National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

814   EXPLORATION OF A WALLED-UP HOARD   [Chap. XXII

â subsequent chapter. But there were also discoveries which I could recognize at the time as throwing light on the history of the whole cache as well as on that of the site. As far as the conclusions drawn from them have received confirmation through expert examination in Europe, and through the results of Professor P. Pelliot's fruitful visit to Wang Tao-shih's ` treasure cave ' a year later, I propose to discuss them here, along with the archaeological indications that could be gathered on the spot.

That the great mass of more or less uniform packets containing rolls of Buddhist texts in Chinese or Tibetan I had belonged to monastic libraries was clear to me from the first. But equally certain it was that prolonged philological labours of competent experts would be needed before accurate data could be derived from them as regards the character and origin of the local collections of which they had formed part. From those packets it was easy to distinguish the ` miscellaneous ' bundles, of quite irregular shape and fastening, the special value of which had already revealed itself to me through the first day's experience. There could be little doubt that the painted fabrics, ex-votos made of textile pieces, and papers of all kinds, which . along with fragmentary rolls of Chinese and Tibetan manuscripts formed their usual contents, had been collected from shrines where they had once been deposited, and stored away here when no longer required for use. Their very irregularity had caused the Tao-shih to put most of such bundles on the top when he built up the wall-like array of what might be called ` library bundles '. There they could be reached with ease, and this proved a special inducement for him to bring them out in steady succession.

It was from these ` mixed ' bundles that I recovered most of the manuscripts with Brâhmi writing and of the Indian Pôthi shape, as well as a portion of the rolls which bear a Chinese text on the obverse and have their reverse wholly or partially covered with Brâhmi script.2 The results of Dr. Hoernle's and Professor De la Vallée Poussin's painstaking examination of these texts will be reviewed in Chapter XXIV below, and a complete descriptive list of them will be found in Dr. Hoernle's Appendix F. It will be seen from the former that the languages represented in these Brâhmi texts are mainly Sanskrit and that ` unknown ' tongue of Iranian type for which the term ` Khotanese ', now recommended by Professor Sten Konow and Dr. Hoernle, appears the most convenient provisional designation.8 The other ` unknown ' Indo-European language of Eastern Turkestan which appears'to have been used chiefly in the north of the Tarim Basin, and for which the term ` Kuchean ', in view of the strong grounds advanced by Professor Sylvain Lévi, may now be safely accepted, is represented only by a few leaves. Among the Sanskrit texts, which almost all significantly enough are of the Pbthi shape and were contained in ' mixed ' bundles, the remarkably well-preserved manuscript on palm leaves, 69 in all, of a redaction of the Prajna para-mita, Ch. 0079. a (Plate CXLII), claimed my special interest at the outset. The material clearly showed that this manuscript must have been written in India, and, as the writing is recognized by

Value of
miscel-
laneous'
bundles.

Manuscripts in Brâhmi writing.

' For a photograph showing a pile of such bundles, mainly of Chinese Siltra texts, in their original cloth wrappers, see Desert Cathay, ii. Fig. 194.

b      2 In Dr. Hoernle's descriptive list of Brâhmi manuscripts
from Chien-fo-tung, Appendix F, the Pôthis and rolls found in miscellaneous bundles can. be generally distinguished from those which were extracted from regular packets of Chinese rolls by the bundle number in small Roman figures (i, ii, xl, etc.) prefixed to the serial numbers (003, oor9, etc.) in the ` site-marks'. Pôthi leaves and rolls subsequently recovered on searching the regular packets in which they were em-

bedded bear only serial numbers (e.g. Ch. 0041, 00271, etc.). Only in a few cases, e.g. Ch. 0079. a, b; 00275, have Brâhmi texts been described without an indication of the `mixed' bundles in which they had originally come to light. I may note here that when the marking with serial numbers was made at the British Museum, no classification of the different objects (manuscripts, paintings, decorated fabrics, etc.) found in the same mixed bundle could be attempted.

For the reasons supporting the use of this designation, first suggested by Prof. J. Kirste, see now Hoernle, Manuscript Remains found in E. Turkestan, i. pp. x sqq.