490 THE ANCIENT BUDDHIST SHRINES OF MÏRAN [Chap. XIII
that the manuscript was of great age at the time of deposition, and still less is it likely that a leaf of such brittle substance could have lain thus exposed for long before the débris came down to cover and protect it.
The chronological evidence furnished by this find has its bearing on the question of the date when the shrine was abandoned. Before, however, considering this, I may complete the description of the surviving structural features. To the south-west and north-west the solid brick mound of the base, as already stated, had lost practically all traces of its original decoration in stucco, arid close to it the ground had been lowered by erosion more than four feet beneath the original level. Nothing could be ascertained here as to the enclosing passage of the Vihara. On the south-east, scanty remains of an outer wall, standing only to the height of a few inches, showed that the width of the passage on this side was merely 3 feet 8 inches. Considering that the north-east side of the structure must have been most exposed to the destructive power of the winds, it was a puzzle at first why the shrine should have retained the sculptured decoration of the enclosing passage just there. But this was soon solved when continued excavation showed that adjoining the outer wall of the passage, here 3 feet 6 inches thick, was another massive brick wall running parallel to it at a distance of less than five feet.
This wall, together with the débris accumulated in the intervening space, had helped to ward off wind-erosion. But, together with the structure which it appears to have enclosed, it had suffered so badly that the character of the latter could no longer be ascertained. It might have been a monastic building or chapel court. Curiously enough, I found the space between the two walls just referred to, and behind the colossal torsos iii and iv, filled with solid masonry to a length of close on seven feet. I cannot explain its purpose unless it possibly served to carry stairs giving access to the upper story of the main shrine. Nowhere did I come upon any indication how this would be otherwise approached. But obviously all points relating to the superstructure can be subjects for nothing more than conjecture.
As seen in the plan (Plate 31), this outer wall turned at right angles to the north-east, and there adjoining it was a small massive structure of which the walls on one side still rose to a height of about four or five feet. Measuring approximately 19 feet square outside, it enclosed a circular chamber, about i 3 feet in diameter and once probably covered with a dome. To my eye the remains, badly decayed as they were, clearly suggested one of those small domed temples, or Viharas, square outside and circular within, with which I had become familiar from the Swat Valley to Kashmir.16 Unfortunately no archaeological relics of any kind had survived within the broken walls, nor could I trace any within a small and still more decayed cella, of which the foundations had escaped wind-erosion, some six yards to the north-east.
When I subsequently cleared the narrow space left between the southern ends of the two parallel walls behind the north-east passage of the main shrine, I found it filled to a height of seven feet with the dung of sheep and horses. But the only discovery made was a large roll of a strong, if coarse, material apparently woven of goat's hair. A similar crust of hard dung was found to have also formed over portions of the broken walls close by, thus furnishing evidence that long after the decay of the shrine its ruin must have been used as a shelter by shepherds and others, while the ground in its vicinity retained vegetation fit for some sort of grazing. In the case of the Niya and Lou-lan sites I have already had occasion to note exactly corresponding proofs of a transitional