On 2006-01-13 13:53, Joao S. Lopes wrote:
> *h1regW-ro-s > rig-ros, thus preffixed ni/rig-ros >nirgros > nigros ?
> Or, more simply, *rigros > dissimilation > nigros
But in this way both advantages of my solution would be lost:
(1) There is no regular change of *e > i before *g(W)r in Latin.
(2) There is no external cognate (Gk. eremnós certainly derives from
*h1regW-nó-).
Another problem with the first derivation above is that the prefix *ni-
is posited _praeter necessitatem_.
As for metathetic changes involving *n and *r in Latin, Anttila's
etymology of re:nes 'kidneys' is worth another look:
*negWHr- (Gk. nepHrós, Gmc. *neGWr-an-) > *nehr- > *ne:r- > re:n-
(probably influenced by lie:n 'spleen', cf. the variant <rienes>).
Piotr