What about Albanian, Piotr? that presents Satemised
form together with non-satemised forms:
(1) Albanian <zjarm> 'fire' < IE
<*gWhermos> [*gWh > zj, satemisation???]
(2) Albanian <pjek> 'I roast' < IE
<*pekWo:> [regular *kW > k]
Albanina <djeg> 'I
burn' < IE <*dhegWo:> [regular *gWh > g]
Albanian <pesë>
'five' < IE <*penkWe> [*kW > s ?????]
(3) Albaninan <thome, thotë> 'I say, he says'
< IE <*k^emi, *k^eti> [regular *k^ > th]
Albanian <dhëndër>
'son-in-law' < IE <*g^em@...> [regular *g^ > dh]
(4) Albanian <kam> 'i have' < IE
<*kap.mi>
Another question, Piotr, we must conclude from
these data that reconstructions with two "gutural" series must be
abandoned?
David Sánchez