In form the fort is a hollow square with round bastions at the angles, and except on the side facing towards Kebeisah, where the centre of the wall is occupied by a gate, there is also a round bastion midway between the angle towers (Fig. 65). All these bastions are much ruined and I may be wrong in representing them as if unequal size. Before the door there has been a vaulted porch, among the ruins of which lies a large block of stone which looks as if it had served as lintel to the outer door; I could see no moulding or inscription upon it (Fig. 66). The existing inner door is arched, the arch being set forward in a curious fashion. It opened into a vaulted entrance passage which communicated with an open court in the centre of the building. The court was surrounded by barrel-vaulted chambers, some of which showed traces of repair or reconstruction, though the old and the new work are now alike ruined.[74] All the vaults are set forward about three centimetres beyond the face of the wall (Fig. 67). Above the outset the first few courses of stones are laid horizontally, inclining slightly inwards, but where the curve of the vault makes it impossible to continue this method without the aid of centering beams, the stone is cut into narrow slabs which are set upright so as to form slices of the vault, and each slice has an inclination backwards, the first resting against the head wall and every succeeding slice resting against the one behind it. This is the well-known Mesopotamian system of vaulting without a centering, which is as old as the Assyrians.[75] It is best adapted to brick, but it can be carried out in stone when the span of the vault is not large, provided that the stones be cut thin, so as to resemble as nearly as possible brick tiles. On the south side, which is the best preserved, there are traces of an upper storey, or possibly of an upper gallery or _chemin de ronde_. A doorway led from it into a small chamber hollowed out of the thickness of the central bastion: I imagine that there was a similar outlook chamber in the other bastions, but in all these the upper part is ruined. I could find no inscriptions; the Arab tribe marks (awâsim) were scratched upon the plaster with which the inner side of the walls had been coated. I do not doubt that Khubbâz belongs to the Mohammadan period, nor that it is a relic of the great days of the khalifate when the shortest road from Baghdâd to Damascus was guarded by little companies of soldiers stationed at Khubbâz and ’Amej, and perhaps at other points. The plan is that of many of the Roman and Byzantine lime fortresses upon the Syrian side of the desert,[76] of the Mohammadan forts and fortified khâns scattered over Syria and Mesopotamia,[77] and of the modern Turkish guardhouse; the structural details are Mesopotamian, dictated by the conditions of the land.