This aqueduct is said by some to be only a branch from the Aqua Marcia from the _castellum_ or _piscina_ at the Porta S. Lorenzo, to the Thermæ of Diocletian, and to have been made by him. Considerable remains of the large _piscina_ for these Thermæ were found when the railway station was built, a plan and section of it was preserved by Visconti; it was not destroyed, but was built over and effectually hidden. The aqueduct was continued, A.D. 330, by Constantine to his Thermæ on the Quirinal, now in the Colonna gardens. Others say that the Algentiana comes from the Mount Algidus, near Tusculum. In the Campagna between Frascati and Rome, there are remains of large reservoirs or _castella aquarum_ of the third century, which do not appear to belong to any of the aqueducts hitherto described. They are usually said to be only reservoirs for the supply of the adjacent fields; but they are all on the highest points, and it is difficult to see how the water was conveyed to them. There appears to be a regular line of them, not indeed a straight line, (the nature of the country would not admit of that,) but still a line of them at comparatively short intervals, always within sight of one another. It seems probable that the Aqua Algentiana was brought in a subterranean channel or _specus_ from the great reservoir at Tusculum, which is on very high ground, and that the water was permitted to run into these reservoirs at frequent intervals, instead of being carried on an arcade across the country, as had been done in the case of the earlier aqueducts. It is beyond dispute that the ancient Romans were well acquainted with the principle of the syphon, and that they were quite aware that water will rise to its level after a very considerable distance: they may have therefore thought it expedient in this aqueduct to avail themselves of that principle. It is difficult to explain how water could be supplied to these numerous high reservoirs in any other manner. XVII. AQUA CRABRA, AND MARRANA, A.D. 1124.