The site, which was until recently concealed by thick vegetation in a marshy area, is located in the Carmel coast eastern trough, at the western foot of the southern Carmel Ridge, between Kibbutz Ma‘ayan Zevi in the east and Kibbutz Ma‘agan Mikha’el in the west. In its vicinity lie the sites of the Northern Dam to the north (Olami, Sender and Oren 2005:15–16, Site 3) and Horbat Kevara to the southeast (Chocron 2018). The excavation revealed a rectangular structure (Fig. 2) that is probably a massive rising pool dating from the Roman period.

The structure (external dimensions 34 × 76 m, internal dimensions 24 × 66 m; Fig. 3) has massive walls (max. width 5.35 m) built of two faces of kurkar ashlars arranged as headers and stretchers with a core of small stones, quarrying debris and some lime; the preserved level of the core is higher than that of the walls’ ashlar courses (Fig. 4). Traces of mortar were discovered in the masonry joints of the walls’ inner faces, along with remnants of light-colored plaster (Fig. 5). The structure is asymmetrical: its western end is slightly wider than its eastern end, so that the northern and southern walls narrow gradually eastward (W101—width 4.20–5.35 m; W103—width 4.0–4.3 m), and the eastern and western walls differ in length (W102, W104; 34 m, 32.7 m respectively).

A probe excavated north of W101 (L117, L136; height 3.55 m; Fig. 6) revealed seven courses: the four upper courses were built of ashlars (courses' height 0.4 m, 0.2 m, 0.55 m, 0.45 m from top to bottom) retaining traces of plaster; below these was a course made of debesh (height 0.6 m); and at the bottom were two wider courses built of high-quality gray cast concrete (height 0.5 m, 0.55 m). The two lowest courses lay on top of a well-preserved sawn timber beam (height 0.3 m). Another probe (L126, L130), also dug to the north of the wall, provided a more complete picture of the wooden beams. At the end of the wooden beam at the bottom of the foundation (height 0.3 m) were fitted hermetically together, one on top of the other, three wooden beams (each 0.2 m high), which overlapped it and were set adjacent to it on the north along c. 0.3 m (Fig. 7). On the inner, southern side of the first wooden beam stood vertically a round wooden pole (preserved height 0.4 m). Based on the casting marks left on the foundation courses, the wooden pole was higher than the beams and reached the top of the debesh course.

The southern part of the outer face of W102 protruded westward, adding the width of a single stretcher (0.3 m), along 4.4 m, ending exactly parallel with the inner face of W103. On the outer face of W102, two square grooves (width 0.25 m) were hewn in the wall’s stretchers 8.3 m apart: the northern groove was hewn 13.0 m south of the end of the northern wall, while the southern groove was hewn 12.4 m north of the end of the southern wall (Fig. 8).

Remains of two clay pipes (L106; diam. 0.12 m, preserved length c. 1.5 m; Figs. 9; 10:8, 9) coated with gray mortar were discovered in W104, 8 m from its northern end; the pipes were set one beside the other and were inserted across the wall, running through its core. Both pipes are very fragmentary, but an entire segment was preserved at the eastern end of the northern pipe. The pipe segments have ‘male’ and ‘female’ ends and the male ends face westward, into the structure.

A probe opened prior to the excavation beside the inner face of the western wall discovered a pottery horizon inside the building. Collapsed building stones were found on the inside of the structure’s corners, except for its southeastern one (Fig. 11).

The probes dug at the site uncovered two layers of particular note. One is a layer of black soil (thickness 0.1–0.3 m; Fig. 12) that covers the entire site to varying depths, in accordance with the modern topography, so that it is shallower in the center of the structure than beside its walls. For example, the soil layer abuts the outer face of W101 at 3.97 m asl and its inner face at 3 m asl, but it also covers the wall at 4.96 m asl. The second layer of note consists of light-colored soil containing crushed kurkar; this layer appears, for example, north of W101 and west of W102 at 3.75 m asl.

The structure’s southeastern corner is abutted on the outside by a level of fieldstones and building blocks of various sizes. The level becomes narrower toward the structure (L121; length 4.85 m, width 1.1–1.7 m; Fig. 13) and cuts into the black soil layer.

On top of W101, c. 7 m west of the structure’s northeastern corner, were the remains of a surface paved with white mosaic tesserae (L133; 3.10 × 5.65 m; Fig. 14). The surface was enclosed on its east and west by walls built of a single row of stones (W132, W134; width 0.4–0.5 m) preserved to the height of a single course. The southern wall was built in a similar manner, on top of the southern face of W101. Meager remains of plaster and a foundation made of potsherds were identified at the points where the mosaic floor abutted W132. The northern part of the mosaic was destroyed, and no opening leading to the surface was detected.

