From some considerable remains of the _Piscinæ_, the system can be clearly made out. The building consisted of four chambers, two beneath and two above. Supposing, for the sake of illustration, in the annexed diagram, the letters [Illustration: A B C D] represent the four chambers. The channel of the aqueduct, coming at a tolerably high level, enters the chamber A. Thence the water passed (possibly over a large waste-pipe) into the chamber beneath, B. Between B. and C. there were small holes communicating through the wall (possibly provided with fine grating). Through the roof of C. there was a hole, and the water passed upwards, of course finding the same level in D. at its exit, as in A. at its entrance; on leaving the _piscina_ the water was carried off into another _specus_. By the aid of sluice-gates the water could be transferred direct from chamber A. to chamber D. Access was obtained by an opening to the chambers beneath, and the mud was from time to time cleaned out. Curious details of the sluices, &c., have been found; but it is not easy to determine their age, as the aqueducts have undergone so many repeated alterations. As a typical example of the size of the _specus_, or channel of the Aqueducts, the Marcian may be taken, which measures in the opening five Roman feet in height, two and a-half in breadth, and the thickness of the wall on each of the two sides one foot. The roof of the Marcian being surmounted by the Tepulan and Julian, no great solidity was needed; but so far as can be ascertained, the roofs were generally of considerable thickness, to prevent, as much as possible, the heat of the sun spoiling the coolness of the water. The channels and general structure of the Anio Novus and Claudian were larger in every respect, those of the Tepulan and Julian somewhat smaller, as shewn by the accompanying diagrams. It will be seen that the period of the construction of the nine aqueducts which have been described, embraces some four centuries, as Frontinus was employed in surveying them towards the close of the first century after Christ, and the first aqueduct was commenced more than three centuries before the Christian era. If we omit the Alsietina, which was on the other side of the river, and which ought scarcely therefore to be taken into account, the order of level follows almost exactly the order of construction, the only exception being that of the Virgo. This increase of height of level as each new aqueduct was added, is pointed out by Frontinus in his eighteenth chapter. It will be perhaps convenient to have the several aqueducts already described represented in a tabular form, shewing the dates when they were made, and the order as to level, together with the distances which they traversed.