» >I have seen examples of both types of epsilon being used to denote set
Lars Hellström :
» No doubt due to "limitations in past typesetting".
Well, that should be checked against Cajori. There has been a time
where the root sign was `r.', i suspect that epsilon (whatever form,
it was on manuscript first...) was the former form of the belonging
to sign. Of course, printing technologies have ironed out some
notations that could come back when it was enhanced (back to the
roots, which I had studied at lengths in Cajori... the precedence of
the form <sqrt sign> (fenced math expr) rather than using the rule
over to denote the scope of the root was imposed by the cost of the
second form. In this respect, reading the book by Chaundy et al. gives
a very interesting picture of the monotype time, when mostly anything
was possible, but any requirement for a special symbol (thus punches
design, etc.) or layout could slow down the process and raise the
costs in discouraging scales.
» >Further, if you want to make it impossible to use \varepsilon and \epsilon
» >side by side in the same document, you will have to make sure that in all
» >of the world literature in the past up till now it has never been used that
» >way, because that is how the requirements of Unicode were set up.
» I'm not saying that it should be completely impossible to use them side by
» side (even though I would question any attempts to do so), but they
» shouldn't be provided as distinct characters in the default set-up.
Yes, I mostly agree. Making it hard for authors to use in a same paper
two versions of epsilon or phi is a good thing for the
readers. Allowing them to choose which shape they prefer is no
problem, especially if the publisher can override this with his
styles. Honestly, when i read maths with a \varphi next to a \phi, i
never know whether it's keying error or semantic subtility. The
problem being that any available glyph in a standard font set will be
used by mathematicians... On the other hand, pi and varpi should
cohabit, but maybe with a more semantic name (pi/doricpi?).