Comments: PAT *mǝza (cf. also Bzyb. á-mza 'torch'; Abzh. has a compound with č̌ǝ- 'wax'); PAK *bzǝ́-jǝ (with a pronominal second part, see Shagirov 1,95); Ub. def. á-mʒa.
Shagirov (1,95) explains the root morpheme in PAK *bzǝ́-jǝ as 'tongue'; this is, indeed, probable because of the compound identity (Ad. maṣ́ʷa-bzǝj = Abkh. á-mca-bz = Ub. mǝǯ́a-bźa 'fire tongue, flame'). Since, however, PAT *mǝza and Ub. mǝʒá 'candle, light' certainly constitute a different root (obviously having nothing to do with *bǝźa 'tongue'), it seems probable that in PAK the two different PWC roots had merged (phonetically this is quite regular).
Comments: PAK *bźa (nowadays used mainly in the meaning 'horn for drinking'; 'horn of animals' is rendered by a compound *bźa-q:ʷa q.v.). It is also used in compounds *ʎá-bźa 'hoof' > Ad. ʎābź, Kab. ʎābźa (*ʎa 'foot') = Abaz. š́a-mza (but despite Шагиров 1, 89 this root has nothing to do with Abaz. msa, Abkh. a-msa 'horn matter' q.v.) and *ʡa-bźá-na 'fingernail', *ʎa-bźá-na 'toenail' > Ad. ʡabźān,ʎabźān, Kab. ʡabźāna,ʎabźāna (*ʡa 'hand, *ʎa 'foot', *na 'eye' - see Шагиров 1, 249; 2, 156).
Despite Abdokov (1983, 106-107) PAK *ʎá-bźa with the meaning 'root' is hard to separate from the homonymous *ʎá-bźa 'hoof'. That is why Abaz. š́ajʒa 'root' (having no Abkhaz counterpart) should rather be regarded as an old Adygh loanword (although -j- instead of -b- is rather hard to explain).
Ub. -žʷa (probably with a late dissimilative loss of initial labial) is present in compounds q̇a(n)cʷa-žʷá 'fingernail' (q̇ancʷa 'finger') and š́a(n)cʷa-žʷá 'toenail' (š́a(n)cʷa 'toe'). Abdokov (1983, 82) cites Ub. bž́a 'horn' which is probably non-existant (while Abkh. a-msa 'horn', also cited ibid., goes back to PWC *pǝ̃ša q.v.).
Comments: PAK *bǝʒǝ́. There are no special reasons to treat Ub. bǝʒ as an Adygh loanword (the correspondences are all right). An expressive stem (with parallels only in Agul).
Comments: In Abaz. there exists also a reduplicated form bǝžbǝž. PAT *bǝžǝ, PAK *bžǝ-ʒá (-ʒá < PWC *ʒ́ǝ 'flea, fly' q.v.). The correspondence PAK *ž : PAT *ž is only pointing to PWC *ž́.
This root (for phonetic reasons) should be kept apart both from Ub. žʷa(n)ḳ́ǝ́ 'flea' and from PAK *bž́a 'bee'.
Comments: The root is preserved independently in Ad. (PAK *ba 'many'; Kab. preserves it only within compounds like ba-ʁʷa-n 'to increase, prosper' - see Шагиров 1,70-71) and Ubykh. AT languages have kept the root only in compounds *macʷǝqA-bA 'thumb' (Abkh. a-načχǝ́-b, Abaz. mačʷqa-ba) and *š́acʷǝqA-bA 'big toe' (Abkh. a-š́ačχǝ́-b, Abaz. š́ačʷqa-ba) = Ub. q̇a(n)cʷa-bIǝ, ʎa(n)cʷa-bIǝ.
Another common WC compound with the same component is *ʎ́a-bIA "foot's bulge" > Ad. ʎa-b 'heel', Ub. ʎa-bIǝ (in compounds) - see Dumézil 1965, 235, Шагиров 1, 248-249).
