Area 2000
Two separate trenches were opened in this area (Fig. 2). The first (L2072; 3.5 × 5.5 m) was within the southeast corner tower of the fort, whose walls were mostly intact, except on the east. Above the original Roman floor in the tower, a substantial burnt patch that contained twenty iron nails was discovered on the west (L2076; 76.82 m). The nails came perhaps from a ladder that led to the roof. Below the burnt patch was a fill (L2077; 76.62 m) of black ash and soot, presumably representing the remains of fallen timbers from the floor of the upper story, as indicated by large chunks of charred wood. When the excavation ended we appeared to be just above the original Roman floor, as suggested by the discovery of a coin (76.20 m).
The second trench (L2071; 2.5 × 6.0 m) lay outside the fort, at the junction of W510 and W514. This trench exposed the exterior faces of the curtain wall (W510) and the northern wall of the southeast corner tower (W514), both of which were constructed from irregular fieldstones and coated with plaster. The two lowest courses of W510 comprised a foundation of large fieldstones, projecting slightly from the wall above. Wall 514 was a much thinner tower wall that did not have a foundation. No trace of an external defensive ditch was discerned.

 

Area 5000
Work resumed in the entrance corridor to the main (east) gate of the fort and in the room immediately to its north (L5022; Fig. 3). Previous seasons had revealed a large semi-circular niche (L5015) in the east wall of the fort (W517) with a built stone platform in front of it (L5025). Two benches (L5044, L5045), abutting W523 on the north and W525 on the south, flanked the platform and extended into the balk under the Mandatory police station. Four successive plastered floors were distinguished in front of the platform, the lowest of which (L5024; 76.1 m) was excavated this season. Finds associated with this floor included animal bones, pottery, glass fragments, a coin and a possible ballista shot.
Four successive Roman floors were also discovered in the entrance corridor. A partition wall (W532) associated with the third of these floors (L5029) separated the entrance corridor from areas to the south. A stone bench (L5028) lined the entrance corridor, abutting W532.
A thickly plastered surface (L5026; 76.35 m), corresponding to the uppermost (fourth) Roman floor in the entrance corridor, was exposed in the area enclosed by the projecting piers of the entrance arch. This surface covered the threshold stones of the gate, indicating that by this time the threshold was no longer in use. Three earlier Roman floors abutted the threshold stones, which came to light at 76.29 m (L5030). A large camel bone had been buried under the uppermost of these three floors (L5032), just outside the threshold.
The threshold was constructed from thickly mortared cut stones, consisting of a rectangular block in the center and two carefully worked stones on either side that served as doorjambs and projected inward (west). The doorjambs were worked so that their bases, indicated by deeply cut circular hollows, were several centimeters lower than the elevation of the threshold. This, together with the evidence of semicircular grooves on both stones, indicates that the original fort gate comprised an inward-swinging double-leafed door. As the two uppermost Roman floors covered the doorjambs, this double-leafed gate presumably had gone out of use by the later phases.
Twenty-four iron nails associated with burnt patches, probably from the double-leafed doors, were found lying on the earliest Roman floor inside the gate (L5034; 76.07 m).  An iron object that might be the gate’s locking mechanism was also found lying on this floor. In addition, the walls just inside the doorjambs of the gate were reddened by intense heat and burning. This is additional evidence for the conflagration that marked the end of this phase, as discovered elsewhere in the fort. A drain (L5036 and L5037) ran under the first (earliest) Roman floor (L5034) in the entrance corridor and out of the gate.
The western balk of Area 5000 preserved evidence of seismic activity after the fort’s abandonment. A mud-brick collapse (width 3 courses, height 13 courses) appears to have toppled southward from W523 to rest on top of the uppermost Roman floor (L5018) at its furthest point. The fact that these bricks fell en bloc suggests a single, catastrophic event.

