Re: Symbol font mappings

From: Robert Herzog (
Date: Fri Oct 22 1999 - 12:46:15 EDT

On Fri, 22 Oct 1999 06:17:54 -0700 (PDT) Markus Kuhn wrote:

> Kevin Bracey wrote on 1999-10-22 12:32 UTC:
> > >
> > > 6.
> > > is a GLYPHlist. Unless it is changed PostScript 'phi' means 'closed phi'
> > > and PS 'phi1' 'open phi', because that's how they appear in the Symbol font.
> >
> > Who says the Symbol font is the reference? Those names apply to any font.
> Good point.

Hmm. I assumed, because it's a GLYPHlist, that PostScript 'phi' means
'closed phi' and PS 'phi1' 'open phi' _per definition_, also indipendently
of the Symbol font. But yes, one could also change these definitions, as
Kevin seems to prefer.

Changing the glyph names in the Symbol fonts might be an
> alternative to do here then. I'm currently reworking the X11 fonts for
> XFree86 4.0 anyway, so I could easily change the glyph name before I use
> the unmodified glyphlist.txt to add ISO10646 encoding information to the
> fonts. The problem is just: If I make the font change in X11, this still
> won't help with the many PostScript printers out there. The entire
> problem remains quite a mess in the end, so I better don't touch
> anything, and we will have to live with the fact that Symbol remains
> mapped according to the Unicode 2.0 glyph tables.

Hmm. Obviously the aim is to get away from the Symbol font as soon as
possible, it's a hack after all - unfortunately a very widespread one.
Unfortunately I am not an expert able to suggest a 'best' route.

The W3C introduced named entities for all letters and symbols with a
Unicode code point in the font to 'eradicate' the use of
<font face="Symbol">a</font> for an alpha. This is clearly a good idea.
Similar efforts should be made in other software where the Symbol
font is used to carry semantic value.

Unfortunately I am not an expert in PostScript to assess which of the
following changes is 'better' (for legacy support etc.):

1. Change the Adobe glyph list, ie:
  U+03C6, 'open phi' (change)-> PostScript name 'phi1' -> Symbol code point x6A
  U+03D5, 'closed phi' (change)-> PS name 'phi'-> Symbol cp x66

2. Change the Symbol font encoding:
  U+03C6, 'open phi' -> PS name 'phi' (change)-> Symbol cp x6A
  U+03D5, 'closed phi' -> PS name 'phi1' (change)-> Symbol cp x66

That the Symbol font mapping has to change to reflect Unicode 3.0
(i.e from U+03C6 -> x66, U+03D5 -> x6A to U+03C6 -> x6A, U+03D5 -> x66)
is clear: the font is _everywhere_ and can not be changed.
Is there someone from Adobe to comment on this?

> Note also, that the "Adobe Symbol" font is completely useless to print
> Greek text. It contains only a subset of ISO 8859-7, the glyphs are all
> in italics, and the entire font was clearly designed for mathematicians
> and not for Greek text processing.

Yes, and even there the x-height of the (Greek) glyphs is 10% larger than
the x-height of the glyphs in Times...
> One comes even to question whether it was a good idea to unify say
> and
> because the typographic styles used by Greeks and mathematical
> typographers differ for some characters quite substantially. (But then,
> So have Greek and Coptic, French and Polish, and other communities with
> style differences been unified, so it is at least consistent).

I agree, but at least the semantics is clear here. So this is mainly
a font issue.
> Recommended reading on the conflicts between Greek (ISO 8859-7)
> and mathematical (TeX, Adobe Symbol, etc.) usage:
> Haralambous: From Unicode to Typography, a Case Study: the
> Greek Script, Proceedings of the 11th Unicode Conference, Boston,
> 1999, <>, 4 MB.

Yes, this is a very good paper.

On Fri, 22 Oct 1999 05:32:05 -0700 (PDT) Kevin Bracey wrote:
> In message <>
> Robert Herzog <> wrote:
> > 3. The Greek seem to prefer the 'open phi' in their texts, which
> > suggests to use 'open phi' at U+03C6, greek letter phi. For
> > the mathematicians it does not matter if 'open phi' is at
> > U+03C6 and 'closed phi' at U+03D5 (greek phi symbol) or the
> > other way around as long as the assignment is fixed (for ever).
> I would say that maths has a preference for the closed form as plain
> "phi" - the open one is rarely used, in my experience. This is not as
> important as getting the basic Greek alphabet in the right order though.

As I detailed in a (longwinded - sorry) mail some time ago (Ancient Greek)
classicists in the Anglophone world seem to prefer the closed form of theta
and phi, wheras classicists on the continent (ie the rest of Europe)
prefer the open forms. This spilled over into the world of mathematics.
At school in Austria we used the closed forms for the spherical co-ordinates.
In any case, I have seen many articles with the closed AND the open forms.
Mathematicians and scientists tend to run out of variable names...

Greetings, Robert

Robert Herzog
LHC division, CERN (European Laboratory for Particle Physics)
1211 Geneva 23, Switzerland

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:54 EDT