The project comprises archaeological, geomorphological and historical research of a unique type of agricultural plots, termed by us a Plot-and-Berm Agroecosystem, developed according to our findings on the coastal plain of Palestine toward the end of the Early Islamic period: the late Abbasid or the Fatimid period (Taxel et al. 2018; Roskin and Taxel 2021b; Taxel and Roskin 2022; 2023). It is possible that this farming method inspired similar methods in later periods along the eastern Mediterranean coast, north Africa and the Iberian Peninsula (Roskin and Taxel 2021a). These plots were created in sand dunes through a process that involved the removal of the natural aeolian sand layer down to about 1 m above the sweet-groundwater table, the mixing of the sand with varying concentrations of refuse, and the piling of this mixture to raise berms around the sunken plots. A layer of dark, brown-gray soil (anthrosol) that contained refuse from nearby settlements was laid over the plots and the berms to enrich the sandy soil and to stabilize the berms. In the plots studied in this project, the refuse was brought from the city of Caesarea, and it included pottery and glass sherds, animal bones, coins and other finds, the majority of which date from the Early Islamic period, and the remainder—from earlier periods. The largest and best-preserved plot area was found south of Caesarea and was part of the city’s agricultural hinterland. The present excavations are the first to be conducted in this area since the pioneering excavation of Y. Porath in 1973–1974. His excavation was the first to identify the area as a system of sunken farming plots (Porath 1975; for a final report of Porath’s excavations, see Taxel and Roskin 2022); the plots were also found to cover the remains of a long Roman- or Byzantine-period ceramic pipe that supplied water to Caesarea from a spring on the north bank of Nahal Hadera (Porath 1990:102–106).
The aims of the survey were (1) to characterize the morphology of the agricultural system and to identify remains of structures and installations within it; and (2) to clarify the nature and date of the finds scattered on the surface of the system, and to identify chronological differences in their spatial distribution. The survey was largely limited to the tops of the berms and to plots that were not covered by dense vegetation. The survey yielded about 30 features (Fig. 1), including remains of structures, mostly on top of the berms or on their slopes. The excavations (below) were conducted in several of these features.
Most of the artifacts found on the surface—almost all of which are potsherds (Fig. 2)—represent refuse that was brought to the site for stabilization and soil enrichment of the agricultural system. The majority of the pottery dates from the Early Islamic period (eighth–eleventh centuries CE), a considerable portion dates from the Roman period (first–third centuries CE), and a small amount dates from the Byzantine period and the beginning of the Early Islamic period (fifth–seventh centuries CE) and from the Crusader period (twelfth–thirteenth centuries CE).
Based on these data, two shallow shovel trenches (1 × 1 m, depth 0.2 m; see Fig. 1) were dug on the slope of one of the berms in the southeastern part of the survey area, where a sizable concentration of Roman potsherds was observed. The pottery retrieved from the shovel trenches comprised mainly Roman-period potsherds and much fewer potsherds from the Early Islamic period. These findings suggest that Roman-period structures (farmhouses?) or concentrations of finds (refuse for enriching fields?) were still visible during the Early Islamic period in parts of the area designated for the agricultural system, and that they were removed in the process of developing the agricultural system.
Following the survey results, seven principal excavation areas were opened (A–G; Figs. 1, 3): Area A was opened at the top of a north–south berm, near the eastern edge of the survey area; Area B was opened at the top of a north–south berm, near the western edge of the survey area that borders on the sea and an adjacent plot; Area C was opened at the top of a berm; Area D was opened at the top of a berm and in a plot to its west; Area E was opened on the western side of a berm, and only a limited excavation was carried out in it; Area F was opened at the junction of two berms and in an adjacent plot; and Area G was opened at the top a mound in the southern part of the survey area, with a commanding view of its surroundings.
