The renewed excavations of the Warren’s Shaft in Jerusalem by R. Reich and E. Shukron brought about a new understanding of the system and its dating to Middle Bronze II (ESI 20:99*–100*). It has been suggested that the Gezer water system also belonged to this period and was hewn in a similar manner. A bottom outlet leading from bedrock to surface was found in the Jerusalem water system and the hypothesis that the Gezer water system had a similar lower outlet was raised.
 
For this purpose, a geophysical survey, carried out by M. Yazersky and A. Beck of the Geophysical Institute of Israel-Lod, was conducted in October 2003, aiming to locate possible underground anomalies in a non-invasive manner. The selected method involved continuous vertical electrical soundings along four long test lines on the southern slope of the tell. The survey detected one distinct anomaly where a presumed hewn opening was present, which constituted a basis for conducting an excavation (Geophysical Institute of Israel Report No. 03/347/278).
 
A long narrow trench (3 × 18 m, depth 1.85–4.20 m from the top of the slope down) was dug perpendicular to the slope. The excavation reached bedrock in all parts of the trench, which was carefully cleaned by hand. No signs of quarrying or other human activity were discovered throughout the excavation area, particularly at the southern end, at the bottom of the slope where the anomaly was recorded in the geophysical survey. The results of the excavation had therefore refuted the theory of a possible outlet from the water system in this region. It became clear that the geophysical survey was not at all suitable for the rock structure that was discovered in the excavation. The surface of the rock itself was quite rugged, with a small natural cavity that was actually located at the top of the slope.
 
No signs of building were uncovered the entire length of trench; hence, it turns out that all the examined area was located outside the boundary of the tell. The section along the trench was fairly uniform. A layer of loose brown earth (average thickness 1.0–1.5 m) was on top of bedrock, overlain with a layer of gravel (0.5 m), a layer of brown soil (1 m) and a layer of stones that included some rectangular masonry stones, randomly placed (0.5–0.7 m). This upper layer is perceived as rock fall at the bottom of the slope and it was probably the result of Macalister’s excavations at the top of the tell, whence he rolled the stones of the ancient walls he dismantled before backfilling his excavation with soil in his distinct method. The stones were covered with a thin layer of soil (0.2 m) that forms the surface. The few potsherds collected mostly from the rock fall and the soil below it dated to various periods (Middle Bronze II, Iron II, Persian (?), Hellenistic). It seems that these too had originated in Macalister’s excavation.