Although I can't locate the posting in
question, I think I once conjectured on this list that the
phonetically regular development of final *-o:r in Proto-Slavic was *-y,
(whatever the intervening stages), parallel to *-e:r > *-i (as in *mati
'mother', *dUkti 'daughter') and *-o:n > -y (as in *kamy 'stone'), cf. Lith.
sesuo 'sister' < *sweso:r. Analogy and the early elimination of the *-r-
declensions have bled Slavic of nearly all potential examples, but *-y- was
occasionally generalised (and has survived) in non-final positions, as in
*c^etyre 'four (m.)' instead of expected *c^etwore, influenced by *c^ety (n.)
(itself undocumented) < *kWetwo:r, or in *pastyrjI 'shepherd' for
hypothetical *pasty < *pa:sto:r < *pah2s-to:r (cf. Lat. pa:stor), with *r
restored from the inflected case forms as the consonantal stem was moved to a
productive declension (the extremely productive agent-noun suffix *-arjI,
borrowed from Germanic, may also have influenced its form).
If the above makes any sense at all, and if
the derivation of Slavic *pero 'feather' from *pterom (Gk. pteron) is correct,
why not analyse the Slavic 'bat' word, *netopyrI (or was it *-pyrjI ?), as
descendent from *nekWto-pto:r, where *-pto:r would represent the compositional
form of *peto:r 'feather (coll.), wing', of which PGmc. *feþro: is a derivative.
Though *neto-pyr- is routinely analysed as 'nocturnal X', where the most natural
interpretation of 'X' would be something like 'flier', I've never seen
a convincing etymology of *-pyr- as 'flier, flutterer' or as anything else for
that matter.
Any comments?
Piotr