Medievalist ligature character in the PUA

From: Wolfgang Schmidle (
Date: Tue Dec 01 2009 - 05:04:51 CST

  • Next message: John H. Jenkins: "Re: Medievalist ligature character in the PUA"

    Dear list,

    I am working with a Latin text from ca. 1500, which contains a set of
    medieval characters. For instance, it contains the abbreviation qꝫ (q
    and "letter et" U+A76B) for "-que" in the form of a ligature. The MUFI
    Initiative has assigned the PUA codepoint U+E8BF to this ligature. Thus,
    an appropriate font (such as Andron or Junicode) can display my text.

    However, I would like to avoid PUA codepoints. In the examples in
    document N3027, virtually all instances of the "letter et" are in fact
    instances of the ligature character. Thus it seems that the ligature is
    not supposed to be assigned an official codepoint in the foreseeable

    In ch. 16 of the Unicode 5.0 book, it says that the "zero width joiner"
    U+200D requests ligatures, e.g. "f+200D+i" should result in a "fi"
    ligature. Although 200D seems to be meant for languages with mandatory
    ligatures such as Arabic, the Latin "fi" is explicitly given as an
    example (p.539).

    Consequently, in a font that contains the ligature glyph at e.g. U+E8BF,
    the sequence q+200D+A76B should ideally be displayed by the same glyph
    as the single codepoint E8BF, i.e. a rendering mechanism for this font
    should replace q+200D+A76B by E8BF. Normal fonts would display the
    sequence as qꝫ instead of the ligature.

    Unfortunately, the rendering doesn't seem to work this way. And even in
    a sophisticated font such as Hoefler Text, the "zero width joiner" has
    the opposite effect: While "fi" is by default displayed as a ligature
    and thus looks exactly like the "latin small ligature fi" U+FB01, the
    sequence f+200D+i is not displayed as ligature, but as the normal
    sequence "fi" with ligatures disabled.

    Is there no implemented rendering mechanism for 200D in the Latin
    alphabet? Or is 200D not meant for cases like these? Is there another
    way to avoid the PUA codepoint?



    This archive was generated by hypermail 2.1.5 : Tue Dec 01 2009 - 11:19:27 CST