Located approximately 80 km. to the south of Amman and 15 km to the northwest of the present-day town of el-Qatrana and the same distance to the northeast of the legionary fortress of el-Lejjun is the fort of Qas Bshir. Remarkably well preserved, this castellum stands as a poignant reminder of the long vanquished might of Rome on this, the southeastern fringe of the Empire.
The fort is located at an elevation of ca. 800 m, in the center of a shallow depression in a broad, undulating plain, cut by numerous shallow wadi (valley) beds, all draining to the west to Wadi Mujib. To the north, approximately 3 km distant, is a ridge of hills overlooking Wadi Su’aydah, a tributary of Wadi Mujib. Approximately 2 km to the east is a low ridge which overlooks the desert to the east. The region itself is best described as arid steppe land.
The siting of the castellum is just below the top of a low rise, on its western side with a southwesterly aspect . Bedrock is very shallow, covered by only a thin deposit of loess. From the summit of its towers the fort dominates the surrounding region, with excellent views in all directions except the south. All movement across the plain or across the hills to the east would have been monitored from here. Cavalry patrols, to which the terrain is ideally suited, would have added to the fort's surveillance potential. To the west, the fertile, inhabited region of the Moabite plateau is visible across Wadi Mujib. Two former Iron Age towers reused by Nabataeans and Romans, Qasr Abu el-Kharaq and Qasr el-Al, are within visual range.
Qasr Bshir is roughly square in plan, with sides of 56.30 m (SE),57.05 m (SW), 56.75 m (NW), and 55.45 m (NE). The corner towers, which are the dominant feature of the building, project from the curtain wall 3.05 m and are from 11 to 12 m square. The main gateway is in the center of the southwest wall and is flanked by towers which also project from the curtain and are 6 m in width. Access to these towers was through rooms opening from the interior courtyard, to either side of the gate passage. These provided access to the towers at the level of both ground and upper floors. Adjacent to the southwest tower, on its northern side, was a postern gate 0.95 m wide. This gate opens into a passageway 1.30 m wide with a barrel vault ceiling, leading to the interior courtyard.
The corner towers were described in detail by Brunnow and von Domaszewski. Their description is basically accurate, although minor corrections have been made to the ground plans of towers III and IV . The towers were of three stories, with three rooms on each story. The largest of these rooms occupied the outer corner and had internal measurements of ca. 5.50 x 4.60 m. The remaining two rooms measured ca. 3 x 3.30 m and 2.90 x 4.70 m. The ceilings of these rooms were carried by transverse arches which supported stone beams, the ends of which were set into the walls . The floor of the upper rooms would have been laid over these beams and probably was of stone slabs. A stairway on the inner corner of each tower provided access to these rooms, via doors opening off landings on each floor, and to the roof. The stairway was ca. 1.10m in width. The treads of the stairs are stone slabs set into the tower wall on the one hand and the central stair pillar on the other. At the tops of these towers was probably a flat platform used for surveillance and signaling. Only the uppermost rooms of the towers had windows, these being small and set high in the walls, although there were embrasures in the outer tower walls at second story level. The height of the southwest corner tower is 10.10 m.
The curtain wall is of megalithic construction in its lower courses, with the courses becoming gradually smaller towards the top. The wall tapers slightly. At its base it is 1.50 m thick. The height was ca. 6.50 m including the crenellated rampart which topped the wall. A walkway along this rampart was reached from doors opening from the corner towers.
The exterior masonry of both towers and curtain was of massive blocks in the lower courses, some of which were dressed flat while others were marginally dressed, bossed blocks. The upper courses were of smaller blocks chinked and set in mortar. There was a rubble and mortar core. On the interior the lower courses lacked the massive blocks. The interior face was plastered.
Ranged around the large central courtyard are a series of rooms, seven each along the northeast and southeast sides and six along the northwest and southwest. The lateral walls of these rooms are bonded to the inner face of the curtain wall in only a superficial manner.
The range of rooms was two stories high, with the exception of the central room on the northeast side, opposite the gateway, which may have served a special function and which is further discussed below. The roof of these rooms, at the same level as the rampart walk, would have served as a fighting platform for the garrison.
