doubt that they could have got there onlÿ by accident. Their distribution in varying depths and places suggests that they had fallen in from an upper storey, while the basement was gradually filling up with drift-sand. This assumption was fully borne out by the small pieces of felt, leather, oilcakes (` kunjara '), and similar refuse which turned up in the same layers. The pagination numerals which I could make out on the margin of some leaves, and which in one instance go up to 132, plainly showed that the pieces thus rescued were mere fragments of larger texts which had probably perished with the destruction of the upper floor.
The earlier these fragments had reached the safe resting-place offered by the sand-covered basement, the more extensive they might reasonably be expected to be. So I watched with growing eagerness the progress my men made on the 23rd of December in clearing the sand nearer down to the original floor. It was no easy task, for the drift-sand from the slope of the dune to the south was ever slipping to fill the space laboriously cleared, and as the wall on that side had apparently decayed long ago, additional exertions were needed. As the work proceeded towards the centre of the room a massive beam of poplar wood, nearly a foot in thickness, was laid bare. Its length, close on 19 feet, and its position showed that it had once stretched right across the room, undoubtedly supporting its roof. Two well-carved octagonal posts with bell-shaped capitals surmounted by . a circular band, in which I easily recognised the Amalaka ornament of Indian architecture, had turned up before ; they had undoubtedly served to support this central beam.
A little beyond the latter, towards the east, the men clearing the sand just above the floor came upon a closely-packed bundle of manuscript leaves, evidently still retaining the order they had occupied in the original ` Pothi.' A little later two more packets of leaves belonging to the identical manuscript were brought to light, practically intact, though the action of moisture to which these leaves must have once been subjected, owing to their position not far above the ground, had stuck them closely together and made
IIIF Color HighRes
