CONSTRUCTION. 173 passing from the square substructure to the dome we know nothing for certain, but we may guess that the system employed by the Sassanids (see Fig. 54) was a survival from it. It is unlikely that timber centerings were used to sustain the vaults during construction. Timber was rare and bad in Chaldaea and men would have to learn to do without it. M. Choisy has shown as we have already mentioned that the Byzantine architects built cupolas of wide span without scaffolding of any kind, each circular course being maintained in place until it was complete by the mere adherence of the mortar. 1 M. Place, too, gives an account of how he saw a few Kurd women build an oven in the shape of a Saracenic dome, with soft clay and without any internal support. Their structure, at the raising of which his lively curiosity led him to assist, was composed of a number of rings, decreasing in diameter as they neared the summit. 2 The domes of crude brick which surmounted many of the Kurd houses were put together in the same fashion, and they were often of considerable size. When asked by M. Place as to how they had learnt to manage brick so skilfully, the oven-builders replied that it was " the custom of the country," and there is no apparent reason why that custom should not date back to a remote antiquity. The Assyrians had recourse to similar means when they built the domes of their great palaces. They too, perhaps, left a day for drying to each circular course, and re-wetted its upper surface when the moment arrived for placing the next. 3 From the existence of domes which he considers to be almost beyond question M. Place deduces that of semi-domes, one of which he assigns to the principal chamber of the harem in the palace at Khorsabad (Fig. 55). Feeling, perhaps, that this requires some justification, he finds it in a modern custom, which he thus describes: " In the towns of this part of the East, the inner court of the harem is, as a rule, terminated at one of its extremities by a vault entirely open at one side, in the form of a huge niche. It is, in fact, the half of a dome sliced in two from 1 Note sur la Construction des Vontes sans Cintrage, p. 12. 2 PLACE, Ninive, vol. i. pp. 266-267. 3 As M. CHOISY remarks (LArt de batir chez les Remains, p. 80), each horizontal course, being in the form of a ring, would have no tendency to collapse inwards, and a dome circular on plan would demand some means for keeping its shape true rather than a resisting skeleton.