Of the ground floor rooms, a total of 23 may have served as stables. These are interpreted as such because of the cupboard-like installations set into the inner face of the curtain, three to a room, identified as mangers, two of which were excavated .The existence of these installations was first discussed by Briinnow and von Domaszewski, who described them as niches, three to a room, formed by the construction of the lower part of the inner face of the curtain as a series of stone pillars, spanned by lintel slabs. The excavated examples measured ca. I m wide, 0.60 m deep, and 1.20 m high. The bottoms of these niches were 0.70 m above the floor of the room. Brunnow and von Domaszewski hypothesized that these were a type of vault, designed initially to reinforce the wall but subsequently not filled in.
They do not speculate upon their ultimate function. There is no firm evidence of their purpose. Their bottoms are flat, lacking the basin or trough that one might expect in a manger; however, there may have been a wooden feed rack or some other device, now vanished. This type of installation is paralleled at Umm el-Jimal, where they have been interpreted similarly.
If the fort was garrisoned by a unit of cavalry (possibly ala II Miliarensis, located at Naarsafari—Wadi Afaris—near Wadi Mujib), as seems most likely, then the interpretation of these rooms as stables and the installations as mangers is probably correct. There are a total of 69 mangers in 23 rooms, providing accommodation for only 69 horses, if we postulate only three to a room. It is unlikely that more than three horses could have been stabled in these rooms, several of which were only 3 m in width, the largest being a little over 4 m. Cavalry units of limitanei in the Late Empire may have numbered ca. 120 to 150 men. Accommodation is provided here for the mounts for roughly half of such a unit. At any one time it is possible that the other half would have been mounting patrols within the frontier region or beyond, or perhaps detachments of the unit also garrisoned nearby watchtowers. Mounts could also have
been stabled in the central courtyard or corralled outside the fort.
The upper range of rooms probably served as barracks for the troops, who may have slept directly above their mounts. Brunnow and von Domaszewski speculated that there was access between the ground floor and upper rooms by means of ladders, although no proof of this has been found. Otherwise access to the upper rooms was via a walkway, which probably extended around the courtyard at this level. Evidence for such a walkway is found in the southwest corner.
Located literally upon the front line of the south eastern frontier of the Empire, Qasr Bshir was a key link in the chain of watchtowers and forts. These were strategically located along the desert fringe in this region, forming the first line of defense in depth of the province of Arabia. To the east lay the desert, the domain of the nomadic Arabs whose political allegiances were often uncertain.
Qasr Bshir is a castellum, constructed as a quadriburgium, with projecting towers at all four corners and flanking the gate . Its large interior courtyard is surrounded on all four sides by barrack rooms constructed against the inner face of the curtain wall .
This style appears to be typical of the end of the third to fourth centuries on the eastern frontier and elsewhere. In the province of Arabia it is attested at many sites which are more or less contemporary with Qar Bshir. These include Deir el-Kahf, which likewise is of Tetrarchic date (306 AD), Qasr el-Hallabat, Khirbet ez-Zona, Qasr eth-Thuraiya, and Muhattet el-Haj (upper and lower forts), all of which follow basically the same plan, although varying in size and in the use of interval towers. Two other similar sites in the south, Khirbet el-Qirana and el-Ouweira, may also belong to this group, although dating is less certain. Of these sites, Qasr Bshir and Deir el-Kahf are securely dated by inscriptions to 293-306.
Lander synthetic treatment of Late Roman fortifications' is a useful collation of the dating evidence for known forts with square towers and plans similar to those of Qasr Bshir.
In Syria Han at-Trab, Han as-Samat, Mleke, and Han Aneybe are all similar in plan but differ from most examples in Jordan (except Khirbet ez-Zona) in that the corner towers all project fully. These forts are sited on, or close to, the Strata Diocletiana and are probably Diocletianic in date. In North Africa the fort of Aqua Viva is dated by inscription to 303. It is larger than Bshir (86 m square) and possesses interval towers but is otherwise very similar in design. A smaller fort of almost identical plan is Aquae Herculis, for which dating evidence is lacking. In Europe parallels are less common and perhaps of later date.