The excavation was conducted along sections of the ancient aqueduct that led from ‘En Shulamit. The aqueduct sections were exposed beneath a level of wadi pebbles (thickness c. 0.5 m) that had piled up as a result of the collapsing cliff. It seems that the location of the aqueduct on a steep cliff caused parts of the cliff to give way. Three segments of the aqueduct (A–C; Fig. 2) were excavated close to the spring and another aqueduct section (D), which led to an open pool, was documented east of the spring.
Section A. The side of the aqueduct in this part was also built of fieldstones; however, it was probably lined with plaster that did not survive. The bottom of the aqueduct was built of wadi pebbles. A stone (0.4 × 0.4 × 0.8 m) affixed with gray mortar was discovered inside this part of the aqueduct. It partially impeded the flow of water in the aqueduct and was probably placed on purpose to prevent the use of the aqueduct. It is possible that this part of the aqueduct was built in a phase that postdated the western part, which was lined with ceramic slabs. This section of the aqueduct (length 12.5 m; Fig. 3) was excavated c. 12 m from the spring. At the beginning of the section, the aqueduct turns at a 90° angle to the southeast and continues along a uniform elevation line. The aqueduct was built inside a foundation trench (depth 0.2 m), on top of a conglomerate rock consisting of wadi pebbles. The aqueduct is U-shaped (L108; overall width 0.8 m, inside width 0.25 m) in the western part of this section. The side of the aqueduct in this part was constructed from medium-sized fieldstones and preserved 0.3 m high; the bottom of the aqueduct was built of wadi pebbles. The aqueduct was lined on the inside with round ceramic slabs (0.2 × 0.4 m, depth 1.2 cm), affixed with grayish white mortar, which were arranged as roof-tiles, whereby one overlapped the other by 5 cm (Fig. 4). The discovered slabs were broken and it was probably done deliberately to fit them to the dimensions of the aqueduct. A layer of travertine was discovered on top of the slabs, indicating that water flowed in the aqueduct. The construction and size of the aqueduct are different in the eastern part of Section A (L111; overall width 1.2 m, inside width 0.5 m, height 0.5 m).
Section B. An elongated heap of different size fieldstones (L116; length 5 m, width 1.5 m, height 0.4 m), probably remains of the aqueduct, was exposed 1.2 m from Section A.
Section C. A wall (W1; preserved length 2.5 m, height 1 m; Fig. 5) was exposed 5 m from Section B. It seems to have been a retaining wall of the aqueduct, which was built as a result of a change in the direction of the aqueduct that caused increased water pressure in this spot. The wall, founded on a layer of wadi pebbles (thickness c. 0.2 m), was built of different size fieldstones bonded with grayish white mortar. A shallow rock-cutting (L114; length 1.2 m, width 0.4 m, height 0.1 m), exposed south of the wall, was part of the aqueduct’s foundation trench. Remains of the aqueduct’s foundation trench quarrying were evident for a distance of another 2 m east of W1; the rock-cutting in this section was preserved c. 0.7 m high. It seems that W1 continued to the east for another 6 m. Remains of the aqueduct’s side (C1; L119; length 1.5 m, width 0.5 m, height 0.8 m), which also included two large fieldstones covered with a thin layer of travertine, were discerned 6.5 m southeast of Section C. This section was not excavated.
Section D. Another section of the aqueduct (length 33 m, overall width 0.7 m, inside width 0.3 m), built of fieldstones and oriented north–south, was documented 50 m southeast of Section C. It terminated in an open rectangular pool (6×12 m) that was built of stone. The inlet supplying water to the pool was not found.
The difference in the elevation of the aqueduct’s bottom in Sections A–C is 1.81 m. The length of this segment is 18 m and therefore the aqueduct’s incline is 10%. It appears that the high inclination of the aqueduct in this section stems from its construction on the precipitous slope of Nahal David. The difference in elevations between the beginning of the aqueduct and the open pool at the end of Section D is 4 m. The length of the aqueduct is 160 m and therefore the overall incline of the aqueduct is 2.5%.
Several fragments of pottery vessels were discovered in the excavation, among them closed cooking pots with a thickened rim on the outside and a handle drawn from the rim to the shoulder (Fig. 6:1, 2), which are common to the first century BCE–the first century CE; a cooking pot lid with a straight rim (Fig. 6:4), dating to the Early Roman period; and a Fine Byzantine Ware jar that has an everted triangular rim and a handle extending from the rim to the shoulder (Fig. 6:3), which is prevalent from the middle of the sixth until the beginning of the eighth centuries CE. In addition, fragments of ceramic slabs (Fig. 6:5) that lined the inside of the aqueduct were discovered.
The ceramic artifacts in the excavation were discovered in the layer of soil that covered the aqueduct and had eroded from the upper part of the cliff. These finds indicate that activity at the site occurred from the Early Roman until the Byzantine periods. Insofar as the ceramic slab lining in the aqueduct is concerned—there is no other such example in the country. These ceramic slabs resemble roof tiles that were used for covering public buildings in the Byzantine period.
According to Y. Porath, the first aqueducts at ‘En Gedi were built in the Hasmonean period and continued in use until the Byzantine period (‘Atiqot 50: 1*–20*). Porath dated the plaster lining the pool at the end of Section D, based on the chronological division of plaster types, and he suggested that the aqueduct conveying water from ‘En Shulamit was built in the Hasmonean period. G. Hadas who excavated the pools along the aqueducts says that it is difficult to date the period of use of the irrigation systems because numerous repairs were made to them over the years. Nevertheless, he suggests that the irrigation systems were first used at the end of the Hellenistic period, but most were built in the fourth–sixth centuries CE. Hadas mentions pottery vessels that dated to the Byzantine period, including a fragment of a Gaza jar rim, which was discovered in the plaster lining the pool at the end of Section D (defined as Pool 5 by Hadas, 2003, pp. 60–61, 279–280, Fig. 29:10–13).
Based on the results of the excavation, it is assumed that the excavated water system was built in the Early Roman period and used until the Byzantine period. During this long stretch, changes were made to the system and it seems that lining the aqueduct with ceramic slabs was done in the last phase of its use.