From: John Hudson (john@tiro.ca)
Date: Mon May 07 2007 - 23:27:41 CDT
Michael Everson wrote:
> See http://www.evertype.com/standards/iso10646/pdf/sharp-s.png where I
> have given an F, J, U, ß, long-s, and two capital sharp esses. The
> second of these is the one under ballot, made in discussion with Andreas
> Stötzner, who suggested taking the stroke and curve of the U and the
> finial of a J. I don't find it unpleasant or inappropriate. I tried
> hacking one based on an F but I like the U-based one better.
You've taken a romantic type with a strong vertical axis and an expansion based stroke
pattern and introduced a letter with a very dominant translation based stroke. Although
these are nominally upercase eszetts, the vertical proportions employed in the treatment
of their right side is clearly based on the proportions of the lowercase letter. Why is
the waist of the uppercase eszett at the same height as the lowercase? There is nothing
else in the uppercase alphabet to which this corresponds.
Leaving aside questions of pleasantness and appropriateness, what these examples
demonstrate is that a kind of bits-and-bobs construction approach to this letter can be
applied in a certain kind of typeface if e.g. it has a certain kind of J. While there is a
certain amount of modularity in a typical Latin text face, that modularity is ultimately
derivable from similar construction patterns in written forms, not from copying details
from different letters to form new ones. And while it might be possible, with some further
work, to make a convincing uppercase eszett in a typeface with a romantic vertical axis
and expansion based stroke contrast, this is very different from being able to make
convincing and harmonious letters in several thousand different typefaces across a very
wide range of styles.
I spent some time today drawing uppercase eszetts, and found it a pretty gruesome
business. In the lowercase form, the transition from the top arch into the right side of
the letter is something that takes place in the zone between the ascender height and the
x-height. In the definition of lowercase letters, this zone is an important one, and what
happens in it determines whether the eszett will harmonise well with other lowercase
letters. In capital letters, there is no corresponding zone, so the uppercase form must be
divorced from the proportions of the lowercase; failure to do this is probably the major
reason why some commentators on the encoding proposal have complained that most of the
examples look more like lowercase letters than uppercase letters. The best success I had
in my experiments was to base the right side of the uppercase eszett on the uppercase S,
and to make the reversal of direction where the top arch meets the S part very high and
abrupt; the kind of double-bowl shape one finds in the lowercase in romantic and
rationalist types, or the flowing form found in humanist types with a greater
ascender:x-height ratio, are quite out of place among the uppercase letters. And if a
typeface lacks any kind of ball or lachrymal terminal in the uppercase, as is often so,
then one can't introduce it for this letter and must allow a vertical serif for consistency.
I tried to Trajan test: can this letter be made in a convincing way that harmonises with
the Roman inscriptional lettering, i.e. with the source of our uppercase alphabet? I'm not
convinced, but I am sure that to work at all it must be made completely without reference
to lowercase details or proportions: http://www.tiro.com/John/TrajanEszett.gif
Of course, none of this serves as or is intended as an argument against encoding this
particular character. It is a general grump about encoding as parts of writing systems
characters that have not followed the evolutionary path of actual writing.
John Hudson
-- Tiro Typeworks www.tiro.com Gulf Islands, BC tiro@tiro.com We say our understanding measures how things are, and likewise our perception, since that is how we find our way around, but in fact these do not measure. They are measured. -- Aristotle, Metaphysics
This archive was generated by hypermail 2.1.5 : Mon May 07 2007 - 23:29:55 CDT
