The four aqueducts, Ashby and Miss Van Deman found, followed the course of the Anio fairly closely from their source to just below Tivoli, where, having lost half their altitude, they turned south along the shoulder of the hills to Gallicano. In this stretch, at Ponte Lupo, the Aqua Marcia crosses a gorge on a bridge that would test the mettle of the most seasoned archaeologist, for it epitomizes Roman constructional history in stone and concrete for almost nine centuries. After Gallicano the intrepid pair traced the aqueducts’ course westward, where, by a system of tunnels, inverted siphons (the Romans knew that water would rise to its own level), and side-hill channels they cross the broken gorges of the Campagna to a point south of Capannelle racetrack, six miles from Rome, whence they proceed on the famous arches to the Porta Maggiore. From reservoirs in the city the water was distributed in lead pipes (one, of Hadrianic date, has walls three inches thick, and weighs eighty-eight pounds per running foot), with a strict priority, first to public basins and fountains (the Aqua Julia alone supplied 1200 of these), next to baths (extensions of the Marcia supplied those of both Caracalla and Diocletian), then to private houses. Surplus was used for flushing the sewers. Attempts were made to control the priorities by running the pipes for private use only from the highest levels of the reservoirs, but Frontinus complains bitterly of illegal tapping. In the Gallicano-Capannelle stretch special archaeological ingenuity is required, first to find the channels, and then to decide which belongs to which aqueduct. Where the channels have entirely disappeared, through the disintegrating action of floods, earthquake, tree roots, or plowing, the course can be defined by plotting the occurrence of heaps of calcium carbonate on the ground. This is the aqueduct deposit. Roman water is extremely hard, and the heaps mark where once there were downshafts (_putei_) for inspection and cleaning the channels, which without such maintenance would soon have become completely blocked with deposit. Frontinus says the downshafts occurred regularly every 240 feet, and Dr. Ashby found many at just this interval.