Although only one of the four main aqueducts is open at the present time, the exact course which each follows can be seen from the dump-heaps marking the positions of the shafts. The longest of the tunnels is the most westerly, measuring 4·6 kilometres from the point of origin to its exit on the west side of the ruined fort, the four together having a total length of 14·3 kilometres. The actual length of horizontal excavation is, however, considerably in excess of this figure, as there are very numerous subsidiary collecting branches ramifying from the main tunnels. Moreover, the total length given is the figure obtained by measuring along the surface from shaft to shaft; whereas, one of the most striking things underground is the markedly irregular trend of the tunnel, which frequently takes a wide bend between two adjacent shafts. [Illustration: THE SUBTERRANEAN AQUEDUCTS OF UM EL DABADIB] The man-holes, or vertical shafts connecting the underground waterway with the surface, served a double purpose. Through them the excavated material was hoisted to the surface, and in addition they afforded a means of ventilation, without which it is doubtful whether the work could have been carried out. The excavation of the man-holes alone must have been a gigantic task, equal to, if not greater than, that of cutting out the horizontal tunnels. The cross-section of an average-sized shaft is, if anything, greater than that of the tunnel; and although in vertical dimension the man-holes near the mouth of the aqueduct are insignificant, the depth rapidly increases in the opposite direction, owing to the upward slope of the ground-surface towards the escarpment. As already mentioned, the particular shaft I descended had a depth of over 40 metres, and subsequent levelling, from the exit of the aqueduct to a point above its origin, showed the most northerly one to have a depth of not less than 53½ metres (175 feet). Along this tunnel, which has a total length of 2·9 kilometres, I counted exactly 150 shafts, so that their average distance apart is between 19 and 20 metres. That the tunnel was most carefully planned and excavated is evident from its very low and gradual slope, as is indicated by the depth of water flowing through and by the generally unrippled character of the stream. Judging by the height of the mouth of one of the man-holes near the point of origin of the tunnel, and by its depth to the stream in the tunnel below, there only appears to be a fall of 1 metre in about 2½ kilometres, or a slope of 1 in 2,670. It is possible that, owing to the elasticity of the line used, there may be a slight error in the measurement of the shaft, but the above figure may be taken as approximately correct.