Sec. iv] DESERT SITES TO THE NORTH OF JIYA 129
the first 4 Tim ' was duly sighted. It proved the remains of a completely ruined Stûpa, rising about twelve feet above the sand and displaying a base about twenty-four feet square. The sun-dried bricks on the top measured fifteen by nineteen inches, with a thickness of three. A bare depression running southward showed near by remains of mud walls, almost completely eroded, and much pottery débris. Specimens of the latter, including two pieces showing a dark green glaze, are included in the list of antiques at the end of this section. I also picked up a small pencil of Chinese ink or graphite and several tiny Chinese coins, without legend and showing much-clipped rims, which Mr. J. Allan is inclined to ascribe to the fifth century A. D. This small ` Tati ' and the adjoining site was said to be known to ` Otanchis ', or collectors of fuel from the oasis, by the name of Kök-kumärish.
About a hundred yards farther south a narrow irrigation cut, only two to three feet wide, was Traces of
traceable for a distance of about thirty yards, running from south-west to north-east. Still continuing ancient
y y g g canals.
to the south I came, after another two hundred yards, upon the marks of an ancient tank, about fifty
feet across. Its earth embankment, once hardened by moisture, still rose clearly above the level of
the surrounding ground which wind erosion had lowered. Even the little earth-cone known as
dömbel, which the villagers to this day invariably leave in the centre of their tanks,6 was well
recognizable. Some sixty yards to the south rose another much-decayed mound of sun-dried bricks,
about twenty-five feet in diameter and standing about eight feet above the surrounding sand.
Though no structural outlines survived, it could scarcely be anything else than the remains of
a Stûpa. Close to the south of it the curving line of a canal, about eight to nine feet wide, could be
followed for over thirty yards. Its banks were still fairly hard and showed in places salt efflorescence
left by moisture. At its bottom I found several much-fissured pieces of ancient wood from the Terek
or cultivated PoJulus alba. Also a narrow side-channel branching off to the north-west was easily
made out. The same canal according to Kasim, my guide, reappeared at several points further to the
south, temporarily left bare by the high dunes. It manifestly came from the Yurung-lash which
with the right bank of its present flood bed touches a line within seven miles or so west of the site.
The ` old house ' of which my guides had spoken, lay hidden away in a curious bay-like depres- Ruined
sion fringed by tamarisk-cones and steep dunes a short distance eastwards. It was easy for me to shrine of
recognize that the scanty remains were those of a rectangular temple cella built in timber and ârish. plaster, measuring inside twenty-seven feet three inches on the north and south and twenty-four feet ten inches on the other two sides. After the fashion of the Dandân-oilik shrines the cella was enclosed on its four sides by an outer passage about seven and a half feet wide. The rapid clearing effected by the small party I had brought with me, showed that the remains of the cella walls rose nowhere more than two feet above the plastered floor. Their thickness was about six inches. The original timber framework had completely perished under the influence of damp from subsoil water, leaving matrices of posts bout six inches broad.
But where accumulation of drift sand in the corners had preserved the stuccoed wall surface this still retained traces of the original decoration in colours. This consisted chiefly of floral tracery painted in black over red ground. A large flower about seven inches across recurred in several places and closely resembled the four-petalled clematis-like flower with which I was familiar from the ancient wood-carving discovered at the Niya Site in 1901, e. g. the decoration of the wooden chair (N. vii. 4).7 Elsewhere the remnants of decorative motifs recalled the fresco ornamentation in the central hall of the ruined dwelling (N. III) at the same site 8 both in design and colouring. This was particularly the case on portions of painted stucco which were found at the foot of the inner wall in
8 See Ancienl Kholan, i. p. 379. 7 See ibid., i. p. 397 ; ii. PI. LXVIII.
8 See ibid., i. p. 333 ; ii. PI. VIL