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

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

' Convent of the Three Regions.'.

help, too, might not prevent Chinese texts of historical or literary interest from being left behind, even in the bundles that we were able to search.

Deprived of the knowledge which alone could guide among this tantalizing embarres de richesses, I had to rest content with a few encouraging glimpses. Thus it was first in a Chinese document dated A. D. 925 and bearing the seal of what evidently was the chief monastic establishment of the site that

I lighted upon the old name of Chien-fo-tung, which here figured as San-chiedi-ssuv   n. 4, the
` Convent of the Three Regions '. The local application of the name could not remain in doubt when we found again and again complete rolls of Buddhist texts stamped with the die of the ` Convent of the Three Regions ' (see e.g. Ch. cv. ooi, Plate CLXVII), and thus clearly marked as having formed part of the monastic library which had supplied the great mass of the manuscripts deposited in the chapel. The designation is no longer known, but the suggestion was made to me locally that there may be some connexion between it and the three divisions still distinguished among the cave-temples of Chien-fo-tung and known as Shang-ssû E , Chung-sss Ill ,

and Hsia-sstc   4. [For another and more likely interpretation, see Add. & Corr.]

Then again there were to be found, folded up or otherwise hidden among rolls of Chinese texts, spirited drawings or woodcuts representing sacred figures or scenes (see; Plates XCI-CI for specimens) and even fragments of illuminated texts, such as the fine booklet seen in Plate XC. Without any expert knowledge I could recognize their artistic value and the interest attaching to the exact dates, mostly of the latter half of the tenth century, which the several block-printed sheets, such as those seen in Plates CI, CII, exhibited. But still more interesting it was when I found that an excellently preserved roll (Ch. ciii. 0014, Plate C), with a well-designed block-printed frontispiece, had its text printed throughout, showing a date of production corresponding to A. D. 868. Here was conclusive proof that the art of printing books from wooden blocks was practised long before the Sung period to which the earliest previously known specimens belong, and also that in the ninth century the technical execution had already reached a level practically as high as the process permitted.

It had cost five days of strenuous work to extract and rapidly search all ` miscellaneous' bundles likely to yield manuscripts of special interest, paintings, and other relics which I was eager to rescue first of all. It was fortunate that these bundles, being less convenient building material than the tightly wrapped uniform packets of Chinese and Tibetan rolls, had been put by Wang Taoshih mostly on the top or in other more or less accessible positions, when he had last stuffed back his treasures into their original hiding-place. But there still remained, rising against the walls of the chapel, that solid rampart of manuscript bundles. I was naturally anxious to have these, too, cleared out in order to be able to search them rapidly, but felt scarcely surprised when this proved a troublesome undertaking in more than one sense. We had so far succeeded in overcoming the Tao-shih's relapses into timorous contrariness by discreet diplomacy and judiciously administered doses of silver. But now„ when faced by the heavy labour of clearing out the whole chamber and by the increased risk of exposure thus involved, the priest became distinctly refractory.

So prolonged efforts and fresh assurances were necessary before, under protest as it were, and after carefully locking the outer gate of the temple, he set to this great toil. Considering how little adapted his slender physique was for it, I felt glad that he now allowed himself to be helped by a priestly famulus whose discretion could be relied upon. By keeping them both steadily to the task in spite of renewed remonstrances, I succeeded in having by nightfall of May 28 the whole of the regular ` library bundles ' taken out and transferred to neat rows, mainly in the spacious cella of the temple. Those containing Chinese rolls proved to number about 1,050 in all, each holding on a rough calculation an average of more than a dozen separate manuscripts. To these had to be

\Voodcuts and earliest. printed text.

Extractions of ` miscellaneous' bundles.

Clearing out `library' bundles.