>the Latin synchronic rule -VGt- > -V:Ct- (G = voiced consonant, C =
> voiceless consonant), e.g. in the formation of past participles.
This is Lachmann's Law, which has been widely discussed in the literature.
There are many examples, and several counter-examples. The consensus at
the moment is that it is a chimaera; there is no such phonological rule,
and the lengthening of the vowel is due to analogical pressure from the
perfect stem. A long vowel in the perfect arises from several sources, and
it is now claimed that a long vowel in the -to forms is found when there is
both (a) a long vowel in the perfect active, and (b) devoicing of a
root-final stop.
There are of course exceptions even to this, and unfortunately, some of the
examples used to prove this new theory are based on -V:s- versus -Vss-, for
example fu:dit ~ fu:sus < fud-to, and fidit ~ fissus < fid-to. But we
know there was considerable variation between -V:s- and -Vss- in any case,
so the fact that there was analogical pressure here does not in itself
disprove Lachmann's Law.
So the consensus is "analogy" - but some of us dreamers leave open the chink
of a hope that Lachmann may have been right!
Peter