circumstances would have completely arrested the completion of the underta- king ; and the more so, as these valleys were neither all, nor the greatest across which th* 10 38 PRELIMINARY ESSAY. water had to be conducted. The resources of the architects here become conspicuous, in their substitution of leaden pipes, forming syphons, already described, which were laid at an expense comparatively trifling, to what must have been incurred by following the other and more usual method. In describing the passage of the valley of the Garon, the aqueduct arriving at the summit of the hill, was stated to deliver its water into a tank, or reservoir, placed in a square tower. This reservoir, fourteen feet long by four and a half feet wide, is seven feet high to the summit ol its arch ; the walls are four and a half feet high to the springing of the arch, and two feet three inches thick. The arch is pierced in the centre by an opening two feet square, which serves as a passage into the reservoir. The bottom is lined with a coat of cement, six inches thick, with a curve at the angles of concourse of the sides and bottom ; there were two ranges of iron rods about one third of an inch in diameter, to strengthen the walls, and probably also to serve as a kind of stair by which to descend into the reservoir. This syphon bridge is disposed in the same manner, and has the same proportions, as the other arcades, the width of the arches being 18 feet, and the height of their opening 36 feet ; but this part is somewhat different from the others, from its width, which is 24 feet, and by its piers apparently terminating at the impost or springing of the arches, forming an elevation of considerable elegance, and a covered passage under the bridge. The arcades pierced in the piers for this road or gallery, are four feet wide, and 21 feet high, their arches being formed of voussoirs of thin stone, alternating with great bricks ; the facing is of reticulated masonry, and being built with black and grey pieces, has much the appearance of a chess-board. However, these open ings having weakened too much the higher piers, the builders were under the necessity of strengthening some of them, by counter-forts of the same sort of masonry. Two arcades in the valley of Bannau fell in 1757, from this precaution not having been taken ; from these piers, after their fall, it was observed that the arches were divided into compartments in their depth, by voussoirs formed of a double row of great bricks. The quantity of water, according to Delorme, which was furnished by this t»queduct in twenty-four hours, was 1,323,000 cubic feet. This author, however, did not appear to have made any corrections for the diminution of velocity on account of friction, and other circumstances affecting the motion of fluids ; after these ,have been made to bear on his calculation, the quantity has been estimated as not being greater than about 172,800 cubic feet, or about 1,209,600 gallons. The receiving reservoir for the valley of Bannau is eighteen feet long, and 6 feet wide. It was pierced, according to Delorme, with twelve openings, for as many syphons, this PRELIMINARY ESSAY. 39 increase in the number of pipes being made on account of the valley being deeper than that of the Garon, and, as he thinks, as the pressure increases, while the depth is augmented, that the architects multiplied the number of syphons, to divide this force by diminishing their diameter in proportion — a circumstance which, if accurate, would clearly decide that the difference between the weight and the pressure of water, was unknown to the Roman architect. The rest of the syphon bridges are similar to that of the Garon. There is now no vestige of the emitting reservoir of St. Foi. The receiving reservoir of St. Irenaeus is also much dilapidated, and also that of Soncieu. Delorme conjectured that its side was pierced for a smaller number of syphons than that of Soncieu ; but these pipes were of a greater diameter than those of the others, as appears by the parts at present in existence ; and Delorme thinks that the emitting reservoir was like the receiving tanks, which are seen near the wall of the city of Lyons, and conducted the water by an aqueduct to the reservoir, now called the Maison Angclique, This reservoir was also furnished with the usual opening. The emitting reservoirs had an opening at a height of four and a half feet above the level of the pavement, to turn, if required, the flowing water to the bottom of the tower, and to facilitate