On the opposite sides of the ravine and in face of the village of Hawârat-el-Maqta, during one of my inspections, I came across the remains of two ancient canals, shown on the sketch map, Plate XXVI. Starting from the present edge of the ravine are two old canals, clearly distinguishable as such by the existing banks, which are of considerable height. In the angle between the two are the remains of an ancient town, and fragments of granite pillars. One of these fragments was part of the shaft of a large pillar of the clustered-stalk design. Both these canals, after a few hundred yards, lose themselves in broken ground sloping down and tailing into the main ravine. Probably the left canal was the first made, and, when it breached into the ravine on its left, the right canal was made to take its place, which in its turn also breached and found its way into the ravine. The take-off was then shifted farther up the Bahr Yûsuf to the position of the present head of the Bahr Sêlah. The dotted lines show the supposed continuations of the two old canals. To feed them, either the ravine must have been dammed below their present take-off from it, or else they must have been continued across the head of the ravine to the banks of the Bahr Yûsuf. The fact that the second diversion diverges from the old canal just where it leaves the ravine, suggests the former alternative, but more probably this was made the point of departure from the first channel so as to utilise the banks, which already existed, for crossing the ravine and avoid the necessity for making a new crossing. Joined on to the breached end of the left canal there exist some curious vestiges of irrigation works, which have failed. It appears that there was originally an earth dam A B joining the banks of one or other of these two old canals with a point in the direction of or across the ravine. In the line of this bank where the height was greatest, was a thick masonry wall, now known as “Hêt Rozma.” This wall is made of brick and rubble stone of a very inferior quality, built in mortar made of lime and mud; it is 90 metres in length, 5 metres thick, and 6·70 metres high. (The top of this wall is at R.L. 21·35.) The bare end of the wall is evidently the original masonry end, as it was built, no part of the wall having been carried away when the bank, which must have joined its outer end, disappeared. The bank, of which this wall formed the centre, evidently breached and scoured out a hole, marked by the pool C below the breach. This breach was repaired by adding an inclined wall D E to the Hêt Rozma, continued by an earth bank E F to the bank of the old left canal. Again another bank G F seems to have been formed above this, and to have breached. The violent action of the water is shown by the circular hollow H, which has been scooped out of the level ground _upstream_ of this breach. [Illustration: PLATE XXVI. _Stanford Geog. Estab. London_