P&G wrote:
> Does this compound *k^red-dHeh1- occur outside Sanskrit (where there are a
> number of problems with this root)? The Latin form seems to be k^re: -
> dHeh1.
Does it? Why not *-ed[z]dH- > *-ezd- > -e:d-, esp. in view of OIr.
cretim 'believe' < *kredd-? Of course we also have Av. zrazda:,
zrazda:iti (influenced by <z&r&d-> 'heart'). I don't seem any problems
with <s'rad-> in Sanskrit if my hypothesis is correct. It's <hr.d->
'heart' (PIIr. *3'Hrd- instead of *c'r.d-) that calls for a special
explanation, but if PIIr. *c'rad- in the compound was no longer
synchronically connected with *c'r.d-, any distortions affecting the
latter could have left the former alone (though note the contamination
in Avestan!). One might object that there is no trace of *k^red(i) as a
free-standing locative. But it wouldn't be the only instance of an
archaic case-form surviving only in composition and being otherwise
levelled out, cf. *dems-poti- 'master of the house'. One could imagine
*k^red(i) as the verbal equivalent of placing one's hand on one's heart
when making a solemn promise ("Upon my soul!").
Piotr