Is there any other reason for
reconstructing '7' as *septh3m? But if there were, the
cardinal would have become *hebda in Greek, wouldn't it? So, apparently,
*sebdh3mos comes from *septm.mos restructured on the analogy of
*og^dh3mos, which means that we can't use <hebdomos> as independent
evidence of *h3-induced voicing. This leaves us without an explanation for
*sedmU '7th' in Slavic, for it's the form of Slavic '8th' that is itself
analogical: *osmU, with *m from the preceding ordinal but _no voicing_! Since
*aCtmas goes back all the way to Proto-Balto-Slavic, one can hardly argue that
perhaps Proto-Slavic once had *ozd(v)U '8th', which caused analogical voicing in
*se(p)tmU and then was replaced by *os(t)mU (why not *oz(d)mU??): in the
light of all available evidence, *os(t)mU was the only form Proto-Slavic ever
had.
Piotr
----- Original Message -----
Sent: Saturday, September 21, 2002 4:22 PM
Subject: Re: [tied] *gWerh3- "to devour"
The evidence is indeed slight, because the frequency of *h3
is much lower than
that of *h1 or *h2. I would like to point out once
again the voicing effect of
*h3 in the ordinal numerals 7 and 8 (*septh3mós
-> *sebd&3mós [Grk. hebdomos,
OCS sedmU-]; *Hok^th3wós ->
*ogd&3wós [Grk. ogdoos]).