Trubetzkoy (1930, 89) compares the WC form with PEC *=ĭmʒ́Ĕr 'to bake, roast' - which should be rather compared with PWC *źʷV id. The correspondences between PEC *=ĕrc̣V and PWC *zʷa are quite regular; the original meaning must have been 'to heat' (whence both 'to roast' and 'to boil' can be easily deduced). [There is a possible parallel in Lezghian: PL *ʔä(r)c̣:V 'to dry', reflected in Tab. je=c:-uz, Ud. ec:-esun. The consonantal correspondence, however, is quite irregular: we would either expect PD *=erc̣- and PL *ʔä(r)c̣V, or PD *=erz- and PL *ʔä(r)c̣:V-. Perhaps, original *=erz- in PD was influenced by *c̣a 'fire' and obtained a secondary glottalisation? In that case we should rather reconstruct *=ĕrʒ_Ŭ for PEC. Note that PN *q-arc̣- and PWC *zʷa can go back to either of these protoforms.]