Piotr Gasiorowski wrote:
> I can only say that Szereményi's rule predicts *bH(o)rw-ós.
P.S. And given its vocalism, it could be an acrostatic u-stem, i.e.
*bHór-u-s, gen. *bHér-u-s (then typically transformed into *bHer-w-os or
the like). All these maybes don't change the fact that the word is not
reconstructible all the way back to PIE. We don't even know if the root
vowel was really *o (rather than *a).
Piotr