On Wed, 10 Oct 2001 22:21:32 -0000, "Piotr Gasiorowski"
<
gpiotr@...> wrote:
>We have *-sor- in *t(r)i-sor-es as well, but it is not parallelled by
>**t(r)i-wor-es.
A case, not really convincing but better than the non-existing case
for an *n in "4", could be made for a *w in there somewhere,
sometimes. We have OIr. trïar "three men" < *triwirom. Maybe rather
suggestive of another explanation for *kWet-wor- are the forms meaning
"thrice": Av. Þriz^vat~, ON Þrisvar, OHG driror, OE ðriwa, ðreowa, and
"ein wo-Suffix auch in av. Þrisva- n. "Drittel" und gr. thri~on
"Feigenblatt" aus *trisFon" (IEW 1091).