In 2005–2006, salvage excavations were conducted at the springhouse in ‘En Zippori (Permit Nos. A-4632, A-4828; map ref. 123740–88/175760–176200; Fig. 1), before preparing the site for visitors. The excavations, undertaken on behalf of the Israel Antiquities Authority and funded by the Kishon River Authority, were directed by J. Sharvit, R. Steinmetz, B. Gatenio and Y. Makovsky (electromagnetic surveying), with the assistance of V. Essman, A. Hajian, V. Pirsky (surveying and drafting), Y. Shaltiel (conservation and planning), O. Marder, A. Yaroshevich and D. Planer (flint tools) and S. Ben Yehuda (graphics).
The first excavation at the ‘En Zippori site was undertaken during the British Mandate by Nazareth’s municipal engineer, who uncovered the top of a massive wall (a dam?; length c. 5 m, height c. 1 m) built of large ashlars; 1.5 m to the west of the wall was an aqueduct (width 0.42 m) that led to the northwest (Archives of the British Mandate Department of Antiquities; Fig. 1:1). Prior excavations also revealed a covered water channel that was part of the irrigation system of nearby agricultural plots and yielded Roman, Byzantine and Early Islamic pottery (Ben-Nachum 1998; Fig. 1:2), and remains dating from the Middle Bronze III to the Ottoman period, including a circular structure (diam. 6 m) with a wide wall (thickness c. 2 m) called the ‘springhouse’ (Porat 2005; Fig. 1:3).
In 2005, work began on preparing the springhouse for visitors, but it was halted because of the high water level in the structure. Subsequent water-jet probes inside the springhouse encountered a solid layer 2.5 m beneath the surface of the water. As a result, six excavation areas (Fig. 2) were opened in the springhouse (Area B) and around it (Areas A, C – F). A geophysical survey was also conducted, probe trenches were dug in the vicinity of the springhouse (Areas Ars–Frs; Fig. 3) and an underwater examination was carried out in the British Mandate pumping station (see Fig. 1).
Area A. Two excavation squares were opened (A1, A2). Square A1 (Fig. 4), excavated to the south of the pool adjacent to the springhouse, revealed a layer of clayey soil (L138) mixed with Neolithic flint items, Roman and Ottoman potsherds, and modern finds. At a depth of 0.11 m beneath the soil layer lay the top of a wall (L144) built of two rows of roughly dressed stones with a core of small stones. In the eastern half of the excavation square, a floor of small stones (L152) was uncovered that was dated to the Neolithic period (Wadi Rabah culture) by abundant finds of flint and pottery vessels and several stone tools. The western side of the excavation square contained a layer of soil mixed with stone fragments and small pieces of stone rubble (L157) with waterlogged mud beneath it.
In Sq A2 (Fig. 5), a layer of black soil was uncovered that was rich in stone flakes, flintstones and Roman pottery vessels (L139). At a depth of 0.23 m beneath this layer were the remains of a wall (W150), to the south of which was a floor made of crushed chalk, pebbles and small stone chips (L146); the floor extended along the wall and abutted it. To the east of the floor and beneath it was black soil rich in stones and stone flakes (L147) on top of a layer of gray soil (L151) that contained several iron nails. The foundation of W150 was clearly visible in L151. When the balk between the two excavation squares was removed, the channel’s covering slabs were found immersed in water (L158; Fig. 6).
Area B. An underwater examination was conducted in the shaft of the springhouse’s inspection chamber (1.8 × 2.3 m; Fig. 7) in order to check the depth of the pool. Fifteen water-jet probes (L137) were conducted designed to only penetrate unconsolidated sediments. Debris of large building stones (L140) that had collapsed from the walls and ceiling of the pool was uncovered and included two dressed stones from the mouth of a well with grooves made by ropes used to draw water. Unconsolidated sediments were also extracted with a dredger operated by divers, revealing a concrete floor (L162; thickness c. 0.15 m) from the British Mandate era that sealed about half of the area of the pool on its northern side. Wooden beams and iron bars were incorporated into the floor. A fill of small fieldstones (L163; thickness 1.55 m) was uncovered below the floor. At the bottom of the fill (L163a), a floor of tamped crushed chalk and small stones covered the entire base of the pool; this was the floor of the original pool built in the Roman period. A section dug in the southern corner of the shaft (L168) beneath the floor uncovered the foundation trench and foundation courses of the pool. On the southern side of the shaft, the inlet of a feeder channel (Fig. 7C) was uncovered through which a constant supply of clear water flowed from the southern channel. On the northern side of the shaft was an outlet (width 0.5 m) through which water flowed out into the northern channel (Fig. 7B), some of whose cover stones had been removed in the past. The two conduits were c. 9 cm higher than the floor of the pool and they were not positioned directly opposite each other but were slightly offset to ensure that the water flowed in a circular motion. Since the water did not flow out in a direct line, it generated movement around the pool and the sediments sank to the bottom. The conduits were clearly visible in the channel walls, which were built of five courses of dressed stones with no traces of cement or plaster. Measured from the upper course, the channels were c. 1.2 m deep.
