A very interesting find made below the niche to the left of the north-west corner was a large and fairly well preserved Pôthi leaf on silk (see PI. CXXI), measuring 12 inches by 5, coated with a chalky substance and inscribed on both sides with a Buddhist text in Sanskrit and in fine Brâhmi script of the Gupta type.° It is, as far as my knowledge goes, the first specimen found of a manuscript on silk in Indian language and writing. In India the use of cotton fabrics as writing material is attested by Nearchos 10 From the foot of the west face of the base numerous tiny fragments of a Sanskrit manuscript on birch-bark were recovered, while near the south-west corner there came to light also some minute fragments of a palm-leaf manuscript. These remains, fragmentary as they are, of Sanskrit manuscripts on birch-bark and palm-leaf, found both at Koyumal and Bash-Koyumal, are of special interest as suggesting import from India by the direct route that still leads from Charkhlik across the Tibetan plateaus to the south. The palaeographic character of their writing suggests that the shrine and the small fort around it were occupied during early Tang times.
The attribution to this epoch finds support in the fragments of stucco relief figures, including the portion of a Buddha head nearly life-size (B. Koy. i. 05, Pl. XX), and other small remains of wood and plaster, found at the foot of the Stüpa base and described in the List below. Special mention may be made of a number of broken fragments of coloured silks (B. Koy. oi), one inscribed with a Chinese character, which evidently belonged to votive banners. There were traces of the broken walls of small structures built against the enclosing wall both inside and outside it. The process of clearing them brought nothing to light except roofing materials and rough pieces of carved wood, of which the pegs or pins found close to the shrines (B. Koy. i. 06-7, Pl. XVI) are specimens." A small detached dwelling, III, of which the much-decayed remains were traceable about 5o yards to the NW. of the north face of the enclosure, was found completely empty.
There is no direct evidence as to the exact purpose of the circumvallation as a whole. But Purpose of
it is noteworthy that it is situated on the track which leads from Charkhlik to the mouth of the circumvalla-
valley drained by the Charkhlik river and thence to the high plateaus behind the northernmost shrine.
K`un-lun range. It lies, moreover, at a point where cultivation would have been very difficult to maintain. I am thus led to conjecture that the ruined enclosure marks a defensible post intended to guard the approach to the oasis from the mountains on the south. At the same time the little Buddhist shrine within it, close as it still is to the head of canals irrigating the present Charkhlik oasis, may well, like the ruined Buddhist sanctuaries usually to be found at the ` Su-bashis ' of Turkestan oases, and the Muhammadan Ziàrats which have in many cases succeeded them, have been an object of pilgrimage to the cultivators wishing to secure adequate irrigation for their fields.'=
OBJECTS EXCAVATED FROM RUINED SHRINE OF KOYUMAL
Finds of Indian MSS. remains.
Approxi- mate date of occupation.
Koy. or. Fr. of carved wood, prob. from openwork detail. Fr. of plain edge at top. Below, part of openwork pattern of intersecting curved bands of pearls or seeds with plain oblong boss over point of section. Probably from canopy over figure of Buddhas. 4f" x rf" x r4". Pl. XVI.
Koy. r. ox. Four frs. of painted wooden panels, in green, black and white. Subject indistinguishable. Wood perished. Gr. fr. 61' x i s-" x
9 For a description of this leaf and other MS. remains, see Mr. Pargiter's Appendix E.
to Cf. Bühler, Indische Palaeographie, p. 88.
11 These pegs may be compared with the decorated pins found at certain Lou-lan shrines (see Serindia, i. p. 398 ; iv.
Koy. r. oz. Part of wooden R. hand of Buddha, gilded,
life-size. Two middle fingers, half little finger, back and wrist remain, palm split off. Through palm is small dowel, and second at base of second finger, evidently fôr attachment of some object to palm ; cf. Ser. iv. Pl. CXXXVII, Mi. xviii. 004. Fingers slightly curved and tapering, with short-trimmed nails. Web between fingers clearly shown. Good work, well preserved. 12" x 3". Pl. XVI.
Pl. XXXV, L. B. u. 002 ; IV. 002), which certainly served for fastening wooden beams or panels to a timber framework. 12 See Anc. Khotan, i. p. 109 ; Serindia, iii. pp. Ir51 sq., n55 55 ; above, pp. 64, 69.