Symbols are distinct from letters in that the latter have a strong
customary identity that maps onto a sometimes surprising range of font
designs. The functional requirement for a glyph for a letter is that it
is recognizable in context so that the underlying letter can be
identified. Users of Fraktur fonts have no problem with a glyph for A
that in isolation might be mistaken for a U by readers not familiar with
that style.
Even then, not all fonts are usable for all purposes. There are Latin
fonts that drop the dot on the 'i'. Those can't be used for Turkish,
where the dotted and dotless 'i' are distinct. However, they work fine
for English, and they are not that rare. Readers who are not used to
looking for typographical design quirks might not even notice.
For symbols, there is usually a lot less context; they don't form part
of words, for example. And the shapes are often very simple or
geometric. Take a simple triangle, pointing right. Is a black (filled)
one a different symbol from a white (open) one? Is an short, tall
triangle shape a different symbol from a long wide one?
Generally, we say yes, so symbols have a much narrower range of
acceptable (expected) glyph shapes.
Punctuation are somewhere in between. Whether a period is square or
round doesn't matter in the context of running text. Both are equally
acceptable so we typically leave that to the font. At the same time, the
character is reused for any dot on the baseline, whether period or
decimal point.
Common to symbols and punctuation is that they can be mapped onto more
than one concept; this is easier if the range of acceptable glyphs is
narrower. This supports the suggestion made here to look for the
intended shape, and if an existing symbol is a good or perhaps even
precise match, then the suggestion would be to perhaps recognize that
alternate use in an annotation (if that use is considered common and you
want to guide users in making a consistent selection).
Yes, it's useful to look at a couple of the most common fonts to make
sure that the actually deployed range of glyphs matches the new usage.
In case where some symbol unexpectedly shows an interesting variation of
appearance, adding another use to it might not work. In particular, if
these are not outliers, but common alternations. But unless that
research has been done and there's conclusive evidence that adopting an
existing symbol for that use case is unworkable, there's not even enough
basis for discussing a new character proposal.
That said, I'm not in favor of adopting an existing character if the
expected glyph for it is only a rough approximation of a preferred
shape. I totally get that not all arrowheads look the same, and that
there is room therefore, for a variety of them in the standard. However,
any proposal claiming that every single existing one is insufficient has
the burden of demonstrating that.
A./
On 7/27/2025 10:56 AM, Doug Ewell via Unicode wrote:
Alex Plantema wrote:
Characters may look useful in one font, but not in other fonts. Are
you going to check them in all fonts?
A character may have somewhat different appearances in different fonts, as long
as the basic identity of the character is preserved. That’s what characters are:
https://www.unicode.org/reports/tr17/#CharactersVsGlyphs
Even emoji don’t look exactly the same in every font. Even characters in the
Dingbats block don't.
If you require such a specific appearance for this symbol that "check them in all
fonts" is considered necessary, then you are not looking for a character; you are
looking for a glyph, and for that an inline image is probably the best choice.
--
Doug Ewell, CC, ALB | Lakewood, CO, US | ewellic.org
