Beyond what you wrote, I would like to add:

• ampersand
• quotes
• at
• numbersign
• asterix
• percent and perthousand

It breaks down into how small caps are used. For books anyway, there are two uses, (1) run into text and (2) used as display, as for subheads.

For (1), you want to maintain most of the punctuation of the regular font — it would look odd to have the quotes change (includes apostrophe), the period change, etc. For most of us book people, it would look odd to have the numbers change, unless someone was so crass as to use lining figures in the text.

For (2), ideally you have a separate set of glyphs that should mirror the full caps in all ways they can occur in display, just smaller, but with a weight that lets the smaller size carry itself when compared to the weight of the text. No telling what words or punctuation might appear in a heading, so you would need everything where the position or size of the basic font was appeared odd — just as with full caps.

As to size, the run-into-text small caps should be just a bit bigger than x height, with the sc “X” roughly 10 to 20 units taller that the topmost part of a lc “n”. They need to blend into the lower-case letters, without appearing optically smaller.

For display, it is a harder call, but I’d say a height should be around 10 percent greater than a l.c. “x” would be most useful.

FWIW

Which special characters schould be activated with smcp-feature, and which with c2sc?

This distinction addresses the “double duty” described by Charles.

What I’m doing now is to put the “text” features (1) in smcp, and the “optical size” features (2) in c2sc.

That is to say, “smcp” only substitutes the lower case alphabet.
So smcp, which is really “caps with small caps”, has default figures (whether cap-height lining or oldstyle), whereas “c2sc/all-small-caps” has small-cap height lining figures.

However, I don’t make c2sc versions of every character, omitting those which I consider to be “text” characters that don’t usually occur in all-capital settings, such as parentheses and brackets. I think the bar character is instrumental in determining the look of “all small caps”, representing the way that setting is most likely to be used, in a folio for instance.

Here are the two styles.

thanks for your answers.

you are right, it depends on text face or display face. Sigh ... all those differentiations.

Concerning smcp and c2sc – I have these sets at the moment:
- upper case
- lower case
- smallcaps

- lining figures
- oldstyle figures
- smallcaps figures

- uppercase currency symbols
- small caps currency symbols (addional characters like bracked.sc would go here I guess)

I just don’t know yet whether to use smallcaps figures when activating smcp or “reserve” them for c2sc. My default figures are lining ones, but I am not sure about that as well.

I think the bar character is instrumental in determining the look of “all small caps”, representing the way that setting is most likely to be used, in a folio for instance.

I don’t get this sentence... might be because english is not my mother tongue.

If lining figs are your default (nothing wrong with this), then I would sub oldstyle figs with {smcp}. I would sub your sc figs only with {c2sc}. You might also consider parens, braces, and brackets to sub with {c2sc} but not {smcp}.

As Nick pointed out above, {smcp} is usually invoked when the designer wants U&sc. Thus the punctuation and figures should accommodate both heights. Standard puncts and oldstyle figs do the best job in this case, in my opinion.

Nick’s comment about the vertical bar has to do with the fact that one of the most common uses of small caps, at least in book work, is as a running element in conjunction with a folio (page number). It is currently a favorite stylistic device to separate these two things with a vertical bar. The standard vertical bar stands too tall when set in a line of small caps and doesn’t look vertically centered against the nice horizontal line of small caps. Book designers usually baseline shift the bar down a bit to look right. Since this is usually done once on a master page, it’s not such a terribly tedious task. But his idea of an alternate form for [c2sc} is a nice detail.

— Kent.

I guess my biggest purpose on Typophile is just to represent those who use type.

So here is a for-instance:

Right now, I’m setting a book with 4,000,000 characters (1,100 manuscript pages). Sequence of events: The files are supplied by the author, then edited. After editing, which includes any coding or change to the text, the book is designed. And in designing the book, the designer will decide that certain elements, some of which (e.g., acronyms) are run into the text stream, are to be set in small caps. The book does NOT go back to the editors after design.

Except for the size, this is a pretty typical workflow.

OK, think it through: in the copyedited files, the elements to be set in small caps may be (1) full caps, (2) a mixture of cap & lower case, or (3) lower case. In the week that I spend typesetting this, when I encounter one of these elements in the text, all I want to have to do is hit “all small caps” (which takes case of 1-3), and be done with it.

Well, almost done. I’ll then run a global S&R for “all small caps” with “0 tracking” and replace the “0” tracking value with a value appropriate to the text font. Small caps in display won’t be affected; they’ll be set either through a paragraph style which will include a bit of tracking, or occasionally “directly formatted,” again with tracking done as set.

If the type designers start messing around with just which small cap function does what, I’ll have a lot more work & confusion.

I believe it was Adam who suggested that “display” small caps really call for a separate font, and I agree. If you’re gonna screw around with just what feature does what, put it all in a readme. And make sure the EULA lets the end user go in and modify the font(s) so they are useful.

So, Charles, are you advocating completely against contextual punctuation for small caps? (I don’t see how that would affect your example, since presumably you’re just selecting the letters of the acronyms.) Or are you advocating against including any figure substitution in either small cap feature? Or all of the above?

So far, I haven’t seen anyone suggesting that “All Small Caps” not change all cases to small caps — {c2sc} for the caps and {smcp} for the lc. Software must implement both for the All Small Caps command.

I can respect the opinion that these two features should perform only the minimum written in the spec (which BTW, suggests that OsF may be included in {smcp} subs).

But it doesn’t seem to necessarily follow from the example that you outlined. Or, at least, I’m not seeing it.

— K.