The two systems still exist side by side over all Western Asia. From Syria to Kurdistan and the Persian Gulf the hemispherical cupola upon a square substructure continually occurs. The timber roof is hardly less frequent; when the apartment in which it is used is of any considerable size it is carried upon two or three rows of wooden columns. These columns rest upon cubes of stone, and a tablet of the same material is often interposed between them and the beams they support. A sort of rustic order is thus constituted of which the shaft alone is of wood. We reproduce a sketch by Sir H. Layard in which this arrangement is shown. It is taken from a house inhabited by Yezidis,[219] in the district of Upper Mesopotamia called _Sinjar_ (Fig. 58). [Illustration: FIG. 58.--Interior of a Yezidi house; from Layard.] We are inclined to think that both systems were occasionally found in a single building. The tunnel vault and the joisted ceiling were equally well suited to the long galleries of Assyrian palaces. In one room, or suite of rooms, nothing but brick may have been used, while in others wood may have had the preference. Still more probably, one architect may have had a predilection for timber, while another may have preferred clay vaults. In either case the general arrangement, what we may call the spirit of the plan, would remain the same. When wooden roofs were used were they upheld by wooden uprights or by columns of any other material? Botta was at first inclined to say yes to this question, but he did not attempt to conceal that excavation had discovered little to support such an hypothesis.[220] Such pillars, were they of stone, would leave traces among the ruins in the shape of broken columns; were they of burnt bricks (and there could be no question of the crude material), those bricks would be found on the spot they occupied and would easily be recognized by their shape, which, as we have already shown, would have been specially adapted to the work they had to do.[221] The points of junction with the pavement would also be visible. If we contend that they were of wood, like those of the house figured above, we must admit that, at least in the more carefully built houses, such precautions as even the peasants of the Yezidis do not neglect must have been taken, and the timber columns raised upon stone bases which would protect them from the sometimes damp floors. Neither these bases nor any marks of their existence have been found in any of the ruins; and we are therefore led to the conclusion that to search for hypostyle halls in the Assyrian palaces, would be to follow the imagination rather than the reason.