On Fri, 3 Oct 2003, P&G wrote:
> [...]
> Have you also considered from *h3ekw, the Vedic an-ak blind?
I'm afraid it's an-aks.a'-. I have been wondering if Av. as^i dual 'eyes',
replacing *ac^i in the analogy of us^i 'two ears' is perhaps a true
example of what you are after, i.e. *H3ekW-iH1 (with ntr.dual ending in
the masc. word copied from 'ears'). Trouble is we do not know what comes
out of *palatalized /k/* + added sibilant in Indo-Iranian. Therefore, Skt.
-ks.- and Av. -s^- could conceivably reflect an already-IIr cluster
*-c^s-, and if that analogical fomr had been created before the operation
of Brugmann's Law, a form */oc^s-i:/ would have /o/ in a closed syllable
and therefore treat it as short. It is generally too easy to find a away
to disqualify what looks like a good example.
Jens