Sec. ii] THE REMAINS OF KÖH-I-KHWAJA 913
northern corner of the circumvallation. Here the plastered wall surface in places showed traces of having once borne decorative painting in tempera ; but designs were no longer recognizable. The area along the enclosing wall, whère it turns from the gate south-eastwards, was found covered with debris apparently of small quarters, and the walls seen in Fig. 462 farther towards the NE. side of the quadrangle are partly those of upper story apartments built above others now completely buried.
There remains a small but very interesting architectural feature to be mentioned. I have already referred to obvious indications that the buttresses built against the terrace wall were later additions. I had observed traces of painted plaster on what obviously was part of the original wall under the arch behind the second buttress on the right of the stairway. On removing the rough masonry at the top of this buttress and arch there came to light the Doric capital on the top of a semi-engaged column, both cased in plaster (Fig. 472 ; Pl. S4). For the upper and lower mouldings of the capital burnt bricks had been used. The buttress next to the right was also found to hide a column and capital of the same order, and on the wall between them there appeared remains of an architrave decorated with volutes in white stucco. Much of this relievo decoration had been destroyed by wasps' nests, the careless masonry of the later buttress not having offered adequate protection. But even thus these scanty remains of unmistakably Hellenistic style offer definite support for the chronological conclusion drawn from the relievo figures above described.
SECTION III.—REMAINS OF MURAL PAINTINGS
If I have left the account of two striking discoveries of pictorial remains to the last it is because the preceding survey of the ruins, summary as it must be, will make it easier to judge of their import. They were made on the very day following my first visit to the site and were largely the cause of the prolongation of my stay there till December 17th. I have referred before to a narrow vaulted passage, i, which, as the sketch-plan, Pl. 53, shows, was built against the wall close to the left of the main gate of the inner enclosure. Fig. 465 shows its position after removal. In the western corner of this passage, only 5 feet wide, one of the men who had come with me from Daudi village had noticed some coloured plaster showing through a crack in the wall. On removing here a little of the rough masonry, a piece of older wall surface behind was disclosed. It showed what I recognized, rightly as it proved, to be the representation of a figured textile. Next morning I had the vaulting of the passage carefully removed in order to be able in safety to examine the surface hidden behind the side wall nearest to the main gate. As soon as a top portion of this side wall nearest to the corner had also been taken down, the legs of a richly dressed figure were revealed. When we proceeded to loosen further the brickwork, only 14" in thickness, most of it became detached and collapsed. The effect was striking. Higher up there was seen a procession of richly dressed figures painted in tempera and preserved up to the waist ; exposure above the later passage wall had effaced the rest. Below, in a separate frieze, appeared some haloed heads. The whole wall decoration, incomplete as it was and poorly preserved, yet at once recalled the arrangement of the frescoed friezes on the passage wall of the shrine M. v at Mirân.1 When subsequently the remaining portion of the later brickwork was carefully lifted off, a group of four haloed figures with remains of a fifth (Fig. 468), extending over a length of about II feet, came to light on the lower frieze.
From the first it was evident that the painted wall surface, owing to the effect of atmospheric moisture and still more to the destructive action of insects, probably white ants, on the mud plaster, was in a far worse state of preservation than the mural paintings of Buddhist shrines in Chinese 1 See Serindia, i. pp. 517 sqq.