Glyph difference between Em dash and Horizontal bar

Peroyomas's picture

Is there any glyph difference between 'Em dash' (— U+2014) and the 'Horizontal bar' (― U+2015)?
There are a number of dashes encoded on Unicode, but this is one of the few that is not very clear its usage. Unicode calls the horizontal bar 'quotation dash', yet is not clear why it was encoded differently, as all bibliography I've read makes no distinction between the dash used for parenthetical text and quotations. Fonts I've seen makes horizontal bar longer, shorter or both characters the same. Recurrently the em dash connect to the sides of the character width, but the horizontal bar not. A difference in the characters properties is that em dash makes line breaks while horizontal bar don't—which may make sense as distinguishing them semantically, but don't give much hints for the glyph side.
I probably won't be very interested by this, but the horizontal bar is included in an old Windows character set for Greek, along Em dash. I'm not sure if is used down there or not.

Peroyomas's picture

I've read those before but they really don't answer my question. Besides perhaps Math characters, Unicode don't encode character based on the semantic meaning and if does something like encoding specifics "glyphs" is for compatibility with other stuff. The em dash don't seems to having to look any different when used either for parenthetical text nor quotations (I speak Spanish and I use it regularly for both) so I'm intrigued about what is the purpose of the 'horizontal bar' at all.

Té Rowan's picture

It's used as a direct quotation device in several typesetting trads, hence its other name, 'quote dash'. What little I have found about it says it should be some 2 ems or so. Usage:

--- Text text text, said Foo. --- More text.

Birdseeding's picture

In Sweden (and I believe other Scandinavian countries) an En Dash is used instead of an Em dash for the conversation dash. I'm assuming it has a separate codepoint so as to enable coutry/language-specific alternates without messing with the "real" Em dash? (Not that it matters much. In Sweden the Em dash is as far as I know only used for one thing these days – omitting the author's name in repeated references in a bibliography.)

kentlew's picture

The length of an em dash in contemporary typefaces these days tends to vary quite a bit and often does not measure an actual em. In my own work, for instance, I generally prefer an em dash which is on the order of 75–80% of an em, and I include modest sidebearings. This seems to work well for contemporary usage (current American practices, anyway).

However, there are some traditions that prefer (or even demand) a nominal em dash — i.e., one that measures precisely a full em. I believe Spanish is one of these.

And there are some situations where a full-length, nominal em dash may be preferable — building three-em dashes for bibliographies, for instance.

The norm in Font Bureau fonts these days (I hesitate to call it a standard) is to provide both a “designed” em dash, according to the type designer’s preference, as the default, plus a “nominal” em dash which measures a full em and has no sidebearings. The default is, of course, encoded u+2014, while the nominal em dash is encoded as u+2015 horizontal bar. We provide a {locl} substitution for Spanish, as well as a dedicated stylistic set {ss18} for accessing the nominal dashes, for those who prefer.

(We provide a similar pair of en dashes, but the nominal en dash is not assigned a unique code point.)

Syndicate content Syndicate content