Re: U+2015 HORIZONTAL DASH, 2-em and 3-em dashes in the Chicago Manual of Style

From: Jukka K. Korpela (
Date: Tue Jan 26 2010 - 14:31:15 CST

  • Next message: Michael Everson: "Re: U+2015 HORIZONTAL DASH, 2-em and 3-em dashes in the Chicago Manual of Style"

    Michael Everson wrote:

    > Normally one just uses two or three —— and ——— and kerns if there is
    > any incidental space between them.

    By “kerning,” you mean manual tuning of spacing between characters, not the
    automatic kerning that rendering software may perform when a font contains
    kerning information for character pairs.

    Well, I don’t do kerning on your message, where I clearly see dashes
    separated by a small amount of empty space. Actually, most people have no
    idea of kerning, automatic or manual.

    Maybe the lack of 2-em and 3-em dashes as characters in Unicode is based on
    the assumption that successive em dashes are joining, either because they
    have been so designed by the font designer (i.e. the advance width is
    exactly the same as the dash width), or because automatic kerning is
    specified and used. But these assumptions often fail.

    This may be related to the American (U.S.) practice of using an em dash for
    punctuation so that there is spacing between it and adjacent letters. This
    had led to font design where em dash has no intrinsic spacing on left or on
    right, so consecutive em dashes form a continuous line. However, some
    typographers have started favoring a design where the em dash has some
    intrinsic spacing, so that it does not hit adjacent characters.

    James Felici writes in “The Complete Manual of Typography”: “Fonts that do
    offer punctuating em dashes typically include kerning values that draw them
    together when dashes are set consecutively, allowing them to be used as
    joining em rules.” But this is too optimistic. There are many situations
    where kerning is not applied, or even cannot be applied. (Word processors
    normally don’t do kerning for copy text, and web browsers haven’t done any
    kerning, though Firefox seems to have some poorly documented or undocumented
    kerning features.)


    This archive was generated by hypermail 2.1.5 : Tue Jan 26 2010 - 14:37:11 CST