Although I had vowed at the time that nothing would induce me to again enter the place, I afterwards reflected that, not having followed the tunnel to its extreme limits, there might still be further information to be gained, and at the termination there might exist an inscription, the deciphering of which would yield the much-desired information as to when and by whom the work had been carried out. Moreover, I particularly wished to ascertain whether the bulk of the water came from one particular point or whether it represented the accumulated flow from the numerous small fissures which a tunnel of such length must necessarily traverse. I therefore took the first opportunity of revisiting the place, this time providing myself with proper gear to facilitate the descent, and with instruments to make a thorough survey of the whole network of underground waterways. The accompanying sketch-map, reduced from the detailed plan made during my last visit, shows that there are four main aqueducts, running nearly parallel, in a north and south direction, along the sides of three separate valleys. It is, of course, quite evident that the tunnels were run under the valleys rather than beneath the intervening ridges with the express object of keeping as near the surface as possible, so as to avoid unnecessary excavation in the sinking of the vertical shafts, and to reduce the labour involved in hoisting the excavated material to the surface. The engineers were at the same time careful to avoid the actual beds of the valleys, as there the soft nature of the ground would have necessitated a considerable amount of stone pitching, without which there would have been constant falls of the loose detritus, consisting of sand and pebbles with large blocks of limestone, forming the actual floors of the valleys. Alignments were chosen along the extreme margins, the mouths of the man-holes being commenced on the sloping sides of the valleys, a metre or two above their pebbly floors. The great majority were thus excavated throughout in solid rock, the latter being, as a rule, sufficiently firm to stand without timbering or masonry supports.