*gwh in Gmc.

[David L. White]

> I still think *pempwe is a better fit for Early PIE. Otherwise the Germanic
> forms require an ad-hoc assimilation of *p__kw__ to *p__p__ mirroring the
> Italo-Celtic assimilation to *kw__kw__. But if this happened, why wasn't
> *perkw- affected (Lat. quercus, OE fyrh, OHG forha)? We don't have
> *firf-trees.

        What about Greek /pente/ and Sanskrit /panca/, which do indeed seem
to contain  the respective reflexes of PIE /que/ (more or less) meaning
'and'?
        As for the non-existence of "firf" trees, all I can suggest is that
numbers are sometimes subject to processes (mostly analogical) that do not
affect ordinary words.  In this case, the "sing-song" rhythm of counting may
have encouraged something not unlike reduplication (or internal
alliteration).   (/ini, mini, maini, mo/).

Dr. David L. White