Comments:Владимирцов 211, Poppe 18, 48, 82. The root cannot be united with *kúma for phonetical reasons, although in modern languages their reflexes tend to merge. Despite Doerfer's doubts (TMN 1, 445: "Urverwandtschaft ist sehr unwahrscheinlich"), it is probably archaic; loan from Turk. into Mong. (TMN ibid., Щербак 1997, 139) is quite improbable. Jpn. has -t- (instead of the expected -r-) probably because of a special cluster development.