populous villages. Reoccupation by Wakhi settlers (Fig. 76) is recent and proceeds but slowly, the total number of houses in the valley probably not exceeding thirty.
This change of conditions has its exact parallel in what I had occasion to observe about the ancient cultivation in the upper Yârkhun valley about Shuyist, the A-shê yii-shih-to of the Chinese Annals, mentioned as the ` capital ' of Chü-wei during Tang times.6 It presents a direct antiquarian interest in view of the probability, which I have indicated elsewhere, that Kao Hsien-chih's route on his return from Little P`o-lii to the uppermost Äb-i-Panja led through Hunza.7 If this assumption is correct, it may be safely concluded on general topographical grounds that the Chinese commander made his way through the Chapursan valley and across the Irshad pass ; for this is the most direct route between Hunza and Wakhan. Besides being some eighteen miles shorter than the alternative route by the Kilik and Wakhjir passes, it crosses only a single watershed, and that not higher than the Wakhjir pass, between the sources of the Ab-i-Panja and the Taghdumbâsh river. The conclusion just indicated finds additional support in what I was able to note of the openness of the Chapursan valley and its former local resources.
It was interesting to observe how the traces of former occupation are also reflected in local legends attaching to particular spots. Thus, after passing several sites of abandoned cultivation below B aba-ghundi, I was shown, to the west of a vast lateral moraine left behind by the now receded Ishkuk glacier (Fig. 64), a well-marked cirque that once held a lakelet ; tradition places here the residence of a nine-headed monster, resembling a Naga by the description given to me, which by its daily food exactions of one human being and seven sheep caused this part of the valley to he deserted. The destruction of the monster is ascribed to Imam Muhammad Baqir, the great saint of the valley worshipped at Bäba-ghundi. The whole alluvial fan below the outfall of the great stream that issues from the Ishkuk glacier is supposed to have once been cultivated. But only at its eastern edge did I come upon the first sign of reoccupation ; this had been started a few years before by three families settled at the old site of Zudakhun. Another extensive area of former cultivation three miles farther down and a short distance above the few houses of Sipenj is known as Kampir-i-dior. Local legend connects it with the story of an old woman (kampir in Wakhi) who alone would offer food to Baba-ghundi and thus saved herself from the saint's wrath, which overwhelmed and buried the whole village under stones and detritus.8
A mile or so below Sipenj, which in spite of abundant arable land contains only seven families, a rock known as Roshtigar bears natural markings distinctly resembling padukc s. They are reverenced now as footprints of Baba-ghundi. We have evidently here a case of continuity of local worship reaching back to pre-Muhammadan times. The fort village of Réshit, some three miles lower down, with its dozen houses (Fig. 59), represents the chief place of Chapursan ; its quaint circumvallation recalls the time, not very remote, when the valleys of northernmost Hunza were still exposed to Kirghiz counter-raids from the north. From this point all the way down to Spandrinj, a distance of some eight miles, I noticed abundant traces of abandoned cultivation, including an old canal, along the fertile alluvial fans which line the right bank of the river.
From the site of Spandrinj, which is bordered by jungle and is now used as a grazing ground, we made our way on September 5th, across the Kermin pass, about 13,60o feet above sea-level, into the Derdi valley. Descending this we struck the well-known route leading up the main valley of Hunza at the old tower known as T6p-khana, some five miles above the village of Misgar. Here I found myself on ground that I well remembered visiting on my first approach in 1900 to
6 Cf. Serindia, i. pp. 50 sqq.
7 See ibid., i. p. S4, note 3 ; also G.J., 1922, February, pp. 128 sq.