and this makes it at present more difficult to determine where the convent of the ` Serpent medicine' and the neighbouring great Stüpa of Su-mo are to be exactly located. They were both believed to have been erected at places where Buddha in a previous birth as I ndra had changed himself into a serpent and allowed pieces of his body to be eaten in order that the people suffering from famine and pestilence might be cured. The much-decayed mounds of Saj5ar and Andân-dheri, in which Colonel Deane had thought to recognize remains of these structures, lie close to the point where the route to Bajaur, and with it the present military road to Chitral, turns sharply west towards the Katgala Pass. I had paid them a visit at the close of 1896, and ascertained then that the supposed name of Süma was not known locally. For excavation, which alone could help definitely to settle the question as to the identity of the remains, there was no time on my rapid passage of 1906.
To the north of the valley, and by the side of a steep cliff, the Hsi-pi-chi mentions a Stüpa near a healing spring which Buddha, in a previous birth as a king of peacocks, was believed to have pecked open with his beak. Colonel Deane has suggested the location of this site at the village Gudai-khwar, high up on the steep slope of the Laram Mountain, and some ten miles to the north of Chakdara, where an abundant spring was reopened in recent times under rather remarkable circumstances. The visit I paid to this spot at the end of 1897 makes me inclined to accept this identification as probable, even though I failed to find remains of a Stüpa above ground, or the traces of the peacock's feet which pious Buddhists' eyes used to see on the rock.20
The only sacred locality in H süan-tsang's narrative still calling for mention is the ` dragon lake ' on the mountain called Lan-o-lo. A legend related at great length connected it with an exiled Sakya who married the dragon or Naga's daughter, and became the founder of the royal house of Udyana. The direction indicated, a hundred and forty or a hundred and fifty li to the north-west of a sanctuary which itself must be looked for within a distance of four or five miles north of Manglaur, makes it appear probable that the lake of Saidgai is meant, which information collected by Colonel Deane places at the head of the Aushiri Valley draining into the Panjkôra from the north-east near Darôra.21 This mountain tract between the Upper Swat River and the Panjkôra still remains unsurveyed, and I am hence unable to add details to the arguments adduced by Colonel Deane for his very tempting identification.
The only other Chinese pilgrim whose visit to Udyâna is detailed in the records so far made accessible, is Wu-kung. The travels of this humble successor of Hsüan-tsang (A.D. 751-79o) fall into a period when Chinese influence in the ` Western Regions ' was rapidly waning, and he must have been one of the last pious travellers to make his way to the Indian North-West through Central Asia. Though he appears to have spent a very considerable portion of his long Indian residence in Gandhâra and the neighbouring Udyâna, his notices here as elsewhere are meagre. But his record of the route by which he travelled from Kashgar to Udyâna, brief as it is, presents points of interest.22 He was attached to a Chinese envoy, dispatched to the ruler of Chi-pin (i. e. the Kabul Valley and Gandhara) in response to a mission thence which had reached the Imperial court in A.D. 75o, and he travelled thus across the Ts`ung-ling Mountains, or the Pamirs, to the territory of the ` Five Ch`ih-ni or Shih-ni'. This corresponds, as a passage of the Tang Annals quoted by M. Chavannes proves, to the present Shighnan on the Oxus.23 Hence the party proceeded through
40 See Deane, loc. cil., pp. 658 sq. The large stone bowl with which the mouth of the spring is there stated to have been closed before its rediscovery, and which now serves for a flower bed in the garden of Government House, Lahore, certainly belonged to some Buddhist Stûpa ; cf. Foucher, L'Art du Gandha'ra, p. 186, note 2.
" Cf. Deane, loc. cil., J.R.A.S., 1896, p. 661. The lake
is said to be more than a mile in length. The popular belief in ` Jins ' constantly seen on the banks of the lake points plainly to the survival of the old local tradition.
22 See Chavannes and Lévi, Ilineraire 'Ou-k`ong, J. as., 1895, pp. 9 sqq. (reprint).
" Cf. ibidem, p. 10, note 3; Chavannes, Turcs occid., p. 162, note 4.