Comments: PAT *śǝ (cf. Bzyb. a-śǝ́); PAK *wa-sǝ́ (a compound with wa- 'sky'). Cf. also Ad., Kab. q:e-sǝ-n 'to snow' (PAK *q:a(jǝ)-sǝ-) = Abaz. ʕa-s-ra id. See Shagirov 1,227; 2,93. The Adygh source is possible for Svan. mus 'snow' (see Dzhanashia 1959, 106-107, 116).
Comments: PAK *qA (cf. also Bz. a-χǝ́), PAK *śħa; Ub. def. a-šá. A cluster with some front fricative must be reconstructed for PWC on basis of the Ub. and AK forms; its precise nature is hard to establish.
Comments: PAT *ʒ́ǝ-śa- (cf. also Bz. á-ʒ́-śa-ra). For the first component (present also in Ub. ʒ́a-śa-) see *ʒ́V-. In Abaz. (and in the Abkh. Bz. variant á-ʒ́-ća-ra) there occurred an assimilation (-ʒ́-śa- > -ʒ́-ća-), perhaps not without influence of *ća- 'to go' q.v.
In PAK (*ja-sǝ-) the root is present without the first component *ʒ́A (but with a pronominal prefix).
Comments: PAT *sǝsǝ ( ~ *śǝśǝ - the Bzyb. form is unknown). In the first part of the compound - PAT *cʷa 'dream' (q.v.); cf. also other compounds in Abkh.: a-χǝ-r-sǝs-ra 'to slumber' (perhaps the non-reduplicated form can be extracted from a-χǝ-r-q̇ʷǝ-s-ra id.). Ub. 1p. s. sǝ-sá-n.
Comments: PAT *xʷǝ- (cf. also Bzyb. xʷ-ba); PAK *txʷǝ. An initial cluster must be with a sibilant must be reconstructed for PWC (it has been regularly simplified in PAT, and changed to *txʷ and *š́x in PAK and Ub. respectively).
Comments: PAK *sʷawǝ́. The AK-Ub. comparison see in Rogava 1956, 87, Dumézil-Esenc̣ 1975, 148, Shagirov 2, 100-101. Although none of these authors could offer a satisfactory explanation of the strange correlation AK *-w- : Ub. *-n-, the parallel still seems satisfactory. All other suggestions (comparison with PAT *cǝxa 'honey' in Abdokov 1973, 71; comparison with PAT *cʷa 'wax' and PK *c̥1̇wil- / *c̥1̇wid- id. in Rogava 1956, 87, Klimov 1969, 291) should be rejected for phonetic reasons.
Comments: PAK *sʷandǝ́-ra. Similar forms in Abkh. and Abaz. (Abkh. a-š̌ǝndǝ́q̇ʷra 'trunk', Abaz. šʷǝndǝq̇ʷara id.) may represent a merger of the genuine root and the Turkic loanword (cf. Kypch. sunduq etc.).
Comments: PAK *pśaśá. The available forms reflect various assimilations: PAK *pśaśá < *p-śasá ( < *śʷasa with a labial prefix); in Ub., besides śasá there exists also a variant sasá (cf. also Ad. sās 'girl (as address)'. The basic variant was obviously *śʷasa (PAK *ś can only go back to a labialized *śʷ).
Comments: PAT *sʷǝ (cf. also Bzyb. a-ś̌-rá). The Ub. word is a compound (for the second part see PWC *bIa), and an exact parallel of PAT *sʷǝ-bǝ "hoar-frost, caked snow" (Abkh. a-š̌ǝ́b, Abaz. šʷǝb). Therefore Dumézil's (1969) analysis of Ub. sʷǝ- as 'light, white' is certainly wrong. Despite Shagirov (1,194) PAT *sʷǝbǝ and Ub. sʷǝbIǝ́ for phonetic reasons can not be related to PAK *č:ápǝ 'hoar-frost'.
Comments: PAK *sʷǝ́zǝ. It seems possible to compare the PAK and Ub. forms as a whole (if PAK -z- < *-c̣- as a result of secondary deglottalisation; perhaps under influence of PAK *bzǝ 'female' q.v.), and not only PAK *sʷǝ- with Ub. sʷǝ-.
Comments: PAT *śǝ-śǝ (cf. also Bz. a-śǝ́ś), pl. *śa-ra (Abkh. a-sá-ra, Bz. a-śá-ra, Abaz. sa-ra); -śǝ is the common suffix for names of young animals. PAK *śǝná.
External evidence suggests that -na is part of the root; reason for its loss in PAT is not quite clear. The correspondence PAT *ś : PAK *ś points unambiguously to PWC *śʷ.
