> When I set this in small caps, it looks like LAS, which is no good.

Which is why smallcaps are best when they're visibly taller than the lc x-height. Unfortunately most fonts have smallcaps that are too short. BTW, many people put an apostrophe when pluralizing an acronym, but I think that's a sad fallback.

As for the initial-acronym situation, I certainly wouldn't "capitalize" it! You might just use full-caps there, especially if your font has smallish smallcaps.

BTW, the richest font I know in terms of smallcaps is FF Atma:
http://www.myfonts.com/fonts/fontfont/atma-serif/
And it's decently spaced.

hhp

I always thought that small caps were supposed to be level with the x-height. The only reason why they recently have not, was because people used shrunken caps and if they did it at the x-height then it would be way to thin. Of course doing it at 70% also thins it out but isn't as bad as doing it at the x-height.

Well, I guess it's back to full caps or I set up bigger (ie. artificial) small caps. Vive la revolution.

I usually make my small caps larger than x-height, more so if the x-height is small. There is a lot of historical precedent for this, Rafael.

But I don't think there is a formula. Goudy even made the SC of Italian Oldstyle smaller than the LC x-height (what a show-off), and it looked OK. So that's worth examining, and it confirms that a glyph contains a sense of scale.

Italian Old Style (the metal version) has very small counters in the bowls of the a and e, and is very tightly fitted. If you compare the counters in the SC "E" with that in the "e", they are of a similar size. And the SC is openly fitted (ie it looks "tracked out"). So the LC looks like a bigger "optical size" setting than it really is, and the SC look like a very small optical size setting of a cap. Hence they work together. (This beforesaid concerns the sense of scale of proportion for letter stem lengths and negative space; detailing such as stem and serif thicknesses is more about consistency of absolute values between SC and LC.)

So, if a typeface has small caps that look too small when they are set beside the lower case, they are not well designed. If a typographer is going to tinker with the setting to compensate for this, I would suggest combining three imperceptible adjustments:
1) increase the SC size by a fraction of a point, such as 0.2.
2) increase the tracking of the small caps, possibly by quite a lot.
3) increase the horizontal scaling of the SC by a couple of percentage points.This is a technique that digital type designers often use when they are developing their small caps from the big uns, but you might decide a little more is required.

There is not necessarily anything wrong with a typeface if you feel it is necessary to do this; it may be that some typefaces require a little more effort to work with than others of more easy virtue, or that typefaces not naturally suited to the requirements of a typographer or a project require some special treatment. Massaging a reluctant typeface to the requirements of a project can often result in more distinguished work than going with the most obvious or most comfortable choice.

By way of a test to determine how thorough a typeface design really is, set the three "O"s (cap, sc, and lc) and the two zeros (lining and oldstyle) beside each other and compare how differentiated they are with how well they harmonize.

Raphael, it seems you're talking about "optimal" fake caps, while the main issue here deals with fonts which have smallcaps provided... But come to think of it, Nick's idea of doing some "fake" stuff with the true smallcaps might be just the thing! :-)

Nick, I think larger apparent size certainly helps, but it's not enough - sort of how an upright italic doesn't work for emphasis: cursiveness usually helps, but alone it's too weak. Consider the "LAs" example: the lc "s" being a bit narrower than the SC one would not resolve the confusion.

hhp

No, I was meaning real small caps. I was saying that according to my understanding people have got used to small caps being larger than the x-height because people used fake small caps and that fake small caps were larger than they "should be" because to make them the "correct" size would make them too thin.

Of course it was always my understanding that true small caps were supposed to be x-height without really thinking about why I thought that. I would be interested to know where small caps came from and what their original use was.

Small caps go way back, as a form of contrast. Before there were sans serifs, or bold faces, and when it was a real pain to change size, printers would use the four main styles: Roman Caps, RomanU&lc, Roman SC, Italic U&lc, as their "palette" of tonality.

Back in the bad old days, Linotype use to make �special 2-letter small caps� for a few designs -- that is, both roman and italic small caps duplexed on the same mat. The roman of these had slightly more height (& usually, also weight & contrast) than the �regular� roman small caps, & were intended mainly for display � running heads, subheads, etc.

The regular small caps were usually x-height, but I�d agree, we are better off when they are slightly larger than x-height. It is the amount of �slightly� that can start an argument.

FWIW

Charles

<font class="dontLookLikeCrap">I just finished typesetting a 180 pg academic text in Adobe Caslon (the new-ish pro OTF version). Tremendous number of acronyms, and, unfortunately, many in quoted text, which had to be set in italics. The font lacks an italic SC so I had to use 16-degree-obliqued smallcaps, slightly fudged to match the weight of the italic - you don't want to know what I had to do.

I love this face but the small-cap height is too low and the plural acronym problem came up again and again.

Unfortunately FF Atma would likely cost me several dozen hours' worth of wages and I just can't afford the investment now, so I end up with another, perhaps finer (that is, finer than, say, "fake" smallcaps), but still less-than perfect way of kludging it.

The other face I'd really like to be using for book work is Mark van Bronkhorst's wonderful Verdigris family, but it doesn't yet have all the features I need it to.

Joshua, figure the cost of the fonts into the job budget, add a mark-up to cover your font management expenses, and send Mark a few bucks on account, to help direct his attention to the features you'd like to see in Verdigris.

You can always sell Adobe Caslon (technically, the term is "transfer"), or perhaps FontFont would accept a trade-in against the cost of Atma.

>several dozen hours wages?
You need a raise!