Latin letter GHA or Latin letter IO ? (was: Pre-1923 characters?)

From: Philippe Verdy (
Date: Sat Jan 03 2004 - 13:34:16 EST

  • Next message: Michael \(michka\) Kaplan: "Re: Pre-1923 characters?"

    From: "Peter Kirk" <>
    > The representative glyph for this character seems to be good. But, given
    > that the name is so misleading but cannot be changed, it is good that
    > there is a note "= gha" in the Unicode character charts.
    > But in the light of naming errors like this one implementers should be
    > advised not to use character names, because they are not reliably helpful.

    In Unicode, the glyphs are normative in a way that they allow character
    identification, but they are not mandatory, so they are mostly informative.
    Names are normative, but if they are wrong, we have no other hint to see
    which semantic is good. The "= gha" notice in the charts is also
    informative, not normative...
    In this case, the assignment is weak, and it allows several interpretations.
    Fixing it in a later specification may break applications that have already
    been created to map legacy charsets to this code point.
    What can be done then? Only a more informative notice in the charts, which
    explicits how this character should be used and interpreted... in this case
    the interpretation as "gha" should be more explicit, and a note should be
    given that says explicitly that "IO" is misleading and just relates to a
    possible approximate glyph. My view about the glyph in Arail Unicode MS is
    that it keeps the weakness of interpretation as either "o with a iota hook"
    or as "gha" as a variant of "g" (which was initially what I read when I
    first found it).

    Concerning the character you were initially describing, I had not noticed it
    in the GIF image you sent to the list which also displays the I with hook
    below, which is so close to the angular L letter with a final hook (in fact
    your exhibit shows only the uppercase letter I with hook, not the lowercase
    one). Thanks for correcting it. So be careful when describing characters, as
    I did not notice you were exhibiting the two letters simultaneously. Notably
    because you described lowercase letters (and the Latin GHA/IO letter has
    nearly the same form in both cases, the lowercase being like a smallcap of
    the corresponding capital letter).

    So we have effectively two letters in your image:
    - the "IO"/GHA uppercase letter,
    - and the L-shaped I with a downward hook on the right (which has been
    deprecated in favor of the "b"-shaped I with upward hook on the right, which
    closely ressembles to the cyrillic soft-sign, but which looks more like a
    reversed Gamma with a upward hook connected on its right, and sometimes is
    handwritten like a reversed-esh with a lower loop or curl by people more
    used to the Latin script).

    This archive was generated by hypermail 2.1.5 : Sat Jan 03 2004 - 13:58:48 EST