joints, he invariably searched with success among the débris of the chamber for remains of the sculptured decorations which had been used to face the walls. Not the least interesting of these discoveries was that of the drains under the palace, portions of which were in very good preservation; and all were vaulted, so that there can be no doubt whatever that the Assyrians were acquainted with the use of the arch. This was further proved by the discovery by M. Place of the great arched gates of the city itself, with an archivolt of coloured enamelled bricks forming various patterns, with a semicircular arch springing from plain jambs. Extreme care was taken by the Assyrian builders in laying the pavements to ensure their being perfectly level: first a layer of kiln-burnt bricks was laid on the ordinary sun-dried bricks forming the terrace; then came a layer of fine sand, upon which the bricks or slabs of the pavement proper were laid, forming in many cases an elegant pattern (see Fig. 33). [Illustration: FIG. 33.--PAVEMENT FROM KHOYUNJIK.] Great difference of opinion exists as to the manner in which the various apartments of the palace were lighted. M. Place suggests that the rooms were all vaulted on the inside, and the spandrels filled in with earth afterwards to form perfectly flat roofs, and he gives a restoration of the building on such an arrangement; but if he is correct, it is impossible to see how any light at all can have penetrated into the interior of many of the apartments, and as these apartments are decorated with a profusion of paintings it is very difficult to believe that artificial light alone was used in them. M. Place thinks, however, that in some cylindrical terra-cotta vessels which he found he has hit upon a species of skylight which passed completely through the vault over the rooms, and thus admitted the light from above. This, however, can hardly be considered as settled yet. Mr. Fergusson, on the other hand, suggests that the thick main walls were carried to a height of about 18 or 19 ft., and that above this were two rows of dwarf columns, one on the inner and the other on the outer edge of the wall, these columns supporting a flat terrace roof, and the walls thus forming galleries all round the apartments. Then to cover the space occupied by the apartments themselves it is necessary to assume the existence of rows of columns, the capitals of which were at the same level as those of the dwarf columns on the walls. Where one apartment is surrounded on all sides by others, the roof over it may have been carried up to a higher level, forming a sort of clerestory. This theory no doubt accounts for many things which are very hard to explain otherwise, and derives very strong support from the analogy of Persepolis, where slender stone columns exist. Such columns of cedar wood would add enormously to the magnificence and grandeur of the building; and if, as seems likely,