Yeah but I just read (on the LINGUIST mailing list, i think) that the Czechs, or should I say C^echs, changed from +z forms to haczek forms recently enough that family names may well be spelt both ways.
‘Cz’, ‘sz’ and ‘zs’ (zh) are equivalent to ‘c’-haczek, ‘s’-haczek and ‘z’-haczek when accented characters are mot available. […] “'” is the “separator” and makes two letters keep the values they have in isolation, where otherwise they’d affect each other. Hence it mostly corresponds to the hard sign. However, when haczeks are not used it has other uses: Luz'skij (rajon) = Luz + skij, Richard fon Vajc'zekker (former(?) president of the FRG).
To expand a bit, WordPerfect’s Character Set 1 includes these modified Rs:
168 R acute 0154 Unicode
169 r acute 0155
170 R haczek 0158
171 r haczek 0159
172 R cedilla 0156
173 r cedilla 0157
218 R grave
219 r grave
For the lower-case t-, l-, and d- haczek, I intend to use the apostrophe, appropriately positioned; for the d-stroke of Croatian, I intend to use the macron or n-dash (whatever the font-designer regarded as appropriate), unless the font-designer felt it was more appropriate to actually DRAW the additional marks for these characters.
I know how to use the techniques in the RedBook, to extend the CharStrings dictionary of the built-in fonts (Times, Times-Bold, Courier, Helvetica, etc.) — what I need is a reasonable set of metrics, so that, for instance, I can position, for each font, the ogonek with respect to the a and the e, or an apostrophe, to form a Czech t or d with “haczek” and so on.
In some cases the accent developed out of a digraph in which the second letter came to be written above the first. I think this is the case of the French circumflex (superscripted “s”), the Czech etc. haczek (superscripted “z”), the German Umlaut (superscripted “e”), the Spanish tilda (superscripted “n” — or was it “h”?), and many others.
The haczek letters are related to those without, but that’s another story. (To start with, the r/r^, d/d^ and t/t^ pairs are not related in the same way as c/c^, s/s^ and z/z^, and … oh well, I’ll leave that for now…)