Sec. in EXCAVATION OF THE ENDERE TEMPLE 427
seems to be little more than a later revision of it, with occasional expansions and substitutions of glosses for earlier and more ambiguous terms '. Yet the philological importance of these fragments is great ; for together with the smaller Tibetan text fragments and the sgraffiti to be mentioned • presently, they are the earliest specimens of Tibetan writing so far known. The archaeological evidence set forth below clearly proves that none of the MS. remains discovered in the Endere shrine can have been deposited there later than the eighth century. In the light of the chronology thus fixed the archaic peculiarities of orthography, first noticed by Dr. Barnett in the Sâlistamba-sutra, and also by the Rev. Mr. Francke in the two religious poems, assume their true significance. But it is, perhaps, even more noteworthy that by the side of these pre-classical spellings the latter pieces, as fully explained in Mr. Francke's General Note, furnish also instances of an orthography agreeing with modern dialectal forms. I am not competent to express an opinion as to the very interesting questions which, as forcibly set forth in the remarks of my learned collaborators, these observations are bound to raise as to the real age of the invention of the Tibetan alphabet and of the first introduction of Buddhism into Tibet 12. But it appears to me that the fact of the Endere texts showing a writing which does not differ from the modern dbu-can script may well deserve consideration in this connexion. Seeing that by the evidence of our Endere texts this script is shown to have already assumed in the eighth century that final form in which it continues to the present day, while the classical orthography so intimately connected with it is proved to have been at that time already archaic, it seems difficult to resist a doubt as to the correctness of the tradition which places the invention of both the Tibetan orthography and characters only about a century earlier.
But these miscellaneous votive deposits of MSS., or pieces of such, in Brahmi and Tibetan scripts were not the only written remains unearthed in this small temple. A very interesting discovery was that of two tiny fragments of birch-bark, showing each a few characters in Brâhmi, which were found sticking to the plaster of the south wall surface about i ft. above the floor and about 6 ft. from the south-east corner. There was nothing to indicate how they had got there, and what the character might have been of the MS. leaf from which they had become detached. The few Aksaras visible in each fragment seemed Sanskrit, and the writing of an upright Gupta type, which looked older than that of any Brâhmi MS. finds at this site or Danclân-Uiliq. Curiously enough, a few thin scraps of birch-bark, with traces of a character or two in Brâhmi, turned up among the sand and débris of the hole which had been dug into the central base. Is it possible that the latter once contained some deposit, as usually inserted into the base of Stupas ?
In different places on the floor, but in no case on or near an image pedestal, I discovered three small pieces of paper with Chinese writing (E. i. 8, 36, 44). Though they are mere fragments, M. Chavannes' translation, contained in Part iii. of Appendix A, clearly shows that they belonged to secular documents, such as those found in the Dandan-Uiliq dwellings and shrines, and not to religious texts. In E. i. 44 we have manifestly the portion of an official record, for it mentions the petition of a certain functionary, as well as the `commander-in-chief of the army of Tso yü lin, Wang Q) Chih-chiang'. E. i. 8, 36 are scraps of papers manifestly relating to private affairs, possibly petitions. None of them contains any date, yet even thus they may claim chronological value ; for in view of what we have learned above as to the events which finally brought to a close Chinese supremacy over Eastern Turkestan as main-