base and arranged in three slightly receding stories. Its ground-plan corresponds exactly to that of the Rawak Stûpa base.' It consists of a cruciform story superimposed on a square, the whole showing twenty projecting angles. A circular drum, ornamented with lotus leaves in low relief and a foot high, carries the cylindrical Stûpa dome now broken at its top. The total extant height of the Stûpa is about 24 feet.
A cutting made from the east into the third base, evidently by treasure-seekers and a long time ago, has laid bare an interesting feature of the Stûpa in the shape of a small inner chamber or well. This is 32 feet square and appears to have had its floor resting on the top of the circular base. This chamber or well, curiously enough, was originally made accessible by a small opening 2 feet wide, of which the plastered vaulting has survived at the top of the cutting. The opening or window appears to have been about 3 feet high, but the treasure-seekers' operation has deepened it to about 6 feet. Owing to the cutting it was impossible for me to determine whether the ancient opening, no doubt intended to facilitate a deposit of sacred objects within the chamber, was subsequently walled up or kept closed in some manner which would have rendered occasional inspection of the contents possible. I may add here that the Stûpa of group III (see Plate 5o), which in all structural features was an exact replica of the one just described, except for slightly smaller dimensions, also had an interior chamber 31i feet square. As its east side was much injured, the opening which it is likely to have had there could not be traced. In the case of the Stûpa in group I treasure-seekers had not been content with effecting an entrance to the chamber from the east, but had burrowed also deep into the circular base from the north side.
The small cella v of I, immediately adjoining the Stûpa from the south, measures 8 feet 4 inches square inside. Its floor was found covered to a height of about 2 feet with sand and plaster débris, which had evidently not been disturbed. But the smoke-begrimed or burned walls and ceiling, and the scratchings visible everywhere on the surviving plaster surface, showed only too clearly the treatment to which the decoration of the shrine had been subjected. On the spherical ceiling of the dome only traces of painting remained, suggesting rows of small Buddha figures. But along the foot of the east and west walls some of the lowest portions of a frescoed frieze came to light from under the protecting layer, retaining strikingly fresh colours. Near the south-east corner it was possible to make out a procession including two richly caparisoned horses—the figures of the riders were lost—and attendants walking behind. The horse of the central mounted figure had its feet raised above the ground by demons ; behind it a reddish-brown animal, perhaps meant for a panther or leopard, appeared led by an attendant.10 What survived of a corresponding frieze on the west wall was even more injured ; but the fragment H. B. oo6, removed as a specimen, shows the fine red background and the brilliancy of the colouring in general. On the north wall of the cella were remains of a large painted vesica, once evidently enclosing a stucco image, and on its right the lower part of a richly-draped figure standing on a lotus." Besides fragments of stucco relievo figures which were found both in the cella and in its antechapel,12 there were also recovered from the former a small Stûpa model and a relief Buddha figure, both of clay and obviously votive offerings (H. B. v. 004-5). Several pieces of paper, bearing in Chinese print the same Buddhist magic formula, were also found on clearing the cella.12
9 See Ancient Kholan, ii. Pl. XL.
10 Removal of this portion of the frieze was difficult; for fragments of it, see H. B. v. oo6 in List below, also Pl. CXXV.
" For a description of this fresco piece, wrongly marked H. B. i. 002, see List below ; Pl. XII.
For other smaller fresco fragments found detached among the débris of this cella and its antechapel, see H.B. 007-9.
12 For stucco relief fragments from the antechapel, see H. B. oor-5, 0010.
" See Chavannes, Documents, Nos. 988-9, Pl. XXXVI.