As regards structure one of the most significant indications of date is the curve of the arches. Ukhaiḍir belongs to the time of transition from the round or ovoid to the pointed arch. This transition must have been accomplished in Mesopotamia during the course of the eighth century. While the Sasanian vault is invariably round or elliptical (I attach no importance to the fortuitous appearance of the pointed vault in the substructure at Qaṣr-i-Shîrîn), the Sasanian arch is, so far as my knowledge goes, invariably round. The arches of Sarvistân are specifically stated to be round,[436] the arches of Firûzâbâd are also round, though where the arch is set back upon the jambs a tendency to give a curve to the angle lends to them the appearance of a horse-shoe.[437] All the arches of the Ctesiphon façade are round, and at Qaṣr-i-Shîrîn the builders knew no other form. It has been contended that the pointed arch is found in the upper gallery on the interior of the east wall at Ctesiphon, but Dr. Herzfeld has shown satisfactorily that the curve assumed by those arches was dictated by their peculiar construction.[438] The pointed arch, like the pointed vault, may have been used sporadically in the pre-Mohammadan era (it is found in the church of Qaṣr ibn Wardân, which must have been built about the year A.D. 564[439]); it was latent in Sasanian architecture; but it was not until the eighth century that it passed into familiar use. In the Umayyad buildings on the western side of the desert, it appears side by side with the round arch, and at Hammân al-Ṣarakh, Ṭûbah and Mshattâ it assumes exactly the same shape in which we have it at Ukhaiḍir, a slightly stilted, pointed ovoid which bears the hall-mark of its descent from the Sasanian elliptical vault. Similarly at Ukhaiḍir it has not yet ousted all other forms; there are examples in the palace of the true ovoid arch and even of the round arch. The builders of Sâmarrâ went a step further. Their arches have shaken off all connexion with the Sasanian ellipse and have taken on the curve which was to become typical from that time forward of the Mohammadan pointed arch.[440] Of the same character are the arches of the Baghdâd gate at Raqqah, which cannot be earlier than the reign of Manṣûr and may with greater probability be assigned to Hârûn al-Rashîd.[441] It would therefore appear to be certain from the evidence which we possess that in the first half of the ninth century, and possibly as early as the close of the eighth century, the pointed arch had come into systematic use in Mesopotamia, to the exclusion of all other forms, and if that be the case, Ukhaiḍir must belong to an earlier period, more closely approximating, as I would suggest, to the period which witnessed the same transition stage on the Syrian side of the desert, a stage which falls there into the first half of the eighth century.