centric squares at once recalled the temple cellas with enclosing passages I had excavated at Dandân-Uiliq. A little experimental digging at the south-eastern corner of the inner square soon brought to light small fragments of soft and much-decayed stucco, which had belonged to a large-sized image. So the whole of my little force, counting over twenty ` Madigârs' from Niya, soon supplemented by shepherd-guides and every able-bodied man whom the latter could impress at the shepherd stations higher up the river, was at once set to work here.
Within an hour I had conclusive proof that my surmise was right. From the sand which covered the interior to a height of about 5 feet on the east, but increased to fully 7 ft. towards the west, there emerged the timber and plaster walls of a square cella enclosed within a passage, no doubt intended for circumambulation. The outer walls of the passage had decayed almost to the ground, and a good deal of excavation was needed there before the clearing of the interior of the cella could be safely _proceeded with. Nevertheless, before work ceased with
nightfall more stucco fragments had furnished some indication of the general decoration of the shrine, while three finds of fragmentary paper leaves with Brâhmi writing, among them three halves of folia clearly recognizable as belonging to a Buddhist canonical text in Sanskrit (E. i. 2), had helped me to form an approximately correct idea as to the age of the ruined shrine. All these turned up on the east side of the cella, lying in loose sand from i to 2 feet above the original floor. On the following day the excavation proceeded sufficiently to show me all the main structural features of the small temple. But it was not until the 23rd of February that the interior was completely cleared and the rich haul of interesting MSS. and other varied finds carefully gathered.
The cella formed a square of 18 ft. 4 in. inside, enclosed by walls of timber and plaster having a thickness of io in. As seen in the detailed plan (Plate XXXVI), the walls were only roughly orientated. Their construction differed from that observed at Dandân-U iliq and the Niya Site by having no matting within the plaster, but the timber framework was massive, as seen in Fig. 49, which shows the cella after excavation, and specially strengthened by diagonally-placed rafters. The main posts of both cella and passage stood at a uniform height of 9 ft. But it may well be doubted whether they indicate the original height of the roof, seeing that the central group of relief statues must have reached close to that height. The inner faces of the walls seemed to have received a rough coating of stucco, but showed no trace of fresco decoration. The entrance lay to the east. The enclosing passage, the walls of which were of similar construction but far more decayed, was 5 ft. wide ; its plastered floor lay 3 ft. higher than that of the cella, a circumstance which had helped, no doubt, to protect the lower portions of the cella walls. On the east side the outer wall of the passage had for the greater part disappeared entirely, together with the portion of the floor facing the entrance, possibly as a result of the early burrowing effected here by treasure-seekers, to which I shall have occasion to refer below.
The four corners of the cella were occupied by plaster images, almost wholly detached and standing each on a base meant to represent an open lotus with the petals pointing downwards. Those in the north-west, north-east, and south-west corners alone survived, and they, too, only in their lower portions, as seen in the photographs reproduced in Plates XI, X II. Of the south-east corner statue nothing remained but the small undefinable fragments which had
turned up at the first trial excavation. By the side of the statues in the north-east and
south-east corners there must have stood in each case a small figure (b, e, in plan), but of these only the lotus-shaped pedestals remained, the sculptures having crumbled away owing to the decay of the east wall behind them. It is impossible to say what divinities all these figures