Hello Vasiliy,
> >> Yes, it works in Welsh (though in modern only):
> >
> >Only in ModW? That's a curious thing - I would have supposed that such
> >things go back to the common ancestor of both Brythonic & Goidelic
> >tongues. An innovation?
> The mutations in both branches are not immediately related (AFAIK).

This seems to be true

> Moreover, the use of mutations in Welsh and Breton seems to be different
> (e. g. in combinations with possessive pronouns). Which may mean that
> the two langs got separated when the phonetic processes that resulted
> in mutations were still active. At any rate, I don't see much space for
> anological levellings here. But I may be wrong; I think there must exist
> some literature on that issue. Pavel?

Yeah, the 4th century was a time when the lenition processes were only
starting to appear. The mutations in Brythonic and Goidelic obviously
differ. As we don't know much on Gaulish mutations, I can't say anything
(though I seem to remember reading that the extant Gaulish fragments exhibit
no sign of any kind of intervocalic voicing and such things). As for Breton,
I can't answer your question, as I know next to nothing about it. Feel free
to ask about Welsh, though.

As for literature, right now I can't give you any special article or some
such, but the books to look for will obviously be Morris-Jones ("A Welsh
grammar, Historical and Comparative"), "A Concise Comparative Celtic
Grammar" (I don't remember the authors, it's a shortened version of some
German imprint), Kenneth Jackson "Language and History in Early Britatin"...
hey, I presume you know about these books!