A probe outside the structure’s southwestern corner revealed a vertical clay pipe (L140; diam. 0.26 m, length 0.34 m, wall thickness 3 cm; Fig. 15) inserted inside an irregular niche cut in W103. At the top of the pipe was an octagonal cement cover (height 7 cm) with a round depression in its base that matched the diameter of the pipe. The center of the pipe contains two holes (diam. 0.11 m) on opposite sides, one in the north and the other in the south. A horizontal pipe (diam. 0.1 m, length 1.6 m) jutting out from the southern hole was composed of segments (length 0.32 m) placed alongside each other in a southerly direction, although they were not hermetically sealed: the first segment sloped downward, and the remainder lay horizontally. The entire array of pipes was encased in a fill of marine shells.

Fragments of modern roof tiles and clay pipes were collected from the surface around the building. The site yielded very meager pottery finds, which include Roman and early Byzantine ware: a cooking pot dating from the late first–early third centuries CE (Fig. 10:1), second-century CE bag-shaped jars (Fig. 10:2, 4–7) and a bag-shaped jar from the third–fourth centuries CE (Fig. 10:3). No sealed strata were discovered at the site by which the architectural remains could be dated, but radiocarbon analysis of two samples taken from the wooden beams in the foundations show that the wood dates from the first century BCE–first century CE. Based on its analysis, the wood is black pine (Pinus nigra).


The excavated structure contained exceptionally well-preserved two-thousand-year-old wooden beams attesting to an ancient method of construction on water-saturated ground: based on the location of the wooden pole inserted on the inside of the wooden beams (the casting side), the wooden structure was probably intended to shore up sections dug into the soil while casting the foundations. The wooden beams at the site are the lower part of these supports, which remained trapped beneath the structure and thus could not be removed for reuse. The wooden beams were excellently preserved from lying in water-saturated clay for the centuries preceding the excavation.

As the wall cores extended above the structure’s ashlar courses, it can be concluded that the walls had at least one additional, upper course that was not preserved. Further evidence for this is provided by the piles of collapsed masonry in the three inner corners. No such collapse was found in the southeastern corner; this absence may be related to the laying of L121, which abutted the structure on the outside.

The layer of black soil found at the site and in its surroundings is probably a marshy sediment that covered the site after the structure was abandoned; the soil has peat-like properties typical of decaying organic matter in swampy areas. It is supple and tends to subside when dry. The light-colored level found in one of the probes outside the building probably represents the working level during construction, i.e., the level of the ancient surface.


This vast, unique structure was built using exceptional construction technology under particularly hostile conditions; the whole structure was cast into natural clay levels in a marshy area that poses considerable construction challenges, and yet it remained stable, with no visible damage. No cracks or depressions were detected in the structure although today the ground drained around it has subsided.

We propose interpreting the structure as a uniquely constructed rising pool, built to isolate water from freshwater springs in a region of brackish water springs. This goal was achieved by casting a massive crate to trap spring water, taking advantage of the high aquifer at the site and preventing the freshwater from being contaminated by water outside the structure. Initial support for this proposal was provided when the groundwater level in the probes inside the pool structure immediately began rising to a higher level than it did in the probes outside the structure. Further evidence emerges from a preliminary inspection of the water salinity at the site and in the immediate vicinity, which shows that the water inside the pool is fresh, while the water outside the pool and in its immediate vicinity is brackish.

The grooves in the western wall were apparently intended to drain any excess water from the pool; the place where the water flowed out of the pool was not discovered. Pottery from a probe opened prior to the excavation near the inner face of the western wall probably accumulated at the bottom of the pool during its use.

Based on the huge investment in a project of this scale, namely the massive construction and the transportation of special wooden beams for the foundations’ formworks, the structure was probably a state-sponsored endeavor. In addition to the radiocarbon dates of the wood samples, relative dating can also point to the time of the pool’s construction. This is provided by the Byzantine-period Tanninim Reservoir, whose water flowed to Caesarea via the Low Level Aqueduct (Dray 2014:81–91). The reservoir made the pool, which was within its boundaries, obsolete, as it did to other structures, such as a Roman mausoleum at the foot of Tel Mevorakh (see Olami, Sender and Oren 2005:45, Site 64) and the remains of a church at ‘Enot Timsah (see Yalqut Ha-Pirsumim [Official Gazette] 1964:1403).

The level of stones (L121) is later than the structure and may even have been laid in modern times, as it contains, among other things, ashlars taken from the structure; moreover, it cuts into the black layer of marshy soil. It was probably a causeway designed to provide access through the marshlands to water stored in the structure.

The system of pipes at the southwestern end of the structure is modern; it is part of the swamp drainage system supervised by the engineers Léon Caron and Dov Koblanov on behalf of PICA in 1924–1928 (Avitsur 1985:186–192; Ayalon 2007:20–25).