Sec.iii] RELIEVOS AND FRESCOES FROM N.W. PORTION OF MING-OI' 1i97
descriptive notes on these frescoes, the pens are clearly held as if they were brushes, at right angles to the leaves, which, moreover, have their narrow end towards the writers, a position suited only for writing in vertical lines. Such writing must be either Chinese or Uigur here, and as there is nothing else pointing to the frescoes being the production of a Chinese hand, the latter interpretation seems more probable. But obviously the argument is not one to be relied upon with full certainty since we do not know what older prototype the painter-decorator may have followed.
The painting from the outer wall of the west passage, Mi. xiii. Io (Plate cxxiv), though evidently of the same period, is the work of a different and distinctly more skilful hand. It shows two rows of Buddhist devotees, evidently meant for monks, though their heads are not shaven and their garments vary in colours and ornamentation. Those in the upper row are represented as older by the indication of hair on their faces. The drawing of the outlines is very clean and firm, but there is no attempt at shading of the flesh such as is very noticeable in the figures of the chamber. Slight variations in the expression, tilt of head, etc., are introduced to break the monotony of the subject ; yet the clumsy way in which the floral patterns are painted over the robes with total disregard of folds, etc., shows the perfunctory character of the work.
Across an open court to the south of xiii is found a group of closely adjoining small cellas. In the northernmost, xiv, there were found several pieces of fine wood-carving. The once painted and gilded finial, Mi. xiv. 002 (Plate Cxxviii), may have formed part of a miniature shrine. The small but very carefully carved piece M. xiv. 003 (Plate Cxxviii), evidently from the side of a circular carved relic casket, shows a Buddha seated under an arcade. The Indo-Ionic column, the horseshoe arch, the bust rising above the spandrel, and other details are purely Graeco-Buddhist in style and can all be exactly paralleled in stone from Gandhâra relievos. The work is undoubtedly of early date, and the excellent condition of the wood even now makes it easy to believe that the little casket had been a cherished object of worship for centuries before it met with destruction. In this cella there turned up also a fragmentary Pathi leaf from a paper manuscript in Brâhmi script, apparently Sanskrit.
The immediately adjacent cellas yielded no finds, but better results rewarded the clearing of a group of small structures which occupy lower terraces on the slope further south. From the shrine xv, a plain cella about 12 feet square, there came to light a considerable number of relievo fragments, closely allied in style to those found in x—xii and furnishing useful supplementary pieces.' The finest among these is the large head of a Bodhisattva, Mi. xv. oolo (Plate CXXIX), which in graceful modelling and careful execution can bear comparison with the best of Gandhâra sculptures of this type. All of them had been hardened by fire and in consequence lost their colouring.
Yet it was from the débris of this cella that two of the best pieces of wood-carving found at the site were recovered. One is the section of a carved wooden arch, Mi. xv. oo29 (Plate Cxxviii), in very good preservation and still richly gilded, showing a series of carefully carved seated Buddhas. It is highly probable that the piece, which has a tenon at one end and a mortice at the other, belonged to the border of a large vesica of a type represented by relievo fragments from Rawak and elsewhere.8 More welcome even is the other wood-carving, the excellently modelled statuette, Mi. xv. 0031 (Plate Cxxvii), close on io inches in height, representing a Lokapâla, in the best Chinese style of the Tang period. Full of character and simple dignity, it is undoubtedly the work of a Chinese artist. In it we have tangible evidence of that reverse current of influence from