The excavation (c. 260 sq m; Figs. 2, 3) was conducted in and near the northwestern water pool of the southern water supply system. This water system, situated c. 1.3 km northeast of ‘En Boqeq, includes an open aqueduct (L120; not excavated); the two water pools, which are 10 m apart (the southeastern water pool was probably damaged by the construction of Road 90); a water channel (L101), which connected the two pools; a second water channel (L108), connecting the aqueduct with Channel 101; and lime-plaster floors on both sides of Channel 101. The excavation was located near the western edge of Road 90, which was constructed over most of the southeastern pool. It seems that when Road 90 was constructed, Channel 101, the aqueduct and possibly the western wall of the northwestern pool were partly destroyed, since the aqueduct and the channel terminate at the same point.
 
Northwestern Pool. The pool is square (internal dimensions 11.65 × 12.00 m; external dimensions 13 × 13 m). Its walls (W109–W112; width 1.0–1.5 m, max. preserved height 3.5 m) were restored prior to the excavation; the original wall in the southern corner (W109) was preserved intact. The walls were constructed of large fieldstones of local limestone, probably collected nearby. They were coated with hydraulic plaster, which was partially preserved. The floor of the pool was laid on a leveled natural layer of chalk, which served as the foundation for the floor; the chalk prevented water seepage. The floor consists of two layers of pebbles. The lower layer (L118), laid on the natural chalk, consists of wadi pebbles bonded with mortar. The upper layer (L114) is a pavement made of pebbles bonded with soil; these pebbles were arranged in lines (length c. 0.5 m), producing a geometric pattern on the floor (Fig. 4). The upper pavement was covered with a layer of plaster and ash, which was covered by two additional layers of hydraulic plaster: the lower layer is dark due to a large amount of ash, and the upper one, which was only partially preserved, is grey (L121; thickness 5–15 cm) and contains non-diagnostic potsherds. Two stairs (L119; Fig. 5) built of ashlars and fieldstones, both of local limestone, remain in the southern corner of the pool, where Water Channel 101 runs into the pool. The stairs most likely extended to the top of the pool and protected the floor where the water descended.
It seems that the pool, and possibly its surroundings as well, were roofed. Outside the southern corner of the pool was a round patch of ash (L116; thickness c. 0.3 m), possibly the remains of a burnt wooden beam or pole. A thin layer of ash was visible throughout the pool; this may be attributed to the burning of a roof constructed of a perishable material, such as palm branches. 
 
Water channels. A section of Channel 101 (preserved length c. 6 m, width 0.12–0.15 m, depth 0.1 m; Fig. 6), which led into the northwestern pool from the southeast, apparently connecting it with the southeastern pool, was uncovered. The channel was dug into the soil; along its walls were preserved some stones that lined the walls. The interior of the channel was coated with a gray plaster (thickness 1–2 cm), which contained a large amount of ash.
Water channel 108 (width 0.12–0.15 m, depth 0.1–0.4 m) led from the aqueduct to Channel 101, reaching it from the southwest; its floor is uneven, sloping toward Channel 101. The channel’s walls (width 0.15–0.25 m) were built of local fieldstones. Its interior and upper sides were coated with grey plaster (thickness 1–2 cm), which contains a large amount of ash. The channel was partially covered with stone slabs (Fig. 7). As the northwestern pool has no direct connection to the aqueduct, it received water via channels 108 and 101.
 
Floors. Sections of lime-plaster floors were preserved on both sides of Channel 101: Large sections north of the channel (L100, L103) and a small section (L104) south of it. It is unclear if these sections used to be one floor, which was cut by the channel, or if they were originally laid on both sides of the channel.
 
The pottery found on the floor of the northwestern pool and on Floors 100/103 and 104 date mainly from the Early Roman period, while some date from the late Byzantine period. Because the excavation was partially executed with mechanical equipment, sherds from floors 100/103 and 104 may have been mixed. Furthermore, since the floors were exposed near the surface, modern debris was also recovered on them. The Early Roman pottery includes a cup (Fig. 8:1), storage jars (Fig. 8:2, 3), jugs (Fig. 8:4, 5) and a juglet (Fig. 8:6). Plaster accretions on one of the jugs (Fig. 8:5) indicate that the plaster was used as building material. The late Byzantine pottery includes a casserole (Fig. 8:7) and a cooking pot (Fig. 8:8). These may have been swept into the pool in later periods, after it ceased to be in use.
Mollusca remains were found in four soil samples (c. 1 × 1 m, thickness c. 0.2 m each) from floor 114 that were dry sieved with a 5 mm sieve. Additional mollusca remains were collected on floor 104 and in water channel 101. The assemblage includes freshwater mollusks (Mollusca, Gastropoda): thousands of specimens of Melanoides tuberculatus and more than a hundred specimens of Melanopsis sp.
 
The southern water supply system of ‘En Boqeq appears to have been constructed during the Early Roman period (the Herodian period) together with what Fisher and Shacham (2002) term ‘the officina’ located further east, between the fort and the Dead Sea shoreline, where irrigated terraced plots with trees and plants were grown for producing perfumed oils and medicines, as in ‘En Gedi (Fisher and Shacham 2002:403; Erickson-Gini 2007:48). The northeastern pool had a capacity of more than 400 cubic m, and together with the second, even larger pool, the entire water system had a capacity of 900–1000 cubic m.
The southern water supply system consists of at least two phases. The construction phase dates from the Herodian period. Based on repairs in the plaster layers in the pool and in channel 101, it is possible that the early phase continued until the second–third centuries CE. According to Fisher and Shacham (2002:403), the site may have been abandoned during this time and resettled only later, in the Byzantine period, when the castellum was built (Fisher, Gichon and Tal 2000:XXV). However, as the recent excavation showed, only parts of the southern water system were used in the Byzantine period and later. It seems that the northwestern pool and Channels 101 and 108 were not used in the later period.