At a later period, apparently during Persian times, two pillars formed of two roughly hewn palm stems were set up in court 36 to support a roof constructed either half way or completely over the court. They stood on the brick pavement, which here as in the adjoining chambers is composed of tiles measuring 40 × 41 centimetres. The lower end of the pillars was encased in a socket of brickwork covered with plaster (Fig. 66). The interior of this base still retains the impressions of the palm stems, the upper portion of which was also plastered. Strabo describes this kind of pillar (xvi. 1, 5): “διὰ δὲ τὴν τῆς ὕλης σπάνιν ἐκ φοινικίνων ξύλων αἱ οἰκοδομαὶ συντελοῦνται καὶ δοκοῖς καὶ στύλοις· περὶ δὲ τοὺς στύλους στρέφοντες ἐκ τῆς καλάμης σχοινία περιτιθέασιν, εἶτ’ ἐπαλείφοντες χρώμασι καταγράφουσι, τὰς δὲ θύρας ἀσφάλτῳ.” Nothing now remains of the reed rope that was twisted round the palm stems, but it is fairly certain that the stems were plastered over. [Illustration: FIG. 66.—Bases of late columns in court 36, Southern Citadel. ] The rear wall of the group of chambers behind the throne-room is toothed in a peculiar fashion. Since the wall joins the building at an oblique angle the series of rooms must either have been oblique, or, if the architects insisted on making them rectangular, the inner face of the wall could not have been parallel with the outer face. The latter could only have been effected by inserting wedge-shaped portions in the single brick courses, which would have imperilled the cohesion of the bricks and would have been very clumsy workmanship. If, on the contrary, the wall were built in retreating steps, the inner chambers could be rectangular and the rows of bricks laid straight, thus ensuring good bonding of the wall. This very characteristic feature of the outside of the building completely dominated the whole of the secular Babylonian architecture of the later period (cf. Fig. 156). All the streets of the town excavated by us in Merkes show these walls faced with remarkable one-sided projections, a method which was still adhered to in the later Graeco-Parthian period, when so much building was done with broken brick, although it was not then justified by technical considerations. It must not, therefore, be regarded as a mere requirement of the workmen, but as a model arising from the technique of an early art, unusual but very characteristic.