It had not needed the discovery of the pictorial representation of Pôthis to make me look out eagerly for finds of ancient manuscripts. None turned up in the shrines excavated during the first three days, but the fragments of leaves which I had found pasted across the two painted panels previously described helped to revive my hopes. The few characters still legible on them showed a bold literary hand, and this, together with the arrangement of the lines, clearly suggested that the leaves had once belonged to some manuscript written in the form of a Pôthi.
The little temples so far excavated had revealed something of the cult and art which this sand-buried settlement had possessed before its abandonment. But it was manifest that indications of the conditions of every-day life and other documentary evidence would have to be looked for elsewhere. A ruined structure close by seemed by its ground-plan, as deducible from the arrangement of the wooden posts that were seen sticking out from the sand, to suggest an ancient dwelling-place, and to the excavation of this I accordingly proceeded on December 22. The structure, D. III, lay about 20 yards to the north-west of the temple-cellas last described, and owing to the height of the dune rising immediately to its south (subsequently ascertained to be fully i6 feet above the original ground-level) there was hope of finding its interior undisturbed by recent diggings.
The excavations, started from the west side, soon brought to light the top part of walls built with timber and plaster after the method previously described, but of greater thickness, viz. io in. Notwithstanding the increased strength of construction the south wall was found to have completely decayed, evidently before the ruin had been protected by its present cover of sand, and the drift-sand ever slipping in from the slope of the dune which rose on that side greatly added to the difficulty of clearing. The walls proved to have belonged to an oblong apartment, measuring 23 feet from east to west and close on 20 feet wide, which had evidently formed the lowest story of a dwelling-house. Judging from some of the posts still intact in the framework the original height of the apartment must have been about Io feet. Fig. 32 shows the north-west corner in the course of clearing, together with fragments of the heavy plaster mouldings which seem to have decorated the upper portion of the walls but were now found almost wholly detached.
At a depth of only 2 feet from the surface a small scrap of paper showing a few Brahmi characters turned up in the loose sand filling the north-west corner. I greeted this small fragment (placed now with D. III. 3 in the inventory list) with no small satisfaction as a promise of richer finds ; and in order to stimulate the efforts of my labourers, who with the sand constantly falling in from the south had no easy task in effecting a clearance, offered a small reward for the first discovery of a real manuscript. Barely an hour later a cheerful shout from a young labourer working at the bottom of the small area so far excavated announced the finding of a ` Khat'.
Carefully extracted with my own hand and cleared from the adhering sand, it proved a perfectly preserved oblong leaf of yellowish paper, measuring about 14 by 2â in., which had clearly formed part of a large manuscript arranged in the form of a Pôthi. The reproduction in Plate CX shows the six lines of remarkably clear writing in Brâhmi characters of the upright Gupta type covering each side. The circular hole intended for the string, which was passed through the separate leaves of the Pôthi and served to keep them in order, is seen on the