Apart from Area E, sediment samples were taken from all areas for OSL dating, as well as for particle size distribution, micromorphology and soil analyses. In Areas B and D, samples were taken from the upper layer of the anthrosol for flotation, to recover botanic samples. Water samples were taken from the groundwater in Areas B and D for pH and salinity tests.
Two excavation squares were opened 15 m apart. The northern square was excavated in order to expose a wall that was identified on the surface. The wall was dry built of fieldstones and roughly dressed stones that survived two courses high (Fig. 4). The wall was not founded on the dark anthrosol but on a later layer of sand. Meager pottery dating from the Early Islamic period was found next to it. The pottery and the layer on which the wall was founded suggest that it was built after the abandonment of the agricultural system or at a late stage during its use. The lack of nearby architecture indicates that this was a single wall that possibly protected a small plot to its east or supported a light structure built of perishable materials.
The southern square was opened at the highest point of the berm. The excavation reached deep into the artifact-rich soil layers that make up the berm (max. depth c. 1.4 m) without reaching the sterile sand at the base of the berm.
This area was chosen to be excavated because remains of a structure were identified on the top of the westernmost berm in the survey area. This berm runs along the seashore and borders a sunken farming plot to its east. The excavation exposed the remains of the structure, an earlier structure under it, and part of the farming plot 30 m east of the structure.
Architectural Remains. Three construction phases were observed, the latest of which was seen on the surface. Two walls were attributed to the earliest phase; only short segments of these walls were exposed (Fig. 5). One of the walls was constructed of kurkar blocks and cement chunks that contained large sherds of wide ceramic pipe segments, which were apparently dismantled from a Roman- or Byzantine-period pipe that was documented in the past along the coast.
Two walls forming a northeast corner that apparently enclosed an installation or a room of irregular outline were attributed to the middle phase (Fig. 6). The walls were constructed of fieldstones, roughly dressed stones and other building materials in secondary use. The northern of the two walls was built over the walls of the early phase and under the foundation course of the late structure. Within the area delineated by these walls were two layers of collapsed stones, one above the other, separated by a thin layer of soil.
In the late phase, a structure (5.0 × 9.0 m exposed dimensions; Fig. 7) was built on top of the berm that included two main units, western and eastern. The western unit was rather small, most of it having been washed away and destroyed by waves or stone robbery. Three walls survived; they were dry built of roughly dressed stones and small fieldstones founded on a layer of sand and shells. No floors were identified, as these must have been at a higher elevation and thus were washed away. A coin dating from the beginning of the Abbasid period (c. 750–830 CE) was found just below the surface next to the structure.
The eastern unit comprised a long, irregular and curved wall with a built stone platform abutting it on the west, jointly creating a crescent-shaped enclosure that bordered the western structure. The wall—which continues north and south—was dry built of fieldstones, roughly dressed stones in secondary use and fragments of architectural elements, such as a large fragment of a Roman-period Corinthian capital made of marble (Fig. 8). The stone platform, constructed of small and medium-size fieldstones, as well as dressed stones in secondary use, sloped from west to east, toward the curved wall; it seems that the wall delineated and supported the platform. Under the remains of the built platform was a fill of dark-gray soil containing much refuse. The function of the platform is not clear; it may have served as a drying surface for agricultural produce or for protecting the western structure from erosion by easterly winds. An excavation along the eastern, outer face of the wall determined that it was built into a dark, brown-gray soil fill replete with refuse. Another wall, found directly under the lowest course of the curved wall and slightly to its east, was built into the same fill; it appears that this wall was part of the foundation of the berm and of the structure built on it (Fig. 9). Another wall that was part of the berm’s foundation seems to have been built on the lower layer of the collapsed stones that were ascribed to the middle phase, as it abutted one of the other walls of this phase (see Fig. 6). The latest pottery and glass sherds found in the brown-gray soil fills were dated to the tenth or the eleventh century CE, and thus the late phase of the structure should be dated accordingly. Nevertheless, it is possible that the middle-phase walls are not much earlier than the late phase, although their relation to the nearby farming plot is not yet clear.
