tapering shape and the rich globular mouldings of its base are derived from the Indo-Persian model in the Gandhara relievos is quite clear. But it is equally obvious that it represents a distinctly later development of this model than our Mirân stucco pilasters. The same observation applies with equal strength to the two fine wooden columns of Indo-Persian style which were brought to light by my renewed excavations in one of the halls belonging to the main quarters of the Endere fort (E. III. iv), and which are shown in Fig. 70.20 In these we find the addition of fillets up the shaft, marking later elaboration, but we can recognize even more clearly the essential continuity of the Indo-Persian type both in the base and in the bell-shaped capital.
Evidence of The archaeological evidence furnished by these Endere columns carved in wood has its special
Endere value for the dating of the Mirân shrine, because we know for certain that the structure containing
them must have been built between A. D. 645 and 719, and probably nearer to the former date.21 If we make due allowance for the time which is likely to have passed before the Indo-Persian column at Mirân, still seen with all the essential features of its Gandhâra prototype, could develop into this later form of Endere and Khadalik, the conclusion seems justified that the construction of the Mirân Vihara M. II cannot safely be put later than the fifth century A. D.
Evidence of With the downward chronological limit thus inferred the evidence of the sculptured remains
sculptured may well be reconciled ; for their type shows no essential difference in style or technique from that so
remains. abundantly represented in the relievos of the Rawak Stûpa, which may be dated with considerable
probability from a period between the fourth and seventh century A. D.22 But it must be borne in mind that, on the one hand, the sculptured relics of M. II are unfortunately very scanty, and that, on the other, conservative adherence to the models derived from Graeco-Buddhist art appears to have remained a strongly marked characteristic of Buddhist sculpture in Eastern Turkestan throughout its existence. Hence, no argument based upon style of sculpture could claim much independent weight in determining the date of the construction of the shrine.
It is obvious that this must be kept quite distinct from the question of the time when the shrine was deserted and allowed to fall into ruin. The only piece of positive evidence available is the fragment of a palm-leaf Pbthi in Sanskrit, already mentioned. As It must have been written within the fourth or fifth century A.D., it supplies a safe upper chronological limit. As regards the lower, I feel inclined to attach some importance to the total absence of any relics in Tibetan writing, and to draw from it the inference that the abandonment took place at some period before the Tibetan occupation about the middle of the eighth century. It is true the evidence in this case is of a purely negative kind. But it seems to me to gather some additional weight from the fact that at the ruined temples of Khadalik and Endere, where the Brâhmi manuscript remains were of distinctly later type than the M. it Pôthi leaf, there were found with them plentiful Tibetan leaves and fragments which proved that Buddhist worship had continued in these shrines under Tibetan domination. Thus the clearing of this Mirân ruin fully confirmed me in the belief that the site had a far older history, and in a way it prepared me for the much more striking proofs of this which my subsequent excavations revealed.