Sec. ii] EARLY CHINESE PILGRIMS TO UDYANA II
and his horse had room to pass.31 Hui-shêng's record notes besides that Shê-mi lay to the south of Po-chih, and that its inhabitants did not believe in the Buddhist religion but served divers divinities.38 I have referred above to the passage of the Tang Annals which describes Shê-mi as adjoining Chieh-shih or Chitral on the west and the south. Reference to the map shows that this description accords exactly with the relative position of Chitral and the Kafir territory, which even in quite recent times reached both sides of the Kunar Valley above and below Arnawai. It is equally evident from the map that it is the Bashgol Valley, with its numerous and large Kafir settlements (Fig. 8), which occupies the position indicated due south of Po-chin or the Zebak-Warduj Valleys, whereas Chitral lies partly to the east and partly to the south-east of them. It would have been difficult among the high and barren mountain-spurs of Chitral for the travellers to believe themselves to be emerging from the Tsung-ling, whereas the description suits well the more open and fertile Kafiristan valleys. Finally the statement about the absence of Buddhist worship would not fit Chitral, where surviving remains actually attest its presence about Sung Yün's time, while on the other hand, in the Kafir valleys, the worship of ` divers divinities ' has continued to our own days without a trace of Buddhism having ever existed by its side.
Though the route over the Mandai Pass has not been regularly surveyed, there is trustworthy information to show that it is practicable for laden animals in the summer and autumn, probably under conditions much like those on the neighbouring Dorah.3s Since the Afghan occupation of Kafiristan a regular trade route appears, in fact, to have been opened up to the head of the Bashgol Valley and across the pass. That even before this a good deal of trade made its way thither from the Badakhshan side is evident from a remark of Sir George Robertson who ascended the Bashgol Valley in 1892.4°
To revert to an earlier period, it is noteworthy that the route in Marco Polo's account, by which the Mongol partisan leader Nigûdar, ' with a great body of horsemen, cruel unscrupulous fellows', made his way from Badakhshan ' through another province called Pashai-Dir, and then through another called Ariora-Keshemur' to India, must have led down the Bashgol Valley.4' The name of Pasiaai clearly refers to the Kafirs among whom this tribal designation exists to this day,42 while the mention of Dir indicates the direction which this remarkable inroad had taken. That its further progress must have lain through Swat is made probable by the name which, in Marco Polo's account, precedes that of ` Keshemur' or Kashmir ; for in the hitherto unexplained Ariora can be recognized, I believe, the present Agrôr, the name of the well-known hill-tract on the Hazara border which faces Buner from the left bank of the Indus.l3 It is easy to see from any
Route across Mandal Pass.
Marco Polo on Nig13dar's inroad.
97 See Chavannes, Voyage de Song Yin, p. 28.
38 See ibidem, note 3.
39 The route leading up the Bashgol Valley to this pass is described quite accurately in the ` Surveys' dating back to circ. 1789-90, which Raverty has edited in his Notes on Afghanistan, p. 149. The pass is there called Apa-luk. Raverty has called due attention to the importance of this route connecting Badakhshan with the Kabul river valley and Peshawar.
10 See Robertson, The Kafirs of the Hindu-hush, p. 305. That Sir George Robertson was able to cross what he calls the Mandal Pass, at the very beginning of June, without exceptional difficulties (loc. cit., pp. 312 sqq.), in spite of the snow which was then still heavy, confirms the estimate given as to the practicability of the route. It must, however, be noted that the pass by which he crossed was clearly on the
side route branching off to the Munjan River headwaters and thus identical with what the map marks as Wulf Pass.
" See Yule, Marco Polo, i. p. 98, with note, p. 104; Ancient Khotan, i. p. 14, note 28.
42 See references in Yule, Marco Polo, i. p. 165 ; also Grierson, Z.D.M.G., lxvi. p. 70, note 1.
" In my note on Râjal. viii. 3402 I have shown that the modern form Agrôr is the direct phonetic derivative of the Sanskrit Atyugrapura, the name by which Kalhana mentions this hill tract in connexion with a contemporary expedition to Ura§â or Hazara. The intermediate Prakrit form *Ayugraura, which the phonetic development there discussed presupposes, would help to account for Marco's Ariora. Cf. also the form '194yovpor in which Atyugrapura > Agrôr is presented by Ptolemy, Geography, VII. i. 45, as one of the cities ' of the "Apvn territory, i. e. Ura0.