Sec. in THE STUPA CELLA M. iii AND ITS WALL-PAINTINGS 495
impression of a direct linking with the Hellenized Near East. There was much in the conventional uniformity of the design, and also in the rapid execution of details, to suggest the brush of a decorator repeating a pattern of distant origin, fixed both in motif and composition. And yet an expression of vigorous artistic feeling seemed still to survive in the lively directness of gaze and pose, and in the simple ease of the outlines shown even by the graceful upward curve of the short fluttering wings. Puzzled as I was by much that this first rapid inspection had disclosed, I could at least feel quite certain that work of this type could not possibly have been produced in the time of Tibetan occupation or in the period of Chinese predominance immediately preceding it. I was still wondering how to account for the distinctly classical style in the treatment of these ` angels ' and to understand the purport of what seemed like a loan from early Christian iconography, when a fortunate discovery supplied definite palaeographic evidence to throw light on the question of date.
From the rubble of broken mud-bricks and decomposed plaster filling the south-western segment of the circular passage there were recovered in succession three large pieces of finely woven buff silk, M. III. ooi 5 (Plate Xxxix), which proved to have formed part of a votive banner or streamer, about four feet long in its present torn condition and originally over one foot wide. It was decorated with two inwoven bands of narrow lines, in harmonizing tints of red and green, running its whole length. Written along and over a red line which divides the intervening space were nine short inscriptions in Kharosthi, five of them complete. Their neat clerical handwriting is not different from that found in the great majority of the Kharosthi documents on leather and wood. from the Niya Sites, or in the more carefully penned Kharosthi records from Lou-lan. Hence the conclusion seems justified that the deposition of the inscribed streamer must date back to approximately the same period to which those documents belong, viz. the third and early fourth century A. D. I was struck from the first with the fact that the ink had remained remarkably fresh and black even without the protection which careful folding in the case of the leather documents, and wooden envelopes in the case of the tablets, had provided. This, combined with the excellent condition of the silk material, where not torn, makes it very unlikely that the gift of the inscribed silk streamer could have preceded the abandonment of the shrine by any long time. And thus I was soon led to infer, as the simplest explanation of the obviously early character of the wall paintings, that the shrine M. Iii belonged to an older settlement, which had been deserted about the same period as the sites of Niya, Endere, and Lou-lan. For the subsequent reoccupation of Mirân in Tibetan times the observations made at the ruins of Endere furnish an exact parallel.
The recurrence of the same words at the end of each of the short Kharosthi inscriptions had helped, even at the time of discovery, to suggest to me that they were of a votive character. This has been fully confirmed by the decipherment which my learned collaborator, the Abbé Boyer, was kind enough to undertake, and the results of which were published by him in 1911.4 It has proved that each of these short inscriptions contains a prayer in Prâkrit for • the health of a certain person, some of them also for that of his family, and that the phrase used in expressing it (arughadachinae bhavadu, i.e. Skr. ârogyadaksivâyai bhavatu) is identical with that regularly found in Indian epigraphic records of the Kusana period where they indicate the benefits hoped for by pious donors in return for their gifts. Of the seven names preserved, three have been recognized by M. Boyer as plainly Indian, viz. Asagosa, Caroka, Samanaya, corresponding to the Sanskrit forms A.vaghosa, Cdruka, and gramaltaka. I believe that he is equally right in classing the two female names Friyâna and Firina as of Iranian origin, Fryâna being the well-known Avestic designation of a Turanian family.4â Mitraka might be either Indian or Iranian, according to the