Syd Bauman wrote:
>
>     <formula notation="http://www.w3.org/1998/Math/MathML">
>   
...
>  My instinct is that this is not attribute abuse, but it isn't
> right, either. That is, it's OK to put a namespace on a data.code
> (especially if it actually resolves to something), but it would be
> much better to have a proper namespace attribute, instead.
>   
If it's not right then it *is* attribute abuse.   And, imho, it is not 
right: we don't define notation as pointing to a namespace. We might, 
but we don't.