In PWC the root has a frequent labial prefix (probably originally a class marker). In EC the original structure of the root was best preserved in Av.-And. and Nakh; in most other languages, where the root was treated mainly as an adjective, it obtained a productive adjectival (*participial) suffix *-rV and underwent metatheses, usual in cases like this: *Hăʒ_ĔmV-rV- > *Hăʒ_mĔ-rV- > *Hămʒ_ĔrV- // *mHărʒ_Ĕ-. Cf. also Hurr. šēɣ-(a)lǝ 'ritually pure', šēɣǝ, šīɣǝ 'pure (water)', Urart. šēχa id. (?) - reflecting probably a variant *ʒ_ămHĔ- (see Diakonoff-Starostin 1986, 35).

    See Абдоков 1983, 149, 168-169.

    The Botl. and God. forms go back to a suffixed *c̣:in-ḳV; cf. also Kar. Anch., Tok. c̣:inḳu= 'clean, pure'.

    See Бокарев 1961, 71; Лексика 1971, 227; Гигинейшвили 1977, 107.

    The Ub. form with glottalisation suggests PWC *ṗ-; the PAT reflex *-z- confirms it (if in PWC we had *bǝć̣A, then even after assimilation PAT *-ź- would be expected; on the other hand, *-ź- normally yields PAT *-z-). However, earlier glottalisation must have resided on the affricate, since the initial labial certainly is a historical prefix and there were no glottalised labial prefixes in NC or WC.