The walls of Bostra must have constitued one of the most imposing features of the city. As I have said above, they were not laid out upon any symmetrical scheme. The Roman walls, of which this brief account is to treat, were probably built upon the lines of far more ancient structures, remains of which are perhaps to be seen in the southwest quarter of the ruins, west of the Theatre, where huge masses of rough bowlders lie as if they had once constituted parts of some work executed by the hand of man. At a few points where these great rocks lie one upon another, in a fashion almost too regular to have been placed by nature, one can not fail to be struck by the resemblance of this crude work to the massive, almost Cyclopean, walls to be seen at a number of sites in the Southern Hauran, like Kôm Ku'aiyid \ Kharab is-Sakhl and ir-Rukes . The walls of the Roman period have disappeared almost completely ; they were broken up, and their material was carried away, to build the Arabic Castle during the Middle Ages. Yet there are sufficient remains of them at three or four points to enable us to determine how they were built. Throughout the entire length of the west wall, and for a considerable distance along the western end of the north wall, there are remains which show that these sections of the city walls are standing to a height of three metres or more; but are so completely buried in debris from the upper parts of the structure which were broken up to be carried away that it is difficult to realize that the lover parts are as well preserved as they actually are. The wall itself, about 4 m. thick, consisted of two faces of quadrated and draughted masonry of unusually good quality, such facing as was used in many of the best government works in this part of Syria, with a core between them of rougher stonework and broken stone. When the higher courses of the draughted blocks and the quadrated blocks of the core were taken to build the Castle, the roughly hewn blocks and the broken stone were left behind, and these create a mass of disordered material which well nigh buries the better preserved portions of the wall, and in certain quarters of the circuit, where the blocks of stone were all removed, is the only remnant by which the walls are to be traced. These western and northwestern stretches of wall are the only parts that preserve even their lower courses, for the reason that they were the farthest removed from the Castle, and it follows that those sections of the wall which lay nearest to that huge structure are not preserved at all. In fact, it is impossible at present, without excavations, to say where the walls of the southeastern part of the circuit were. A small bit of the Roman wall is preserved in the structure of the walls of the Djami il-Mebrak, the mosque at the northeast angle of city. The lower courses of this wall still remain, though partly buried, in a section of considerable length to the west of the mosque ; but the wall preserves a much higher fragment which actually forms a part of the west wall of the mosque, and is set at right angles to the other section, showing that it was part of an angle tower. This bit of the great city wall may be assumed to have been spared because the mosque is earlier than the castle. The section of wall that is shown in black on the plan of Bosra, a short distance to the northwest of the Theatre, is not of the same type of wall building as the great Roman wall. It is poorer work, and appears to belong to the mediaeval period. There is a round tower in the wall on the west side of the city and a square tower in the east wall, both of which are shown in the plan of the city; but they are so deeply buried in debris that it was impossible to secure a photograph of them or of any part of the city walls. All my observations of them were made by excavating small holes in the debris which now covers them, at different points along their course.