From the point of view of the special study on which we are now busy, the inhabited and visible part of an Assyrian building is less interesting than those channels hidden in the substructures which acted as drains. These channels existed in all the palaces. Layard encountered them at Nimroud and Kouyundjik,[286] but it was at Khorsabad that they were found in the best condition and most carefully studied.[287] We shall make use chiefly of the observations of MM. Place and Thomas in our explanation of a curious system of sewers that does, perhaps, more honour to the Ninevite builder than any other part of his work. Every detail of their construction is full of interest,--the general arrangement, the choice of materials and the various methods of vaulting brought into play. In nearly all the rooms there is an opening in the middle of the pavement towards which the rest of the floor has a gentle slope. It is a round hole cut through the centre of a square stone set among the bricks and leading to a circular brick conduit. In the first specimen described by M. Place, this descending pipe is five feet four inches deep, and rather more than eleven inches in diameter. It leads into an almost horizontal conduit with a similar section and of the same materials. This latter channel is gently inclined through the whole of its length; it terminates in the main drain of which the cut on the next page gives a section in perspective (Fig. 92).[288] The floor of this sewer was formed of large limestone slabs overpassing the inside width of the channel by several inches. By this means the internal joints were reduced to a minimum, and a further precaution was taken by placing the slabs in a bath of asphalte, which was also used to coat the oblique channels and the foot of the vertical pipe. The low perpendicular walls upon which the vault was to be placed were built upon the outer edge of these wide slabs. They were of four-inch bricks, carefully laid.