The constructive details of these old water courses are as interesting as are their general design. At the mouth of each aqueduct there generally was constructed a reservoir in which to collect water from the springs or streams that supplied it, and in which impurities could settle before the clarified water was delivered into the channel. The water channel was usually formed either of stone or brick coated on the inside with cement to make it water-tight. It was arched over on top, and at certain intervals vent holes were provided through which access could be had to the channel to make repairs. When two or more channels were carried one above another, the vent holes of the lower ones were placed in the sides. When possible, aqueducts were carried in a direct line, but frequently they were given a tortuous course either to avoid boring through hills, where their construction would have entailed too great expense, or else to avoid very deep valleys or soft marshy ground. In every aqueduct, besides the principal reservoirs at its mouth and terminal, there were intermediate ones at certain distances along its course, in which any remaining sediment might be deposited. In addition to serving as sediment basins, these reservoirs made it more easy to superintend and keep in repair the different sections, and provided service reservoirs to furnish irrigation water for fields and gardens and water for stock. The principal reservoir was that in which the aqueduct terminated. This reservoir or castella, as it was called, far exceeded any of the others in grandeur of architecture, or in magnitude and solidity of construction. The ruins of a work of this kind that still exist on the Esquiline Hill at Rome, are about 200 feet long by 130 feet wide, and had a vaulted roof that rested on 48 immense pillars disposed to form rows so as to form 5 aisles and 75 arches. From the description of this interesting reservoir, the interior must have greatly resembled many of the covered slow-sand fillers recently constructed in this country, in which elliptical groined arches form the roof, which is carried on brick columns spaced as in the reservoirs at Rome, about 15 feet from center to center. Judging from the fact that not only the aqueducts but also the reservoirs were covered to exclude light, it seems reasonable to conclude that Roman engineers were aware that absence of light prevented or altogether checked the growth of algæ and other objectionable forms of water vegetation. Nowhere in the writings of the early historians is any mention made of trouble due to this cause, but as the water supply of Rome was obtained from both ground (spring) and surface sources, which in many cases were mixed together, the resultant mixture would have furnished the best possible soil for algæ, the ground water providing the necessary mineral food and the surface water furnishing the seed. It is quite probable, therefore, that the aqueducts and reservoirs were covered to