Of cardinal importance to us is the point at which the conduit debouches, because near to that point we may hope to find the fountain-house ‘Nine-Spouts.’ The conduit ends in an arrangement which is somewhat surprising, and which will be best understood by reference to Fig. 38. To the extreme left, at a point near letter B, the conduit emerges. It here consists of a massive channel built of blocks of _poros_ stone, indicated by the thick black lines on the plan. At point a⁴ it ends in the Pnyx rock. But, and this is the odd thing, at a³, about eight metres before the channel ends, a pipe issues from the stone channel and running parallel to the Pnyx rock conducts the water to the main reservoir (Haupt-Bassin). A similar arrangement has been observed in the aqueduct at Samos. There, too, the conduit pipe leaves the rock channel before it ends. It is conjectured[272] that this was a plan intended to mislead an enemy who might desire to cut off the water-supply. The conduit actually debouches at a⁵ into the great reservoir from which the new fountain-house Nine-Spouts must have been fed. Here, at the reservoir, we find indications of three successive structures. First a structure of very early date, possibly of the time of Solon. Second that of Peisistratos. Third a late Roman structure. Of the two earlier structures no masonry remains, but the position and dimensions can roughly be made out by markings on the Pnyx rock, out of which the West side of the basin was hewn. The exact size of the original basin, which was smaller than the later one, cannot now be determined. In the time of Peisistratos it was enlarged and deepened; the floor of the basin was sunk nearly 1·50 metres deeper. The great basin of Peisistratos was lined with masonry, the blocks of which have now disappeared. In Roman days the place of the great basin of Peisistratos was taken by a quite small structure. This change must have taken place before the building of the late Roman villa which occupied the place where once the ‘Nine-Founts’ stood. When the villa was built the great reservoir had for some time been disused, and the water from the aqueduct, not being needed on the spot in any large quantity, was carried by pipes to the lower city to the North for the supply of the new Roman market-place. These alterations as to water-supply, it should be noted, are of the first importance in questions of topography, and change in the direction or the extension of an aqueduct is naturally the index of a shifting of population. [Illustration: FIG. 38.] The restoration by Professor Dörpfeld (Fig. 38) is, it must clearly be understood, to a large extent conjectural. It must be consulted strictly in conjunction with the plan in Fig. 35, where the actual remains of Greek date are clearly marked in solid black lines. So used it can be of great service in helping us mentally to reconstruct scattered fragments of masonry that would otherwise be unintelligible.