In talking about the idea that IE neuter u-stems were originally r/n stems, I made mention of my point that one would expect these stems to end in -h. (visarga) in Sanskrit and Vedic, since final *-r becomes -h. in these languages. However, I have since realized that I am at least partially wrong about this statement: the words in Sanskrtit/Vedic that end in -h. from a hypothesized *-r actually go back to an ancient *-rs, viz. in consonant stems such as gi:h. "praise" (stem gir-) and in the 3rd pl. ending of the perfect -uh., which goes back to *-r.s. So Miguel Carrasquer Vidal, your idea that an original *-r was simply dropped may be more correct than I originally thought. (However, r-stems in *-e:r and *-o:r go back to original *-ers and *-ors, and these do
not have -h. in Sanskrit, simply rather -a:. It may be that only original syllabic *r. before final *s gave rise to -h. in Sanskrit, and not diphthongal *-ers and *-ors, which early lost the *-s and later lost the *-r in Sanskrit.) However, why don't these stems preserve /n/ or /r/ in their declension in languages outside of Sanskrit/Vedic and Armenian (unless there are examples that I am unaware of)?
Andrew Jarrette