passing by the populous commercial town of Zebak, and is consequently a pass to be reckoned with in spite of its altitude. It is, in short, the chief pass over the Hindu Kush directly connecting India with Badakshan; but a pass which is nearly as high as Mont Blanc affords no royal gateway through the mountains. Wood had sufficiently indicated the nature of the Kokcha valley between Jirm and Minjan. At the point where the mines occur it is about 200 yards wide. On both sides the mountains are "high and naked," and the river flows in a trough 70 feet below the bed of the valley. We know that it is not a practicable route. It is, however, much to be regretted that no modern explorer has touched the valley of Anjuman to the west of Minjan, which, whilst it is perhaps the main contributor to the waters of the Kokcha, also appears to have contained a recognised route in mediæval times. "If you wish not to go to destruction, avoid the narrow valley of Koran," is a native warning quoted by Wood, which seems to apply to the upper Kokcha. As a passable khafila route, Idrisi writes that from Andarab to Badakshan _towards the east_ is a four days' journey. Andarab (the ancient site) being fixed at the junction of the Kasan stream with the Andarab River, the only possible route eastwards would be to the head of the Andarab at Khawak, and thence over the Nawak Pass into the Anjuman valley. Nor can the Nawak (which is as well known a pass as the Khawak) have any _raison d'être_ unless it connects with that valley. There is, however, the possibility of a wrong inference from Idrisi's vague statement. "Badakshan" (which was represented by either Jirm or Faizabad) is actually east of Andarab, but to reach it by the obvious route of the lowlands, following the Kunduz River and ultimately striking eastwards, would involve starting from Andarab to the west of north. But just as the Mandal leading into the Minjan valley opens up no useful route in spite of being a well-known pass, so may the Nawak lead to nothing really practicable in Anjuman. This, indeed, is probably the case, but Anjuman remains to be explored.