> Haukur, thank you for your reply. I had also developed the theory
that /z/ could have still been a voiced sibilant at the time of the
second voicing. It seems the most probable thing to me. What I really
hate is that scholars who treat the subject of second voicing do not
mention the possible reasons why /s/ is excluded from voicing, but
simply state that it affected only /f, th/.
Anyway, I feel safer now. Thanks again.
Yours, Lily