Comments:See АПиПЯЯ 286, Цинциус 1984, 96-97, Дыбо 316, Лексика 243, 245. The comparison is quite reliable phonetically; *-jl- has to be assumed to account for the development in Kor. ( > -i-). The semantic side ('arm'/'leg'/'thigh') can be explained if we suppose that the word originally designated a part of the animal body (front or hind leg together with the thigh). Cf. also ТМ *xūl-kse 'sleeve' (probably an original derivation, although the length is not clear), *xul-ŋsi 'shank, shin', PT *koltuk 'armpit' (Лексика 243, TMN 3, 557-558, ЭСТЯ 6, 52-54). It is interesting to speculate on the subject of the identity Kor. *kūi-mǝ̀rí = Jpn. kuru-(n)pusi. In Kor. -mǝ̀rí is certainly to be analysed as "head" (*'leg-head'); the Jpn. form in this case may reflect a dissimilation < *kuru-n-musi, where *musi could be the remnant of PA *mĕ́ĺǯu 'head' (q. v.). The same element in fact may be also present in Jpn. *tu(m)pu-(n)pusi 'ankle, knee' and *kǝmpusi (if, with haplology < *kǝmpu-(n)pusi) 'fist'. One has, of course, to reckon with the possibility of having here rather PJ *pusi 'joint' ( < PA *bŭ̀ĺi q. v.), which would explain the constant emergence of a stop in Jpn.; but the Jpn.-Kor. match (*kūi-mǝ̀rí = kuru(n)pusi) seems to be not accidental.