494 THE ANCIENT BUDDHIST SHRINES OF MIRAN [Chap. XIII
the dome had suffered on the south. As the elevation reproduced in Plate 32 shows, the present height of the Stûpa was close on 13 feet above the cella floor. That originally it is likely to have been considerably higher is proved by the remnant, visible in Fig. 119, of what I take to have been a square pedestal, corresponding to that discussed in connexion with the Stûpa rock carvings of Chitral.' The peculiarity of this Stûpa lies in the circular shape of the base instead of the square one usually found in the Stûpa ruins which I examined elsewhere in the Tarim Basin. It may be assumed that the choice of the round base, amply attested among the Stûpas of India and the North-west border, was due to considerations of the space available within the small circular cella.
On the lowest division of the base, 9 feet in diameter and 9 'inches high, there is placed a receding drum, z foot 6 inches high, terminated both below and above by a uniform series of mouldings. Then follows another plain division, 9 inches high and of the same diameter as the lowest, and above it a succession of small step-like mouldings, with a total height of less than 6 inches. From this rises the cylindrical member, about 2 feet 3 inches high, which carries the dome, both being 6 feet 8 inches in diameter. The projecting frieze-like moulding, about 4 inches thick, which intervenes between drum and dome, is a feature seen with equal clearness in the Stûpa carving of Pakhtoridini, the wooden Stûpa models of Niya and Lou-lan, etc.2 It is difficult to make sure whether, and in what way, the disposition just described was intended to give expression to that arrangement of the base in three stories which we have reason to consider as prescribed by tradition for the Stûpas of the Tarim Basin.3 But it is worth pointing out that the circular base of the Stûpa in the neighbouring shrine of M. v reflects such an intention much more clearly. The plaster surface was fairly hard, and strengthened along the projecting mouldings by layers of twigs inserted between the masonry courses. Low reliefs, showing what evidently is meant to represent leaves of the Bodhi tree and the Triralna symbol, could be made out on the plaster of the drum (Plate 32) and were the only traces of ornamentation on the Stûpa.
That it was different with the interior wall of the cella became increasingly certain during the afternoon of the same day that the clearing of the circular passage proceeded, and fragments of painted stucco kept cropping up rapidly from the débris in its north-eastern and south-eastern segments. Yet when the digging had there reached a level of about four feet from the floor, and the topmost portion of a cleverly painted dado, showing the heads of fine winged angel-like figures, began to disclose itself on the wall (Plate XL, XLI), my surprise was so great that at first I found it hard to believe my eyes. Not here, close to the desolate salt-wastes of Lop-nor and among the ruins of what seemed the very last outpost of Buddhist Central Asia towards China, could I have expected to come upon what looked like late classical representations of Cherubim. And what had these graceful heads, which seemed to recall figures familiar to early Christian art, to do here on the walls of what was beyond all doubt a Buddhist sanctuary ?
As I eagerly cleared head after head with my bare hands, in order to prevent any chance of damage, in the surviving portions of the dado along the north-east and south-east walls, I could not long remain in doubt that the classical influence was far more marked in these frescoes than in any remains of ancient pictorial art which I had so far seen or knew of, whether north or south of the K`un-lun and Hindukush. There was much in the vivacious look of the large fully-opened eyes and in the expression of the small dimpled lips and the slightly aquiline nose to call up in my mind recollections, gathered long years before in the distant West, of those fine Levantine-looking portrait heads preserved for us on painted panels from Fayyûm mummies of the Hellenistic period. A faint suggestion of Semitic strain in the features presented by some of the heads seemed to support the
Cf. above, p. 38 ; Pl. 2. ' See above, pp. 38, 247; Pl. 2, XXXII.
3 Cf. Ancient Khotan, i. pp. 83 sq.; above, pp. 37 sq.