> Looking in the archives I see that you've been answered before.
I wasn't asking a question.
> I don't think you can take one word as evidence of kW/p variation
> related to later changes in IE languages.
In this case, I think you can.
The akWa/apa alternation recurs in many European river names.
W.P. Schmid
Alteuropa und das Germanische
in
Beck(ed.)
Germanenprobleme in heutiger Sicht
juxtaposes the river names
Alpe, NWGerm. Alupe,Lith. / Albæk, Den.
Asphe, NWGerm. Asupis, Lith. / Asbæk Den.
Gellep, NWGerm. Geldupe, Lith. / Geldbæk, Den.
Marpe, NWGerm. Marupe, Lith. / Marbæk, Den.
Vidapa, NWGerm. Vidupe, Lith. / Vidå, Den.
Elsewhere, Germany has river names in -ach.
Obviously, a word with such a wide geographical distribution must
belong to a language which must have mattered one way or another to
invading IE-speakers. Since the distribution of apa and akWa names
doesn't follow any known IE borders it must have been a substrate
language for large parts of Europe, thus giving rise to loans of or
preferences for p- or kW- variants later on.
> For 'river, water?' I'd say
> that *xakYw+ with reduplicated *xaxkYw+ > *xaxpY+ (this assumes that
> h2 = x = velar fricative).
Reduplication is not productive in IE nominal morphology. It's a verb
thing.
Torsten