On the etymology of PAK *ba see Шагиров 1,70-71 with literature (all attempts to separate this morpheme from Ub. bIǝ and AA *-bǝ seem to us superfluous).
Comments: PAT *jǝba; PAK *jǝbá. The common PWC form *jǝ-bIa contains a pronominal prefix *jǝ-, but the same root without it is also attested: cf. PAK *sʷǝzá-ba 'widow' (a compound with *sʷǝzǝ 'woman') > Ad. ś̌ǝzāb, Kab. fǝzāba.
Comments: PAT *bVga; PAK *baǯ́á. The development *g > *ǯ́ in PAK is not clear (the antiquity of *ǯ́ in Adyg is proved by the Ubykh loanword baǯ́á 'fox'); nevertheless, the regular correspondence PAT *bVga : Ub. bIaǵa- leaves no doubt in the reconstruction of PWC *bIaga.
Comments: PAT *(bǝ)la (in Abkh. dialect variations: á-bla and á-la). Certain problems are connected with the PAK form *na : it is certainly related to PAT and Ub. forms (even several identical compounds with the component 'eye' may be observed in AK and other languages), but reasons for nasalisation are not clear (regularly *da would be expected; note that in this case the root would be homonymous with *da < *ĺA 'ear' q.v.).
Comments: PAT *sa- (cf. also Bzyb. a-sa-rá); PAK *bzǝ-. As pointed out by Shagirov (1,96) Ub. bzǝ- may be borrowed from Ad. In that case it would be possible to compare instead Ub. sʷa- (a-s-sʷá-n) 'to cut, clip', reconstructing PWC *(b)-sV (with secondary labialisation in Ubykh).
Comments: The Abkh. form is probably most archaic; not quite clear is the origin of nā- in Shaps. and -r- in Ub. (cf. also Abkh. Bzyb. a-bráč̌). It is also possible that all the WC forms reflect an original form like *bVnc:ʷV or *bVmc:ʷV with exceptional preservation of the old medial nasal (cf. external evidence); in this case the Ad. form should be treated as a metathesis ( < *banca), and the forms with -r- - as a result of denasalisation (-n- > -r-). In any case, the coincidence in Adyghe between nāpca 'medlar' and nāpca 'eye-brow' is, of course, purely fortuitous.
Comments: PAT *bVgVʒ́V (cf. also Bzyb. a-bgǝ́ʒ́). The form is an old compound (with contraction) from *bVga 'fox, jackal'(q.v.) + *bVʒ́a, with the second part corresponding to PEC *bɦĕrc̣ĭ 'wolf'.
Comments: PAK *ja-bʁa- (with a pronominal prefix; in Kab. the stem is usually in a compound with jǝ-gʷ 'heart': jǝ-gʷja-bʁa-n); Ub. sǝ-baχá-n. PAK *-ʁ- < *-χ- through assimilation.
Comments: PAT *mǝć̣ǝ (cf. Bz. á-mć̣; the Abaz. form originally meant 'gadfly' /lit. "ox"+"fly", cf. Abkh. Bz. á-ć̌-mǝć̣ 'gadfly'); PAK *baʒá. Glottalisation in PAT is secondary (probably, under the influence of *mac̣a 'locust, grasshopper' q.v.).
The dissolution of the AK root and its derivation from PWC *ʒ́ǝ 'flea, fly' q.v. (Rogava 1956, 13-14) is untenable.
WC > Osset. bǝnʒ / binʒä 'fly' (despite Trubetzkoy 1921, 249, who suggested the opposite direction of loan; Dumézil (1963,15) argued correctly with Trubetzkoy, but his reference to the root's expressive nature is not quite appropriate, since the Osset. and WC roots are specifically close to each other. Other etymologies of the Osset. word (see Abayev 1958, 280) are unsuccessful.
Comments: PAK *bza (Kab. bza-ps is a compound with -psǝ 'string, rope'). The PWC reconstruction is tentative (because PAK *z can go back to several prototypes: PWC *z, *ź, *ʒ, *ʒ́, *ž, *ǯ).