frequently combined into a triad along with Maitreya, I am inclined to recognize the latter Bodhisattva in the middle figure, especially as two of Maitreya's common attributes, the Cakra and the holy water flask, seem to be represented in his hands. The scarf-like garment wrapped around the arms and the deep red colour of the undercoat also point to Maitreya 8. The thin Chinese-looking moustaches of the two side figures are a curious un-Indian feature. Of the four seated figures shown originally on the reverse only the two middle ones are still fairly distinguishable. The third from the right proper is shown in the ` Dhyânamudra' attitude. It deserves to be noticed that both in this figure and in the one adjoining it the outlines marking the eyes are continued by a straight black line extending from the outer angle of the eye to the top of the ear, and giving precisely the effect of spectacles. Is it possible that an indication of the latter was intended ?
It is interesting to find in three of the panels subjects treated which are known to us likewise from paintings discovered in another shrine of Dandân-Uiliq. The comparison enables us to realize to what extent details of pictorial representation had become fixed even where of no apparent mythological importance. On the reverse of D. x. 5 (see Plate LXII) we meet with the scene of the horseman and bird to which reference has already been made in connexion with a fresco of D. n. A comparison of this panel with the far better preserved one D. vii. 5 (Plate LI X) shows that, notwithstanding the considerable difference in artistic execution and care, both must be directly or indirectly derived from the same prototype. The pose of rider and horse is identical in both ; the uniformity of treatment extends to the dress and accoutrements and even the saddlery. The high sugar-loaf shaped cap with its ` vandyked ' corners is borrowed from the camel-riding figure of D. vII. 5. Leaving other details for mention in connexion with the latter panel, attention may be called to the distinctly Persian look of both rider and horse. The elaborately painted figure of the three-headed divinity, perhaps a Tantric form of Avalokitegvara, which we see in D. vii. 6 (Plate LX), and which we shall have occasion to discuss below, is represented by two replicas on the obverse of D. x. 5 and on D. x. 8. The latter painting (Plate LXII) is of interest notwithstanding its very poor preservation, because it shows the god combined with his Sakti, an arrangement common in Tibetan representations of the special tutelary divinities or Vi-dam 9. The Persian treatment of the female head shows that *Iranian influence had affected even those sacred figures and groups which the Buddhism of Khotan, like that of Tibet, must be supposed to have received ready-made as it were and fixed in all essential details from the Mahayana imagery of India.
The manuscript finds in this shrine were slight, but the place and distribution in which they turned up make them instructive. Numerous fragments of leaves written in Brahmi characters were found in front and east of the image base lying on the previously mentioned rim in little packets distinguished in the list as D. x. 9, D. x. 1o. a, io. b, io. c. a, io. c. ß, io. d. From the way in which they were placed between various painted panels it was clear that they had been deposited as votive offerings by some of the last worshippers at the shrine. The marks of red paint which I found sprinkled over the leaves fully agreed with this conclusion ; for they also appeared, as we have seen, on several of the panels. Dr. Hoernle's analysis 10 shows that the fragments distributed in packets D. x. 9, D. x. 1o. c. ß, D. x. io. d, all belonged to one leaf, and that similarly D. x. to. a, D. x. io. b were made up of portions of a large leaf, but from some other manuscript. The fragments D. x. io. c. a, which were lying