I've never believed that t>ts anywhere in PIE, so I
haven't consistently organized my evidence against it.
With this discussion I've been surprised just how
much there is to say on the subject, so I'll be
presenting my argument systematically over time.
*ke>ka (optional)
*kart+ 'bend (wicker), weave'
*kàrttló+s kr,ttl(e)+ 'basket'
*k(a)rttí+s 'weaving, wickerwork, shrubbery'
*k(a)rttís+Ló > *k(a)rttiLós 'small _'
The PIE form of these words can't include *tst
because their outcomes often show dissimilation of one
*t>0, leaving the other as t, not *ts.
*karttis > *kar_tis> *kra_tis > cra:tis
*kàrttlós > *kàrt_los > *kàrtllos > kártallos
The difficult cluster *rtt is fixed in Latin by
removing the first t, the mora left caused metathesis
so the V could link to the mora and become long (long
r not allowed at the end of syllables).
The difficult cluster *rttl is fixed in Greek by
removing the second t, the space left was filled by
the *l linking to the space and becoming long (long l
not allowed after C so V inserted).
*kr,ttís > *kr,stís > *xurstiz > hyrst
*kr,ttís > *kr,_tís > *k_,rtís > *kurtís > kurtía
In Germanic t>s before t, as in most. This is
preserved after r, probably because it caused
retroflexion of s and only *st>ts later.
The difficult cluster *rtt is fixed in Greek by
removing the first t, the mora left caused metathesis
so the r moves to the mora and leaves an empty nucleus
which is filled by a V (u was the V associated with
retro. at the time).
*karttiLós > *kart_iLós > *kertLyós > certle 'ball of
yarn'
The difficult cluster *rtt is fixed in Old Irish by
removing the second t, the space left was filled by
the *L moving to the space and i>y before a V per
lasting constraints.
The Sanskrit ev. is consistent, but more difficult
to describe.
____________________________________________________________________________________
Park yourself in front of a world of choices in alternative vehicles. Visit the Yahoo! Auto Green Center.
http://autos.yahoo.com/green_center/