One of the structural features of Ukhaiḍir has a value which is not only structural but also decorative. I allude to the use of masonry tubes between parallel barrel vaults. Obviously it is a scheme which was born of the systematic use of the vault. It is to be found at Hatra, where it appears in some of the tombs.[378] The same system is present at Firûzâbâd, where there was a masonry tube between the barrel vaults of the side chambers of the entrance lîwân and the domed chamber.[379] In later Mohammadan architecture I have found masonry tubes at Khân al-Khernîna above Tekrît.[380] A second device for the lightening of the wall mass between parallel barrel vaults is employed at Ukhaiḍir in the east annex and in the buildings to the north of the palace. It takes the form of a number of narrow tubes. I saw it also in a fourteenth-century khân at the foot of the Djebel Sindjâr (Plate 88, Fig. 1), a khân which is famous for the dragon reliefs on its doorway,[381] and in a mosque of the early fifteenth century at Ḥasan Kaif (Plate 88, Fig. 2). The decorative importance of the first scheme, the large single tube, lies in the effect which its opening produces on the façade. This can be observed in the courts on the ground floor at Ukhaiḍir, as well as in the court on the upper story of the gate-house. The arched openings of the tubes between the arched doors of the lîwân and its side chambers form an essential part of the façade, and they are retained when vault and tube are alike absent. The existence of tube openings in the façades round the central court, the ṣaḥn, of the mosque of Ibn Ṭulûn is sufficient to show that the Egyptian mosque was copied from a vaulted prototype (Plate 89, Fig. 1). I do not doubt that it was modelled on the vaulted buildings of Mesopotamia, though vault and tube are absent from its structure. The great mosque at Sâmarrâ was not vaulted; unfortunately the data are insufficient to determine the scheme of the façades of its ṣaḥn. Nor was the mosque of Abû Dulaf vaulted; it had a flat roof carried on arches, like Ibn Ṭulûn; but the tube openings appear in the form of niches on the façades of the ṣaḥn (Plate 89, Fig. 2). As at Ibn Ṭulûn, they have become purely decorative. I do not know whether there are tubes between the vaults of the Bait al-Khalîfah at Sâmarrâ, but the openings are simulated upon the façade by shallow blind niches. The same system holds good in the ṣaḥn façades of the Azhar at Cairo, a building which has no other connexion with Mesopotamian architecture than this traditional use of a decorative motive, the true significance of which had long been forgotten. CHAPTER VI THE MOSQUE