RE: The glyph of the CAPITAL SHARP S

From: Philippe Verdy (
Date: Thu May 10 2007 - 14:07:28 CDT

  • Next message: Andrew Lipscomb: "Re: How is the glottal stop used in some languages?"

    Remember that the sharp s is already derived from a long s, and that a long
    s is also an s semantically; there is also a separate proposal to make a
    capital long S which is most likely to adopt a simple “turned J”
    representative glyph (with no descender, like J)

    Unicode has made all its Latin letters adopt a modern Times-like style, and
    this includes *very light* serifs and variable stroke weight (with verticals
    bolder than horizontals.


    Changing from a long s to an S for the first part of the glyph would cancel
    the origin of the letter. If you want an S, it’s just enough to show it on
    the right part, given that DIN also suggests to make a compatibility
    decomposition (for NFKC) into a double S to match the current usage in
    Germanic languages.

    But some questions remain about the usage of the glyph as a ligature (not as
    a letter) in other languages than High German: if you want a SZ semantic,
    this is most likely for another language where sharp S is not a letter but a
    ligature, to be handled as a font style (like it is when handling œ/Œ or æ/Æ
    as letter pairs in French and English, but as a a separate letter in some
    other languages).


    Given that most usages of the character will be as a distinct letter, not as
    a letter pair (because its ethymology may link it to other pairs than a
    double S, such as SZ or even TZ that have been unified into a single form)
    it can’t be a ligature: it has its unique meaning, and its best to keep it
    linked to the way it was formed in lowercase, i.e. keeping the unconditional
    long s form in some way; how would you capitalize the long s? The arch of a
    turned U is the only solution that keeps readability and avoids the
    confusion with a capital I.


    However, if the right part should adopt the S form, this S should really use
    the capital height, to make the final ligated glyph look like a true

    So the only question is about which right part to adopt: an S, a Z, a Yogh,
    or an Ezh?

    Given the current usage, and the fact that Unicode/ISO10646 representative
    glyphs have unambiguously favoured the modern usage by chosing a modern font
    style, only the form that maps directly to the modern usage is suitable
    here, i.e. an S for the right part.


    This does not remove the possibility to make a different choice for
    older/traditional font styles. And may be we could allow encoding optional
    variants for letter forms based on <long S+Z>, <long S+Yogh> and <long
    S+Ezh> (but with caveats, because these are used as typographic ligatures
    for other languages than German, including those that use Yogh or Ezh as
    part of their orthography.)


    It’s true that, historically, the <long S+Ezh> was used, but this is in
    documents printed before the complete changeover in the German typography
    after WW2 (and before WW2 there was no real tradition in Germany for a
    Times-like font style, given that older styles like Fraktur was still
    popular, as well as specific Germanic styles used in cursive handwriting,
    and widely advertised during WW2 as a distinctive and more Germanic style,
    for political reasons, rather than stylistic reasons.)


    Today, there’s no Yogh and no Ezh in German, why would we choose them for
    the capital sharp S? For the representative glyph of the small sharp s, they
    were not used, and the lowercase z was excluded too. So the lowercase s was
    clearly chosen to match the current usage as meaning a double s.


    If we were to encode variants for the capital sharp s, the same variants
    should be encoded too for the lowercase letter to make case mappings equally
    valid and reliably mappable to each case.


    The chosen glyph must also be easily mappable to other modern styles,
    notably Helvetica/Arial/Verdana-like sans-serif styles, without creating
    confusion (another reason why the right part should not use a reduced S, but
    should take the full height to avoid the ambiguity). This is only possible
    if we don’t allow the arch of the capital long S on the left part to slide
    down on the top right (otherwise the letter will not be drawable with a
    single contour, without raising up the pen, and thus would break the strong
    ligation of the two parts into a single letter, or would require reducing
    the size of the S on the right part, making it confusable with a lowercase




    De : [] De la
    part de George W Gerrity
    Envoyé : jeudi 10 mai 2007 16:17
    À : Unicode Discussion
    Cc : George W Gerrity
    Objet : RE: The glyph of the CAPITAL SHARP S


    I don't find any of the proposals too satisfactory: the Capital Sharp S
    glyph should be derived from a ligature of Latin Capital letters SS or SZ,
    not from a cursive lower-case ligature or from a lower-case long s. The
    problem is that neither pair form a nice ligation, although I did try the


    However, suppose we use the glyph form Latin Capital Letter Ezh (U+2169) for
    the Z. In that case, a rather pleasant-looking and unmistakable glyph can be
    formed with Latin Capital S on the left, Latin Capital Letter Ezh on the
    right, with the ligature on the top.


    Could one of the typographers paste together such a glyph using a serif font
    style — with several variations — for comment?




    Dr George W Gerrity Ph: +61 2 6386 3431

    GWG Associates Fax: +61 2 6386 4431

    P O Box 229 Time: +10 hours (ref GMT)

    Harden, NSW 2587 PGP RSA Public Key Fingerprint:

    AUSTRALIA 73EF 318A DFF5 EB8A 6810 49AC 0763 AF07


    Orange vous informe que cet e-mail a été contrôlé par l'anti-virus mail.
    Aucun virus connu à ce jour par nos services n'a été détecté.


    Orange vous informe que cet e-mail a été contrôlé par l'anti-virus mail.
    Aucun virus connu à ce jour par nos services n'a été détecté.


    This archive was generated by hypermail 2.1.5 : Thu May 10 2007 - 14:08:58 CDT