> >To which one might argue that whoever designed Latin orthography
> >didn't want a stem to alternate as much as *-em, *-in-, and therefore
> >made an allowance for "etymological spelling" in such cases.
Latin orthography was not "designed" in that sense, but reflects the actual
pronunciation at a certain period, and changed as the pronunciation changed.
Therefore if an original *-em *-in- stem noun were altered to -en -in-, it
would be speakers who did it, and the cause would be analogical levelling.
But there is no evidence for such a paradigm (*-em *-in-) and a lot of
evidence against it. So I think we're stuck with Latin distinguishing PIE
*m. from PIE *n. like it or not.
Peter