Areas C and D. The topsoil had been disturbed by the installation of water pipes (L141). A probe (L148) was dug, in which flint tools and a few potsherds from the Pottery Neolithic period were found in situ.
In Area E, the northern channel was excavated. After removing surface finds of flint items that were not found in situ and removing the channel’s cover stones, subsurface testing determined that the channel was blocked up with alluvium. A concrete belt (thickness c. 0.2 m) cast in the British Mandate era was revealed along most of the length of the channel. Some of the cover stones that were removed had been replaced on top of the concrete. The concrete belt was also uncovered in a previous excavation at the site (Porat 2005).
Area F. Two courses of dressed stones belonging to the outer wall of the circular springhouse were uncovered. Beneath these courses and slightly inside the building were foundation courses built of medium-sized fieldstones. The cover stones of the southern channel were also revealed, and the circular wall was probably built on top of the channel and incorporated within it. The small finds from the excavation include Roman potsherds and Neolithic flint tools.
The Geophysical Survey
The survey was conducted to the east of the road leading to the settlement of Zippori, between Nahal Zippori and Highway 79, with the aim of locating underground water channels. An EM31-2MK device was used, which measures the electrical conductivity of the subsurface without contacting the ground, leaving the survey area undamaged. A preliminary survey was first conducted over the southern part of a water channel between the British Mandate pumping station and ‘En Zippori, between two trial excavation areas that were dug to locate the route of the channel. The preliminary survey examined the device’s ability to locate an underground water channel and to recognize the anomaly formed by such an underground channel for later comparison. Following the preliminary survey, six areas were selected for electromagnetic surveying (Ars–Frs; Fig. 8, see Fig. 3). The survey in Area Ars investigated the existence of a channel branching off toward Nahal Zippori from the channel leading from the inspection station ‘En Zippori toward the north-northeast. In Areas Brs, Drs and Ers, the survey examined whether a channel entered the springhouse from the east. The area to the south of the springhouse could not be surveyed due to the presence of a large water pipeline and a road that traverses the area. In Areas Crs and Frs, the survey examined whether there were channels leading from the springhouse and ‘En Zippori toward a ditch to which the water was pumped, which is located near the entrance road to Zippori. To link Areas Crs and Frs, two more lines were surveyed between them, and a third line was surveyed perpendicularly across them. These lines are marked as dotted lines (see Fig. 3).
In Area Ars anomalies caused by the channel in the south and west of the area, probably leading to the building at ‘En Zippori are clearly visible. Also visible are two less pronounced anomalies (marked with a dotted line) corresponding to the direction in which the field was cultivated, and that were probably caused by it. No trial excavations were conducted in the area.
Area Brs. An anomaly was observed in the southern part of the area, roughly parallel to a water pipe leading southward from the pumps. The interference from the pipe makes it difficult to determine the exact width of the anomaly (6–12 m), which varies from north to south. This anomaly may be caused by a trench dug for the pipe that leaves the springhouse. An east-west trench (B1; length 10 m) was excavated 2.8 m beneath the surface that encountered a fill of stones and soil containing arrowheads and fragments of flint tools of a type known from the Wadi Raba culture.
Area Crs. A section dug on a north-south alignment (C1; length 10 m, depth 3.5 m) uncovered dark clay soil rich in organic matter that is probably a confined aquifer layer. No ancient remains were discovered.
Area Drs presented an anomaly (length 25–30 m) characteristic of a shallow feature (max. depth c. 1 m) consisting of two parallel lines. The width of the anomaly (c. 10 m) is similar to that of the trench anomaly in the preliminary survey. The northern part of the anomaly continues northwestward and has a uniform shape in this section. In its southern part, the anomaly veers toward the mandatory pumping station. A north-south section was dug perpendicular to the anomaly (G1; length 10 m, max. depth 3 m). In the center of the section, 1.6 m below the surface, was a large deliberate fill of small fieldstones and stone flakes; the fill was V-shaped in section (Fig. 9, see Fig. 7: Section B). This is probably a concentration of fieldstones that was used as a foundation for the built water channel that was uncovered. Moist soil was found beneath the fill. The section did not reach bedrock. The fill of stones probably continues along the entire northern part of the anomaly. After the anomaly veers toward the springhouse, it changes its shape, either because the fill has a different depth or composition, or because the anomaly here is caused by another factor.
Area Ers presented a relatively weak, narrow anomaly (c. 4 m), which is probably the result of field cultivation, similar to Area A. The anomaly was not excavated.
Area Frs presented marked differences in electrical conductivity from north to south, as well as some weak anomalies appearing as straight lines. A north-south section was dug (F1; length 15 m, max. depth 3.8 m) and another section was dug perpendicular to it (max. depth 3.8 m). Heavy clay soil containing a few medium-sized stones was found at the bottom of the sections. Dry mud was found near the bottom. Several Wadi Rabah-type potsherds were discovered 0.6 m beneath the surface. Another section (F2; max. depth 3.8 m) dug in the center of a field to the west of ‘En Zippori uncovered two dressed stones placed beside each other at a depth of 0.5 m below the surface; they may be the remains of an east-west wall. Based on the finds, the wall can be attributed to the period of the Wadi Rabah culture.
