Re: superior numerals

[Thierry Bouche <Thierry.Bouche@ujf-grenoble.fr>]

» Sadly no; there is nothing in Bringhurst about it, as I recall. In my
» experience, hanging figures tend to be variable width and lining figures
» tend to be fixed width, but as you point out, this need not be a hard
» and fast rule.

especially because it's a wrong one;-)

most available figures are monowidth. We made some statistics on
comp.fonts some years ago, it's funny how everybody thinks that
lining=tabular, and hanging=proportional, but this is really wrong. 

Well, indeed, Titling should be proportional lining figures, (usually,
you only get `onefitted' from the expert set), Text should be
proportionnal & kerned hanging (usually, they are fixed width &
kerned, but thyis yields a very bad spacing in sequences like 1\,998
where the excessive side-bearings of `oneoldstyle' can't be corrected
by kerning), Tabular should be monowidth and compatible in style with
the ambiant typography... I strongly believe that digits in maths
should be hanging and proportionnals, just as the letters are.

Maybe not a reference as barbara is seeking, but Mr Hudson, from Tiro
Typeworks has published lengthy postings on comp.fonts regarding this
issue. 

Thierry


PS It is indeed very interesting to see that the Bell digits are the
kind of digits compatible with small caps, the fact that such digits
are so rare, may be related to the fact that SC being old-fashionned,
we use roman numerals with them?

PPS i've read in this thread that footnote markers should not be set
as superscript: we french never do that;-)

PPPS when you use hanging figures, you'd like that \MakeUppercase{123}
prints lining figures...