Abstract: The paper deals with the description of the Ursidae specimens from the Ramapithecus fossil site of Lufeng. Three kinds of the Ursidae fossils were collected by the Lufeng Team in 1981: Ursavus depereti, Indarctos sinensis and Ursinae.
The fossil materials of Ursavus depereti include a broken right DP4, a broken left M1 and right M1, a broken left M2 and right M2, a complete right M2; a piece of left tooth row with P3—P4, a broken right P4, a broken right Mi, a piece o￡ right tooth row with M2—M, a complete left M3.
DP4 The paracone relatively bigger than the metacone. The deutercone ridge-like. There is a small basin between the posterior part of the paracone and deutercone. Many wrinkles on both internal and external sides of the cusps.
Upper molars moderately elongate, low-crowned, with reduced main cusps and prominent internal cingulum, the occlusal surface much wrinkled.
M1 The External cusps higher than that of the internal side. The paracone higher than the metacone. The paracone protrudes antero-externally. The metacone relatively small and connecetd to the hypocone by a cross-crest. The internal cusps fused into a wedge-like ridge.
M2 The external border of the metacone considerably reduced relative to paracone border. The talon somewhat developed.
P3 and P4 with double roots. The main cusp developed, with a distinct anterior accessory cusp and a faint posterior accessory cusp. The internal cingulum developed.
Mi The paraconid relatively large. The paraconid-protoconid forming shear. The metaconid relatively small. The talonid broken.
M2 Short antero-posteriorly relative to Mi. The trigonid broad relative to the talonid. The metaconid is more prominent than the protoconid. The external and internal borders slightly constricted postero-medially.
M3 Small, nearly circular. Three cross-ridges run from the protoconid to the centre ot the tooth.
Only a dight M3 of Indarctos sinensis was found from Lufeng. It is nearly circular. There is a system of enamel wrinkles on the occlusal surface. The middle field is surrounded by a slightly raised wall without any distinct cusps. Only the protoconid forms a ridge swelling bounded posteriorly by a notch. Two short ridges run from the protoconid to the centre of the field. There are six small shallow vertical grooves which are equal distance each other on the internal wall of the tooth. The length and breadth of the tooth crown are 18.5 mm and 17.0 mm respectively.
The specimens of the Ursinae include a piece of fragment of upper jaw with a deciduous premolar, two left M1 and a right M1. Three broken lower canines, a left Pi and P2, a piece of left tooth row with P3-M1, a right P4, a broken right Mi and a right M2. The chada~ cters of Mi and M2 were discussed in the paper.
Mi The trigonid longer than the talonid (ratio of the length of trigonid/the length of the talonid 1.7/1). The latter is broader than the former. Without cingulum and any accessory cusplets.
M2 The length smaller than that of Mi. Its maximum width is at the middle of the trigonid and it tapes slightly posteriorly. The long axis of the tooth is somewhat curved. The external wall is concave and the internal, convex. The protoconid and metaconid are the largest cusps. A transverse ridge from the protoconid to the metaconid is interrupted by the deep valley between the cusps. The trigonid consists of four cusps: the protoconid, the metaconid, the small paraconid and another small cusp at the antero-extemal comer of the tooth. The talonid with five cusps: two lie on the external border, two on the internal border, the fifth on hind border. In comparison with M2 of Protursus, they seem to have some resemblances, but the size of the former much bigger than that of the latter. In comparison with U. boeckhi and U. edensis, their size of Mi and M2 is about same and Mi longer than M2, but the trigonid broaded than the talonid in the specimen of Lufeng, the trigonid narrower than the talonid (in U. boeckhi )or both subequal (in U. edensis).
The specimens of Lufeng were not referred to any genus and species in this paper. If they are referred to Ursus, it is difficult to explain in age and horizon because they were excavated from the same layer with Ursavus and Indarctos mentioned above. If they are referred to Protursus, its size is too big on the one hand, and the establishment of Protur- sus itself is still a problem on the other hand. Thenius (1977) suggested that it was neither the earlist and most primitive representative of the Ursinae nor a member of the Ursidae. Protursus is a junior synonym of the genus Simocyon. In my opinion, it is undoubted that the specimens of Lufeng is a member of Ursidae even if Protursus is Simocyon. There is a clear pressing trace on the hind wall of its M2. This is a mark of connection to M3. Its phy- letic position may correspond to Protursus.

Key words: Ramapithecus fossil site of Lufeng; Miocene fauna; Carnivora; Ursa- vus; Indarctos