Comments:EAS 146-147, Poppe 60, Rozycki 222 (Mong.-Tung.), Martin 247, Menges 1984, 291, АПиПЯЯ 15, 81, SKE 284. This is a well known Altaic root, and the Mong.-Tung.-Kor. match seems undeniable. Doerfer first tried to abolish the obvious Mong.-Tung. parallel by attempting to postulate the original meaning in Mong. as "opposite" (see TMN 1, 168-169), then included it into his list of Mong.>Tung. borrowings (Doerfer MT 25): both positions are certainly untenable. However, the Turkic and Japanese reflexes here are not devoid of problems. The Turkic reflex reveals an exceptional preservation of *i̯u- as jü- (regularly *ɨg- would be expected) - which may be explained by the inner Turkic confusion of the synonymous roots *jüg- and *jok- ( < *ŋi̯ṑk`è q. v.). The Japanese match can be (and has been traditionally) explained as a compound of *u- 'top' with *pa (*ba) 'place' (v. sub *bi̯ŏ̀ga). However, in case of a compound we would rather expect *u-m-pa. It should be also noticed that the form *úpa- itself is very frequent as a first component of compounds, while the root *ú- alone is never attested. This all may mean that the Japanese form actually reflects a different root (with a medial labial consonant), or a merger of PA *i̯ugu with some different root. Such a root may be perhaps discovered in TM (*ebu-ri- 'to lift, raise', see ТМС 2, 471) and Mong. (Kh. övx- 'to rise'), with a provisional reconstruction of PA *ébu {(*épu)}.