At 19.15 +0100 2004-01-16, Vladimir Volovich wrote:
>The following message is a courtesy copy of an article
>that has been posted to comp.text.tex as well.
>
>"LH" == Lars Hellström writes:
>
> >> 3) The fontinst package contains some code to put the ring close
> >> to A (the glyph ringfitted defined in latin.mtx, and used in
> >> ot1.etx).  (and this package is meant to be used for installation
> >> of "arbitrary" fonts)
>
> LH> This is mostly because the (implicit) definition of the OT1
> LH> encoding made in ot1enc.def, since the definition of the OT1
> LH> \r{A} (\AA is obsolete notation in LaTeX) assumes that the width
> LH> of the ring glyph is equal to that of the A glyph. In other
> LH> words, this is effect, not cause.
>
>agreed; BTW maybe the t1.etx should also use the ringfitted: it will
>not cause any harm, when applied as an accent, and will "survive" the
>trick too (though this trick is not used in T1 encoding); e.g. EC fonts
>have ring's width equal to the width of A.

I would look odd if the character was produced in isolation, though. I
believe \r{} is perfectly legal LaTeX (even though I cannot see why one
would want to write it), and in that case the width of A would look wrong
(especially in comparison with other accents in isolation).

> >> 5) my memories about the look of the Angstrom sign in the books
> >> suggest that it should be written as plain TeX does it.
> >>
> >> Arguments against the close placement of the ring (from Alexander
> >> Lebedev and Lars Engebretsen) state that from aesthetic reasons
> >> the ring should be put on the same height as other accents.
>
> LH> This might actually be the reason to _make_ it touch the
> LH> A. Recall that accents over capitals are usually quite close to
> LH> the letter (closer than in the case of lower case letters). Also
> LH> note that the height of the ring in many fonts is significantly
> LH> larger than that of the dieresis. Vertically centering them both
> LH> on the same axis may well result in the ring touching the A.
>
>is it "desired" that the ring touches the A?

It would simply be a consequence of other aspects (height of capitals,
position of "accent axis", thickness of lines, ...) of the font design.
Probably nothing that one would strive for from the start, but certainly an
acceptable way of handling geometrical realities.

>note that the design of the ring accent in the EC fonts as a separate
>glyph and the ring used in Aring is somewhat different: the ring in
>Aring is "the real ring", while the ring accent is really an ellipse;
>but it looks like the height of Aring pre-built glyph and A with the
>ring accent (using \accent) is the same (as can be seen from the latex
>example i posted in my original message). This is not so in CM fonts:
>the "artificial" aring has smaller height and ring looks closer to A
>than in other accented letters.
>
>so the questions are:
>
>1) is it acceptable/desirable to have the ring in Aring placed closer
>   to A than in other accented letters?

This is probably not only an issue with Aring. You might also want to
compare e.g. \"{O} and \"{\empty O} (the \empty defeats the
composite-lookup mechanism in LaTeX, so that it defaults to the definition
via \accent). I'm pretty sure that the dots won't be in the same position
there either.

Capital accents are different from lower case accents. The loose accents in
OT1 and T1 fonts are lower case accents. Using the placement of the latter
as a rule is probably not a good idea. fontinst could also do better in
that department.

>2) shall the artificial accent placement be applied to other combinations
>   of capital letters and accents on the level of encoding definition
>   files?

? Please clarify.

>3) you wrote: "Vertically centering them both on the same axis may
>   well result in the ring touching the A" - but what i'd like to know
>   is whether the ring SHOULD be touching the A (not MAY).

Gap and no gap are typographically equivalent. As I wrote above, it
probably depends on how other aspects of the font design works out whether
there in the end will be a gap or not.

Some fonts have both a gap and a touching position of the ring. They manage
this by cutting out a bit of the top of the A, as if there was a circular
margin around the ring.

> >> 2) is this rationale only a design decision of the Computer Modern
> >> fonts, or it can be applied to other font families? I.e., shall
> >> the ring in the Aring glyph be put with a gap of the same widths
> >> as all other accents, or it is preferred to use "gap-less" Aring?
>
> LH> This is probably nothing CM-specific.
>
>I.e., do you mean that "gap-less" Aring is considered to be the best
>practice of using of this letter in Scandinavian languages and/or in
>physical literature?

No, I mean it is probably up to the font designer to decide which looks best.

Lars Hellström