tunnel connecting the antechapels had been driven through the rock, obviously at a later date as proved by the broken wall-paintings.a The uniformity with which this disposition is repeated in all shrines points to execution upon a definite plan and at no great distance of time, and with this
the uniform style of their decoration seems in full accord.° .
Throughout these cave-temples the walls bear paintings in tempera closely resembling in style those of Chien-fo-tung. The walls of the porches ordinarily display on the one side rows of men dressed in dark-red robes with wide-brimmed black hats, and on the other processions of ladies wearing rich wide-sleeved dresses and that elaborate coiffure with flowers, bands, and pendants around bulb-shaped caps which the examination of the paintings brought away from Chien-fo-tung has allowed us definitely to associate with the fashion prevailing in the Tun-huang region during the tenth century. There can be no doubt that these figures carrying offerings, which also recur on the side walls of the entrances to the cellas, just like the corresponding types seen on the walls of some of the Chien-fo-tung shrines, are intended to represent donors and donatrices. The walls of the cellas and antechapels are decorated either with processions of large, richly robed Bodhisattvas moving under ornate canopies or with a variety of panels, showing familiar scenes from Buddhist Heavens ;8 Buddhas enthroned among rows of Bodhisattvas ;8 large representations of Manjugri and Samantabhadra on their traditional Vâhanas, etc.10 A comparison of the photographs illustrating a few specimens of these decorative themes with the reproductions of corresponding mural paintings at Chien-fo-tung will suffice to bear out the very close resemblance in style already referred to. The same is the case also with the representation of the curious ' wind scene' legend which I had occasion to describe above as seen on the wall of Ch. xvi at the ` Thousand Buddhas', and which is found here again with identical scheme and details on the back wall of two cellas."
The great uniformity of style and technique displayed by the frescoes in this whole series of cave-temples is striking, and suggests that all of them are more or less coeval reproductions of the same prototypes. That these were to be found among the mural paintings of Chien-fo-tung I could scarcely doubt at the time, with the recollection of the latter still fresh in my mind. The execution of the paintings in this main group of Wan-fo-hsia grottoes impressed me as generally inferior to that of the ' Thousand Buddhas'. But whether this is due to distinctly later production or else to the employment of less skilful hands I was unable to determine. I noticed particularly the often careless drawing of outlines, and the coarse washes which replaced them in frequent instances suggested production by stencils. I may add here that the relative lowness of the walls—they seemed rarely to rise much above Io feet—and a corresponding flatness of the ceiling seemed evidence also of more limited outlay and resources on the part of those who dedicated the Wan-fo-hsia shrines.
I may now proceed to record a few notes about particular grottoes. In x11 the left wall of the porch has numerous sgrafti in Chinese, Uigur, and Tibetan characters, and among them also one in badly faded Arabic writing. Those in Chinese, which on account•of their dates I had copied by Chiang Ssii-yeh, proved to belong to the Caih-shun and Chih-chêng periods (A. D. 1330-3, 1341-68)
6 The opening of such a later tunnel, walled up again, is seen in xvu, Fig. 247.
6 There is no evidence that these upper story shrines were ever accessible except from the northern end. Hence it might be assumed that these excavations were commenced from that side and successively extended. But it must be remembered that their level above the terrace ground below is not so great as to preclude the use of scaffolding, which
would have expedited construction or at least permitted it to
be carried on simultaneously at different points.
See Fig. 246.
8 For a specimen, see Fig. 247.
See Fig. 246.
10 See Fig. 247 on right.
" In xix and in shrine n of the left bank (Fig. 245);
cf. above, pp. 935 sq.; Figs. 233, 234, 236.