On 2007-06-14 02:46, tgpedersen wrote:
> Let me rephrase: does "nom.sg." appear in the rule text
> (this is just a complicated way of asking whether the rule is ad hoc)?
It is not ad hoc. It's highly regular, and if it's practically
restricted to the nom.sg. of thematic stems it's simply because *-ros is
difficult to find anywhere else. The *-os apocope should also be
expected in the declension of neuter es-stems, but unfortunately neuters
of the shape <XCVrus> or <CVCrus> (X = a syllable) just happen not to be
attested. Note, however, that we have apocope in *bHaros > *faros > far
'emmer', which behaves like <vir>, and that here the dropped *-os is
part of the stem, not a case ending. Cf. also a similar reduction in the
distributive numeral quater < *kWatrus < *kW&twr.-s (cf. Ved. catuh.,
Av. caTrus^).
Piotr