THE ARCH. 231 If the full benefit of the natural cohesion between one brick and another was to be obtained, this method of laying them was absolutely necessary. Internally, the drain we have, been studying was four feet eight inches high from the floor to the crovn of the vault. Its width was three feet nine inches, and its general slope very slight. It may be followed for a total length of about 220 feet, after v/hich falls of earth have carried away the arch and the whole northern part of the esplanade, so that no trace of the mouth by which it opened on the plain can be traced. The other sewer described by M. Place may be more summarily dismissed. In spite of their drawings and minute descriptions, explorers have not yet succeeded in explaining the eccentricities of construction it presents. It has two channels, one above the other, which are similar neither in slope nor section. Moreover this double sewer is abruptly interrupted in the middle of the artificial mound through which it runs. Must W 7 e believe that it was never finished or used ? We shall not attempt to answer this question, but shall content ourselves with pointing to the similarities between this tunnel and the last described. The same large stone slabs upon a layer of bitumen, the same inclination of the body of the vault, the same bricks formed in different moulds according to their place in the vault, are found in each. Our Fig. 93 shows the two channels and their position one above the other. The pavement of the terrace, which consists of a double bed of large bricks, rests upon the extrados of the upper channel. This vault is semicircular ; it has three voussoirs on each side, which, with the key, make seven in each vertical course. But in consequence either of an error in measurement or of a mistake in calculating the shrinking of the bricks, there was a gap between the third voussoir on the right and the key. This gap was filled in by the insertion of a stone cut into the shape of a wedge. But for this fault which, however, had no appreciable effect upon its solidity the vault would be perfect. 1 The narrow triangular opening of the lower channel may be seen below it. The semicircular vault gradually and insensibly changes into an elliptical one. The side walls become lower, at each yard their height is diminished by the thickness of a brick, and finally they 1 The slope, the height, and the width of this channel are not the same through- out. In some places it is wide enough to allow two men to walk abreastjn it.
