the "elfin mills" and probably some of the undecipherable signs are purely magical. The oldest level has the most varied subject-matter, and it seems as if this level covered a wider range of conceptions than the rest. Otherwise we have no doubt to deal with imititative magic and especially hunting magic. The hunters passing here have drawn the outline of their game on the rock, by which action they got power over the animal in question and would be successful on their hunting-expedition. TALLGREN writes : "The primitive rock-pictures in Northern Central Asia express, I suppose, an idea which once prevailed among the Eurasian peoples of the Old Stone Age, and which still lives among the primitive hunters in the Siberian Taiga and on the mountain chains of Central Asia. The pictures of this really "timeless" culture have been executed by pocking or painting. The main elements in its repertory are animals and conventionalized men, worked in the same manner as in the corresponding late rock pictures in Africa, Spain, Arctic Scandinavia and among all primitive hunting nomads in general. It must be noticed, however, that the Central Asian rock pictures of the primitive group are seldom naturalistic or lifelike. They are mostly conventional, being products of a "frosen" shamanism rather than of a hunter's imagination, as the Palaeolithic pictures are. In Siberia we have to do with an inherited art, not with one which depends on the personal observations of the painter."
How far our carving can be labelled as an inherited art is hard to tell, but it obviously has connections with the Siberian and Mongolian rock pictures, above all with the group that TALLGREN calls the primitive. There is, however, one difference in the subject-matter of the Siberian-Mongolian rock carvings and the one in Quruq-tagh. The latter contains several incised hands, whereas this element seems to be lacking among the former groups. I have not been able to trace any hands in the publications of Siberian and Mongolian rock pictures available to me, and Prof. TALLGREN of Helsingfors, the famous specialist on the prehistory of these regions, has kindly confirmed this observation. Otherwise the pictures of hands have a very wide distribution. On the Panja rock pictures there are hands, both large naturalistic ones (Tallgren 1933, Fig. 19) and short and broad ones executed in the same stylized manner as are some of the Quruq-tagh hands (Tallgren 1933, Fig. 17, our Pl. XVII b). Exactly the same curious shape is found on an Indian seal from Barenrah, Hamipur (Wilke 1913, Fig. 210 b) and on one of the Lang-shan carvings (Bergman 1935 a. Fig. 6). Some of the hands in our carving have only four, or even three fingers, but it is far from certain that they are meant to represent mutilated hands. The general meaning of the hands is explained as apotropaic gestures ; in other instances they are interpreted as symbols of proprietorship and as means of keeping possession when they are over or next to animal figures. This latter explanation would fit in perfectly with the general meaning ascribed to the other elements, and on Pl. XVI a a little to the left of the centre there is a hand covering a part of an animal, though this is incomplete, only the hind part being executed.