Comments: PAT *rǝ; PAK *ʒa; Ub. def. á-La. The correspondence PAT *r : PAK *ʒ : Ub. *L points to palatalized *ŕ in PWC. Because of this reconstruction, Charaya's (1912) comparison of the WC forms with Georg. eri 'people, nation' seems possible (despite Shagirov 1).
Comments: The word is isolated in Ub. (for phonetic reasons a comparison with PAK *c:ǝʁʷá 'mouse' and PAT *cǝɣǝ 'marten' is not possible), but has very probable parallels in EC languages. The Ub. form may reflect PWC *d-,*t:- or *r-; *-ʁʷ-,*-Gʷ- or *-ɣʷ-; the form *rǝɣʷǝ is thus only one of the possible variants of the PWC reconstruction, fitting best for the EC data.
Comments: PAT *rǝqa (cf. also Bzyb. á-rχa). Ub. def. á-tχʷa. Not quite clear is the loss of labialisation in PAT; otherwise correspondences are regular (in Ub. *rǝq:ʷa > *dǝqʷa > *tqʷa, regularly fricativized to tχʷa).
Comments: PAK *dǝ́śa. PAK *-ś- can go back to PWC *-śʷ- or -š́ʷ-. Despite Shagirov (1, 158) it is impossible - for phonetic reasons - to compare PAK *dǝ́śa with PAT *x́ǝ 'gold'.
Comments: PAK *tχʷǝ. Ub. tχʷǝ is hardly a loan from AK because of semantic difference. Despite Abdokov 1973, 68 and Shagirov 1,78 PAT *xʷǝ-ša 'melted butter' can not be related (the correspondence PAT *xʷ : PAK *χʷ does not exist).
Judging from PAK t- and Ub. t- in Anlaut *r-, *t- or *d- can be reconstructed; *r- is most probable because of external correspondences.
Comments: PAT *raʒ́ǝnǝ (cf. also Bzyb. a-rǝʒ́nǝ́); PAK *t:ǝž́ǝnǝ́. The correspondence PAT *ʒ́ : PAK *ž́ : Ub. šʷ is quite irregular. It is possible that the original form *rVš́ʷVnǝ (regularly preserved in Ub. as dašʷanǝ́) was modified in PAK and PAT, perhaps, under the influence of Iranian forms ( *arzata-/*razata- < PIE *rǵnto-): the WC forms can reflect an original variation *rVʒ́Vnǝ (>PAT) / *rVǯ́Vnǝ (>PAK).
Comments: PAT *sa(-ra) (cf. also Bz. sa-rá), PAK *sa. The root (in the Ablaut grade *s(ǝ)) is used in all languages as a pronoun prefix of the 1st person sing.
Comments: PAT *śa (cf. Bzyb. á-śa); PAK *sa. Dirr has also recorded Ub. sa 'sabre' which is most probably an Abkh. or Ad. word. The correspondence PAT *ś : PAK *s points to PWC *ś or *š. Shagirov's linking this root with PWC *śV 'to beat, hit' is most probably faulty.
Comments: PAK *sǝ- (the Bzhed. and Shaps. variant śǝ-d is still unexplained). The second part of the stem (Ad. -d, Kab. -t) raises some problems. In Kab. -t could have been obtained under influence of xa-t 'who' (q.v.); thus we should probably reconstruct *-dA for PAK. The precise nature of this *-dA is not clear (perhaps it reflects the PWC demonstrative *da).
Comments: Ub. def. á-s. Further connections of this word within WC are not quite clear. Shagirov (1972, 137; 2, 60) and Abdokov (1983, 132) find a cognate in Abkh., Abaz. -s(ǝ) 'suffix denoting material' (Abaz. qǝlpa 'hat', qǝlpa-s 'stuff for making hats' etc.) - on the analogy with a synonymous Adygh morpheme -pχa going back to a root meaning 'wood'.
A probable derivate in Ub. is sǝ-χǝ́ 'horizontal bar, bolt, pole' (from sǝ- 'wood' + -χǝ 'a possessive particle'). If this derivation is true, we must admit an interesting fact of the further expansion of this term throughout the Caucasus: to Adygh languages (Ad. sax 'bolt', Kab. sax 'post,pole'), whence Abkh. a-sáḱ 'bolt', Abaz. (Ashkh.) saḱ 'a pole for locking the gate', and further - Osset. säux 'post, pole' and Avar s:ak 'bridge support, abutment'. See Shagirov 2, 60.