> Well, from a brief perusal of the MathML spec at W3 it seems that
> while they do (of course) recognise and need to encode numerical
> types such as complex, real, rational, and interger, I've seen no
> evidence that a datatype has been defined anywhere for expressing
> this in a single attribute.
> So, what do we do?
> (a) Assume that this is something that other TEI-users are going to find
> useful and propose the addition of a new attribute on num (or the
> numericLike class, if there is one) in a future revision of P5
Council did not think this was likely, but we'll find out, I suspect.
> (b) assume that no one else would ever need anything so silly and
> so customize my local schema
No need to make any drastic assumptions -- both customizing your
local schema and (roughly) simultaneously submitting a feature
request for a future revision of P5 are not only not mutually
exclusive, it's expected and encouraged.
> (c) abuse @n to hold the formulaic fraction
I don't like this one.
> (d) import the MathML namespace and use <mathml:cn
> type="rational">1<mathml:sep/>12</mathml:cn> inside a <reg> (vel
> sim.) child of <num>
Ummm... what is it you're trying to do?
> Any thoughts? (d) seems a bit over-the-top to me. (e) other?
So <num value="0.08333333333">1/12</num> really won't do, eh? Could
or some such in your customization (and simultaneously in your
feature request :-)
 Could add facets to ensure absolute value of significand is
between 1 and 10 if you want normalized notation.