Times wide, can't remember if that was Monotype or Linotype, and if it had long descenders available.

As a sidenote, somebody had an absolutely beautiful 14-point version of Times that used long descenders. I came across a book set in it the stacks at Duke library some 40 years ago, & now wished I'd written down the publishing information.

As for Goudy Oldstyle, Lanston Monotype offered a PostScript version that had "long" descenders.

* * *

For so many metal fonts, it could/can be hard to know what was drawn & then cut, and with photocomp anyway, what was available over time.

Remember that at Monotype, little errors happened -- the wrong capital R was sent down with Bembo, and (perhaps) the wrong version of Fournier was what was cut. My acquaintances from the Linotype U.S. drawing room are silent on details that might embarrass management, but I do know similar things happened. I believe Kent knows the old New York drawing room people better than I. And of course David Berlow, who sometimes posts here, worked there in it's later years.