O dear, I think this list moves too fast for me to keep
up. Anyway, I have gathered some more material to back
up my suggestion:
re sanskrit nouns on -ij: Mylius knows of only 3 besides vaNij-:
Mayrhofer (etym. Woerterbuch des Altindoarischen)
connects nirNij- with NEJ (*neigw) and Rtvij- with YAJ
(*Hiag') while bhurij- has no etymology at all.
for vaNij- Mayrhofer does give "Gueter gewinnend" from
*van-ij- < *ven-Hg'- even though he has no explanation
for cerebral N (vaNij < vanij)
Now, the fact is that vaNij is a real word in sanskrit
and does mean "merchant" whether we like its formation or
not. Consequently, if we can make a formal (phonetic)
connexion with wanax, I will not quibble over a semantic
shift from "merchant" to "priest" (or vice versa) because
we know these things happen, and I will prefer a IE etymology
_with_ _comparative_ _evidence_ from other dialects any day
over a loanword etymology, never mind semantics. (because
way too many words have been labelled Pelasgian and what not
in the past. I know loanwords are real, but for reconstruction
they must be a last resort). Still, do note that LIV gives
as the meaning for *ven "ueberwaeltigen, gewinnen" (oe. 'winnan'!) _not_
"desire"; the original meaning does not seem to have been sexual.
So *ven-Hg'-t- stands or falls with the formal connexion
of wanax with vaNij. I agree it's not perfect. The
cerebralization is an indian thing and may be done away
with with some handwaving (Mayrhofer: "wohl mit
kontextfreiem -N-"). Greek -t- (absent in vaNij) is a
popular member of root compounds and also accounts for
greek -k- (rather than -g-). The only serious objection
is Nom. Pl. Anakes which would need to be secondary (-k-
but no -t-). But why not? Anax, Anakes would sound ok to
a greek anytime.
>Tocharian has (B) ñakte, (A) ñkät, Gen. ñäktes 'deity', sometimes
>connected with *wanakt-, as Mark Odegard once pointed out on this list.
clearly I would welcome a tocharian wanax!
Douglas Adams (dictionary of toch. B):
"~nakte 'god' [voc. ~nakta often used as a respectful address
to a king] [...] V[an ]W[indekens] suggests a derivation from PIE
*h1nek- 'obtain, take' (Sanskrit ashnoti [...]
Normier, however is probably right in taking PTch *~n"ak("a)te
to reflect virtual PIE *ni-g'huH-to- '+/- the one called down'"
this seems good enough to prevent us from bending ~nakte to
venHg't- (where is the v-?). So I don't think tocharian will
help me any here -
I'll agree my case is not very strong and I would never
stand up and defend this as the one correct etymology.
But I think the evidence is good enough to keep it in
mind as a possibility.