Area 7000
Work continued in the southwest corner of the fort. A new trench (L7040; 2 × 4 m) was opened north of the Roman room (L7036), west of the staircase by the southwest corner tower (Fig. 4). An entrance with two steps led from this room to a corridor on the north side of W533.
Excavations in the postern gate (L7060) revealed that the staircase consisted of seventeen large stone steps, seven of which were still in situ (Fig. 5). Two sets of stairs were at either end of a central mud-brick platform (W534). Both sets were accessible from the postern gate and turned as they ascended to the central platform.
The postern gate was filled with collapse (Fig. 6). Its frame was built of three courses of fine ashlar stones, each cut in an L-shape to form the doorjambs. The stones were carefully cut and fitted into place, without any visible mortar. The doorway was 0.8 m wide and c. 1.8 m high. Square slots were cut into the east and west walls (W519, W550) of the postern gate, just inside the entrance and c. 0.5 m below the top of the doorway. The slots presumably held a bar to lock the gate. On the west side the slot extended through W550 into the southwest corner tower, so that the gate could be locked from inside the tower. Two horizontal grooves, c. 0.25 m below the slots, seem to have been formed by the repeated blocking of the gate with a wooden panel, which was locked in place by the bar.
The discovery of a thickly plastered Early Islamic threshold (L7082; 77.29 m) indicates that the gate still functioned in the last major occupation phase of the fort. Beneath it, the original Roman threshold was composed of one large stone between the doorjambs (L7090; 76.92 m). At least one floor surface was identified on the inside of this threshold (L7085; c. 76.8 m). Finds associated with Floor L7085 included pottery, glass, and animal bones, as well as a number of iron fragments representing at least four nails and other door fittings, clustered just inside the gate and on its east side. There may have been earlier Roman floors associated with the gate, but the end of the excavation precluded further exploration.
A new trench (L7041;3 × 6 m), opened in the southwest corner tower, was accessed from the postern corridor. Excavation in this area was limited to the tower’s entrance (width 0.94 m) and the inner face of its north wall (W551). An Early Islamic floor (L7066; 77.62 m) and four Roman floors were exposed in the tower. At the level of the lowest Roman floor (L7084; 76.70–76.65 m), three crudely cut stone blocks had been laid as a threshold at the entrance to the tower. At this point the excavation ended.

 

Area 9000
The area to the east of W536 was a room with a series of Roman floors (Fig. 7). At the beginning of this season, excavations revealed a compact dirt surface that apparently represented the second to lowest Roman floor. The material associated with this surface (L9027 and L9030; c. 77 m) included coins, animal bones, pottery, glass, and a spindle whorl. The room contained several built features, the most prominent of which was a large plastered hearth (L9026), enclosed by mud bricks that abutted W539. It was filled with charcoal, ash, burned bones, a date pit, pottery, glass, iron fragments and three coins. The lowest original floor (L9035, L9037; 76.92 m) included a plastered flagstone pavement on the west side of the room that yielded a large number of iron objects, three coins and a gaming piece. Soundings below the earliest floor revealed the gravel leveling bedding, deposited when the fort was originally constructed.
Excavations north of W536 indicated that all but the lowest course of W539 had been removed after the initial construction to create a continuous corridor along the inside of W518 (the west curtain of the fort). 
The trench (4.5 × 4.5 m) was extended northward to locate the staircase at the entrance to the northeast corner tower. An Early Islamic surface (L9043) was found at 78.40 m. The topmost step of the corner tower staircase was uncovered in situ, immediately west of W553. At the same elevation, the top of a mud-brick platform (L9045) appeared between W552 and W518, analogous to the platforms associated with staircases in other areas (L2019 in HA-ESI 117). A second step was found immediately to the north of and below the first step (elevation of top step c. 78.19 m; second step, c. 78.03 m).
Another extension (1.25 × 3.50 m), north of the curtain wall (W558) and east of W553, revealed the top of an Early Islamic tabun (L9047; diam. c. 1 m). Its walls, built of concentric rings of tiles, were covered with plaster; the opening faced west. At this point the excavation ended.

Area 1000 – the Bathhouse
Excavation continued in the frigidarium and apodyterium, enclosed on the north by W545, on the east by W554 and W556 and on the west by W546 (total dimensions 8.5 × 9.0 m; Fig. 8). The bath complex was entered on the east through a courtyard paved with finely cut stones (L1032). A low stylobate (W556) of finely cut stones and covered with plaster, marked the boundary between the courtyard and the bathhouse to the west. Six large stone piers, arranged in two rows of three, supported the roof of the frigidarium (L1008) and apodyterium (L1033). The space between the piers formed a passage between the two rooms and presumably served as the entrance to the complex from the courtyard. The walls and floor of the frigidarium were coated with thick hydraulic plaster, whereas the plaster in the apodyterium was thinner and non-hydraulic.
A plunge bath (L1031; depth over 1 m; Fig. 9), enclosed by W554, W557 and W559, was exposed in an extension to the northeast of the frigidarium. Four steps descended to the bottom of the plunge bath along its western side. A crude semicircular mortar and stone wall (W555) was built over the plunge bath after it went out of use. This secondary wall is presumably associated with Bedouin occupation, which was evidenced throughout the bath complex.