3 feet. Embedded in this there was found in excellent preservation the fine wooden panel Mi. ix. -001 (Plate CxXVIi), with groups of relievo figures delicately carved in pure Gandhdra style. The panel, close on I I inches high, is semicircular at the back, and was evidently fixed to a wall or other background by nails, of which one in bronze still adheres. The flat upper portion of the front shows two scenes, obviously from the Buddha legend, one above the other. A projecting portion at the bottom is divided into two niches, the one on the right holding two kneeling worshippers, perhaps meant for donors, the other the figure of a standing Buddha with staff. The interpretation of the two scenes above, in both of which the haloed Buddha appears nude except for a loin-cloth, is uncertain. But in the lower scene the pose of one of the worshippers, prostrated below the Buddha's feet with the crown of the head touching the ground, suggests the Dipankara jâtaka, so frequently represented in Gandhâra relievos.11 The modelling of the figures and the execution of the whole follow the style of Gandhdra sculpture so closely that direct import of the little relic from the Indus region would suggest itself, were it not for the head-dress of the two worshippers below, which looks Central-Asian, and for the fact that the great mass of stucco relievo fragments found at other ruined shrines of the site shows the impress of Graeco-Buddhist art quite as strongly. In the same cella was found also the well-carved wooden bracket Mi. ix. 002 (Plate cxxviiI).
The remaining cellas of group I were cleared without any finds of interest being made. In the shrine ix. a, however, which is somewhat larger and belongs to the first type, partial clearing showed, in the passage by the side of a central image base, remains of relievo figures badly decayed through damp.12
In group II, occupying the middle ridge, finds were made mainly among the small cellas to the south, the walls of which were badly decayed and evidently had fallen early. From a narrow passage, i, adjoining a small cella at the southern end of the row that has been completely destroyed there were recovered numerous fragments, Mi. i. 004. a-h (Plate cxxxviI), of a valance in richly painted stucco which must have adorned the border of a projecting corner. The design, with its floral band on the top and the rows of triangular hanging ` swags ' and tassels below, looks distinctly like an elaboration of the pattern found painted valance-fashion on the walls of the central hall in N. III of the Niya Site.13 Among the fragments of small stucco relievos, which probably occupied projecting friezes on the passage wall, the well-modelled face of a Buddha, Mi. i. 001 (Plate Cxxxii), may be specially mentioned. Before a small niche in i, retaining part of a lotus base in plaster, there were found embedded in the flooring four Tang coins, just as they had been deposited as a votive offering. Two bear the legend K`ai yiian, two the nieiz-hao Ta-li (A. D. 766-79) ; all are in excellent preservation, showing practically no wear. On the north side of the passage i was another small and badly destroyed cella, ii. This yielded a number of small relievo fragments, some from statues of, or over, life-size (see e.g. Mi. ii. 007, Plate CXXxIX), and two well-carved hands from wooden statuettes, Mi. ii. ool, 009.
Beyond this again there was found a narrow passage, xxiii, enclosing what appears to have been the base of a small Stupa, now completely destroyed.. Here the débris had helped to preserve a considerable number of interesting wood-carvings, such as the ornamented brackets Mi. xxiii. 008I I (Plate CxXVIII), including two with grotesque dragon heads ; panels with fine floral decoration like Mi. xxiii. 0015 (Plate cxxviiI) ; the upright Mi. xxiii. 0012 (ibid.), reminiscent of Lou-lan designs. The staffs with elaborate lathe-turned heads, Mi. xxiii. 0013, 0019-22, may have formed part of a balustrade round the Stûpa. The fine architectural tile, Mi. xxiii. 1 (Plàte Cxxix), shows
" Cf. Foucher, L'art du Gandhdra, i. pp. 273 sqq. p. 192, and mentioned among those where tentative diggings
'= This shrine appears to be the same as the one were effected without special results.
numbered 3 by Professor Grünwedel, Altbuddh. Kullställen, 13 See Ancient Kholan, i. p. 333; ii. Pl. VII.