jambs of the south gate are standing, the width of the opening is only two metres, and the lintel here and in the east gate (where it has fallen to the ground) is unadorned and uninscribed. The character of the masonry and the existence (as is proved by the lines of street and ruin heap) of a town carefully planned upon an ordered system, point to a date prior to the Mohammadan conquest, and I am inclined to seek for a Byzantine origin for Za’ferân. Perhaps it may be a relic of the triumphant, though brief, re-occupation by Heraclius of the provinces ceded to the Persians by Jovian. I followed my caravan back to the Môṣul highway and so across the hills to Zâkhô. We climbed up the pass by as good a road as any in Turkey, but while we were rejoicing over its excellence, it broke off short and left us to find our way down the opposite side of the pass as best we might along a bridle-path strewn with boulders. So we came down into the valley of the Khâbûr and saw before us the snowy wall of the Kurdish Alps (Fig. 180). At the gate of the pass stands Zâkhô, “old and isolated,” as Ainsworth says, and it would be difficult to better the phrase.[173] The more ancient part of the village is built upon an island in the Khâbûr. The right arm of the river is spanned by a masonry bridge, the left arm washes round the castle, a fortress which must have had a long and checkered history, though I can find no record of it.[174] The masonry is of many different periods. The finest and probably the oldest part is an octagonal tower which juts out into the stream on the south-east side. The outer walls are all fairly well preserved and make an imposing appearance, but the interior is terribly ruinous. In the upper part of the building there is a large hall with windows opening on to the river. The engaged columns which support the interior pointed arches of these windows are covered with a delicate tracery of carving very like Seljuk work of the thirteenth and fourteenth centuries. This part of the castle cannot be dated later than the fourteenth century, but the foundations and the octagonal tower must be considerably older. Last of all the Turkish garrison has supplemented the ancient work with wretched structures of rubble and mortar, and these, too, have fallen into ruin and have been given over to the storks, who nest contentedly among them. In Zâkhô lies buried the first missionary to Kurdistân, the Dominican Soldini, who died here in 1779. The quarter that stands upon the right bank of the Khâbûr is mainly Christian and contains, I believe, two small churches of no very great age, but my curiosity was quenched before I reached them, by a violent thunderstorm which drove me back to my tents. It swept down the valley from Amadîyeh, and rolling away, left the mountains so magically beautiful that I could give no further thought to any architecture but that of their white pinnacles and spires. CHAPTER VIII ZÂKHÔ TO DIYÂRBEKR _May 10--June 4_