Sec.ii] THE SHRINE F. xii AND THE CHRONOLOGY OF THE SITE 1253
The pilgrim I-tsing, after telling the pious legend which accounted for the ogress' conversion by Buddha and her subsequent worship, informs us that ` the image of Hâriti is found
either in the porch or in a corner of the dining-hall of all Indian monasteries, depicting her as
holding a babe in her arms, and round her knees three or five children '.5 It is thus that we find the goddess represented by the side of the entrance to the shrine. One of the children
around her embraces her left breast ; three others bestride her shoulders and right forearm,
while the fifth, a boy, is. seen _dancing by her right side. Besides these figures seen in the panel
there could be faintly distinguished, on the almost completely abraded surface below by the left leg, two little figures in tight-fitting dress gambolling about, while near her right foot a small naked boy seemed to be warding off a blow struck probably by another little figure completely effaced. It is curious to observe that exactly the same number of Hâriti's children which our fresco appears to have shown is found also in the Turfân linen painting, which otherwise differs greatly in composition and style.
Apart from its iconographic significance, this painting of the deified ` demon mother of the Features of
children ' presents also other points of interest. In her short rounded moon face ', combining âoddess'
matron-like sadness and sweetness in its dreamy expression, it is impossible not to recognize the influence of a Persian type of beauty. But it is far less strongly marked here than in the faces of the princess and her attendants in the painted panel from Dandân-oilik, representing the legend of the origin of sericulture at Khotan,° or in the faces of Vessantara's queen and of the girls in the daclo at the Mirân temple M.. v.7 The symmetric love-locks of these figures appear, however, here too, as do also the strings of pearls in the hair. In strange contrast with these quasi-Western features are the frightfully distended lobes of the ears and the strongly marked folds of the neck, which M. Foucher calls ` classic in India '. Among the details of the elaborate dress 8 it may suffice to mention the resemblance of the short-sleeved close-fitting jacket to that worn by the Princess in the Dandân-oilik panel just referred to.
The wall on the opposite side of the entrance was found, unfortunately, broken. As M. Foucher has justly pointed out, ` we should have expected to see there the genius of riches, the usual counterpart of the goddess of children '. This ` genius with the golden bag ', whom I-tsing's above-quoted account mentions as usually found seated at the porch of monasteries in India, and whose identity with Kuvera, the god of wealth, has long ago been established, is often found represented side by side with Hâriti in Gandhâra relievos.9 His conjunction with her would have been particularly appropriate at a shrine of ancient Khotan ; for we know that Kuvera in his aspect as the ` guardian king of the North ', or Vaigravana, was of old particularly worshipped at Khotan as the ancestor of the royal race and the protecting divinity of the country.'°
The excavation of the shrine F. xii, which completed my work at the site, was rewarded also by Find of
a manuscript find of value. The initial clearing along the south-east wall brought to light over half Sanskrit
a dozen fragments of Pathi leaves in Central-Asian Upright Gupta script and in Sanskrit, and later I had the satisfaction of extracting myself, from the sand near the low platform once bearing the images on the north-west side, an excellently preserved packet comprising 33 complete folios of a Sanskrit text. The leaves, measuring about 14 by 41. inches with a string-hole at about one-third of their length, stuck close together, but could subsequently be separated without damage at the British
s Cf. Takakusu, A Record of Buddhist Practices, by I-tsing,
6 See Ancient Khotan, ii. Pl. LXIII, D. X. 4.
' See above, pp. 518 sq., 525 sqq. ; Figs. r34, 135, 138-40, 143.
8 For these the three-colour reproduction in Desert Cathay, ii. Pl. XI, may be consulted with advantage.
8 Cf. Foucher, The beginnings of Buddhist An!, pp.141 sq., 282.
10 Cf. Ancient Kholan, i. pp. 156-8.