Werner Winter (personal communication)
suggested that *potni:h2 > *podni: by irregular voicing assimilation, and
that *podni: > *po:dni: via Winter's own Law, then > *po:ni: > *panji.
I don't think he was quite sure himself that it was a plausible derivation. West
Slavic *panji 'lady, Mrs' cannot be divorced from *panU 'lord, Mr', and treating
the latter as a back-derivative of the feminine form would be a
desperate idea, given the Proto-Slavic social realities. The words pan and pani
had (as you observe) an initial h- in older Czech, which suggests the mysterious
prototype *gUpan- (Trubachev derived it, perhaps too romantically, from
[attested] Iranian *g(a)u-pa:na- 'cowherd' > 'protector' > 'lord'). Here
for once Bankowski is probably right in suggesting that pani < haplologically
shortened *gUpan-ynji (with a very popular feminine-forming suffix), just like
Polish ksieni 'abbess', originally 'duchess' < *kUnINg-ynji (cf. uncontracted
Ukrainian knjahynja).
Piotr
----- Original Message -----
Sent: Monday, January 15, 2001 7:49 PM
Subject: Re: [tied] Re: About the etymology of *nepo:t-
"nephew/grandson"
Speaking of which, I have this crazy idea that Pol.
<pani> is from
*potnih2. I know about all the objections (potnih2
should have given
poni, loss of -t- before -n- is not known to cause
compensatory
lengthening, Old Czech <hpán>, etc.), but it would be so
nice if it
were
true...