parapet of the Ukhaiḍir gangway. The upper story, with its plain wall and its row of niches is the same in both façades, and the upper battlements may safely be restored at Ukhaiḍir. At Ctesiphon the capitals and bases (if bases there were) of the columns and colonnettes were moulded in stucco and have disappeared. Bases seem to have been absent from the slender engaged columns on the outer walls of Firûzâbâd and Sarvistân, but at both places the state of the ruins renders an exact determination of such details difficult. The engaged columns seem to rest upon a low plinth. The decoration in those palaces is, however, far more nearly connected with oriental than with occidental tradition. We have not much information as to Sasanian capitals. The columns and double columns of the inner rooms at Sarvistân are covered by rectangular imposts,[338] and de Morgan gives a drawing of a stucco capital from Shirwân.[339] It is scarcely necessary to allude to the famous impost-capitals of Bîsutûn and Iṣfahân, which belong, in all probability, to the end of the sixth or the beginning of the seventh century. They show far greater skill in the handling of the rectangular impost than the capitals at Sarvistân, but whether they are a natural development out of the latter, or borrowed directly from Byzantine art, existing material does not enable us to decide.[340] The latter theory seems to be the more probable, and it is supported by the fact that the evolution of the Mesopotamian capital did not proceed upon the Bîsutûn-Iṣfahân lines. At Ukhaiḍir there is a reversion to the simple impost of Sarvistân, nor did the development there go beyond the elementary impost-capital of rooms 30 and 40. The capitals of the swelling columns on the north façade of the central court may have been more like those of Bîsutûn and Iṣfahân, but unfortunately they are completely ruined. At a later date, in the church of Mâr Ṭahmâsgerd at Kerkûk (eighth or ninth century), the scheme of the Sarvistân halls is repeated, but the pairs of columns are without capitals or bases, and the colonnettes of the niches in the spandrels are similarly treated (Plate 75, Fig. 1). I should be inclined to reconstruct all the columns and engaged columns at Ukhaiḍir and Sarvistân, and possibly at Ctesiphon also, without bases.