Underwater Examination of the British Mandate Pumping Station Shaft
The pumping station was built as a deep round shaft (diam. c. 3 m) with a central cavity in which iron pipes were installed for pumping water (Fig. 10). Remains of a rectangular pool (c. 1.8 × 2.0 m) built of ashlars with no traces of plaster were discovered 0.2 m beneath the shaft’s concrete floor. On the southern side of the pool was the inlet of a channel that fed the pool with water, as in the springhouse, while an outlet to a covered channel leading northward was found on the northern side of the pool. Four courses of the pool’s walls were identified, although further courses probably remain buried beneath the soil accumulated on its bottom.
Pottery. The excavation yielded body sherds from the Neolithic period (Wadi Rabah culture; Fig. 11:1–5), a Roman jug (Fig. 11:6) and an Ottoman jug (briq) and a tobacco pipe fragment (Fig. 11:7, 8).
Flint Items. 577 flint items were discovered in Areas A and C–F. Since the cultivation of the farmland made it difficult to separate ancient strata from more recent layers, the flint items were treated as a single assemblage. The assemblage contains 241 examples of knapping debris in which flakes are more frequent than blades: 81 flakes were collected (14% of the assemblage), compared with 42 blades (7% of the assemblage). Six irregularly shaped flake cores (c. 1% of the assemblage) are almost completely exhausted (Fig. 12:1–4). The core debitage comprises nine items (c. 2% of the assemblage; Fig. 12:5, 6), all of which are ridge blades (average length 0.05 m) attributed to the initial stages of core work. A total of 230 tools were found, including blades with notches and denticulates (Fig. 12:7–9), drills (Fig. 12:10), sickle blades (Fig. 12:11–13), axes (Fig. 12:14) and scrapers (Fig. 12:15). The sickle blades in the assemblage are delicately denticulated on long blades—a familiar type from the Pre-Pottery Neolithic B. Some of the tools were produced at the site itself.
The excavation in Areas A–F shows that the springhouse was built in three phases: in the earliest phase, in the Roman period, a water system was built that included a pool (Area B) through which water flowed from a channel in its southern part and out through a channel in its northern part. The pool functioned as a settling pool, in which the sand and debris sank to the bottom. To prevent the channels from clogging, the sand and dirt that sank in the pool were occasionally cleared. In the middle phase, during the Ottoman period, the settling pool continued to be used in the same way. Pottery, iron fasteners and wooden components from this period discovered on the pool’s original floor are probably associated with a simple water-drawing installation used to fill troughs outside the installation. At the end of the Ottoman period, the settling pool was blocked with silt, the water inlet from the southern channel was narrowed, the water outlet to the northern channel was partially blocked, the water level in the pool rose and the pool became a kind of lifting pool. The water was pumped from above to basins or small pools on its eastern side. In the final phase, during the British Mandate era, a concrete floor was laid in the pool and a pump was installed on its eastern side, which was used to irrigate fields and water livestock. It was probably in this period that the pool’s northern wall was breached, and the sides of the northern channel were raised with concrete beams to enable the passage of water; some of the channel’s cover stones were not reinstalled, leaving the channel partially open. No external saqiye system was discovered while excavating around the springhouse. There is no natural source of water inside the springhouse; the water source lies to the south of the pumping station, probably on the opposite side of Highway 79. The finding of flint tools indicates that the site was occupied during the Pre-Pottery Neolithic B and the Wadi Rabah culture; the remains of a settlement from these periods were destroyed during construction on the site. Past excavations in the area uncovered remains of a large settlement from these periods (Barzilai 2010; Milevski and Getzov 2014; Yaroshevich 2016; Getzov and Milevski 2017).
The geophysical survey showed several linear anomalies caused by the digging of a water channel. The anomaly in Area Drs was the most marked—the only one associated with ancient construction. No water channel was discovered between the area of the settling pool in the springhouse and the pumping ditch to the west of the survey area. In Area Ars, no channel was discovered that linked the settling pool to the site of the dam across Nahal Zippori to the north.
The examination of the British Mandate pumping station shows that it was built as a deep, round shaft with iron pump pipes in the center that led through a rectangular opening at the bottom into a settling pool from the Roman period, which is exactly the same size as the pool excavated in the springhouse. This pool also has a water inlet from the south and an outlet to the north. After the pumping station was built in the British Mandate era, the water level inside the building was evidently high and water could constantly be pumped and supplied to nearby localities. The builders of the pumping station were well-aware of the water system and the settling pools from the Roman period and the connection between them and sealed the settling pool beneath a concrete floor. Its blockage allowed the water level in the pumping station to rise, simultaneously reducing the supply of water to agricultural plots via the settling pool. The construction of such an advanced underground water-supply system could only have been carried out by a central authority concerned with ensuring a regular supply of drinking water.