Farming Plot. In the depression of the ancient farming plot, the aeolian sand was removed from the dark, brown-gray anthrosol layer over an area of about 5 × 8 m (Fig. 10). Two probes were sunk from the anthrosol surface level down to the natural sand layer, on which the anthrosol lay. North of the soundings, a trench was dug down to the shallow groundwater level, about 1 m below the bottom of the anthrosol layer. In the southern probe, two wide horizontal bands filled with anthrosol (Fig. 11) were observed on top of the sand layer. These seem to be evidence of lignified roots of plants (vines?) that were cultivated on the plot, but they may have grown wild after its abandonment. The anthrosol layer contained numerous mixed finds, but not as many as in the anthrosol layers that made up the berms. Here, as in the berms, the latest pottery dates from the eleventh century CE, including, for example, a double-slip glazed bowl from Beirut.
This area yielded a large, round limekiln (inside diam. 4.5 m, outside diam. 5.1 m; Fig. 12). Its lower part was dug into the ground (depth c. 4 m) and lined with fieldstones and roughly dressed stones; its upper part did not survive. It appears that its original depth was not much greater. Most of the stones of the inner lining were found covered with a thin, uneven coat of lime. The stones of the upper part of the kiln collapsed into the kiln and were found at its bottom with an accumulation of a porous, friable substance that was apparently produced by the burning kurkar stones. Above these, the kiln was found filled with aeolian sand, almost devoid of finds. The kiln was surrounded by an embankment (width c. 3 m; Fig. 13) made of earth layers topped by a layer of fieldstones, which apparently served to bolster the kiln. The stones were probably laid to prevent the erosion of the embankment and to afford convenient access to the kiln. A large, square opening (c. 0.5 × 0.6 m; Fig. 14), constructed of large dressed stones, was set in the northern side of the kiln’s wall. Used for stoking and cleaning the kiln, the opening was approached through a trapeze-shaped north–south corridor dug in the ground and lined with stones. The floor of the corridor was made of a thin, whitish layer of lime, which also covered the threshold and the lower courses of the stones lining the corridor’s walls. The latest potsherds found in the fill layers of the embankment date from the eleventh or twelfth century CE, and thus date either the construction of the kiln or one of its latest phases of use or maintenance. A preliminary conclusion is that the lime produced in the kiln was used to enrich the soil of nearby farming plots with calcium.
Two sections were cut through this area: one across the entire width of the top of a berm in a north–south direction, and the second on the western slope of the same berm; the farming plot west of the berm was also excavated.
The Berm (Fig. 15). The section cut through the western slope of the berm exposed the fill layers from which it was constructed (Fig. 16) and revealed for the first time the meeting point between the anthrosol covering the berm and the anthrosol covering the farming plot. The anthrosol of the berm covers the edges of the anthrosol of the plot, indicating that at least technically, the plot was prepared first.
A low wall was found at the base of the berm, constructed of large, dressed stones in secondary use. The wall was built into the anthrosol layer, indicating that it was part of the agricultural system and not an earlier construction.
Farming Plot (Fig. 17). As in Area B, the aeolian sand layer was removed from the anthrosol layer over an area of approximately 3 × 8 m just west of the berm. A trench cut through the western part reached the shallow groundwater level, which, as in Area B, is about 1 m below the base of the anthrosol layer. A probe cut into the anthrosol reached sterile sand. Here too, the anthrosol layer contained much mixed material, mainly potsherds, the latest of which date from the eleventh century CE, such as a double-slip glazed bowl from Beirut.
A limited manually dug section on the eastern slope of another berm identified the upper deposition layers of the berm, which resembled those in Areas A, B and D.
In Area F, two subareas were opened, eastern and western (Fig. 18). The eastern subarea, comprising one excavation square, was opened on top of the junction of two berms running north–south and east–west, at a point where remains of a wall and clusters of stones were identified on the surface. In the western area, about 22 m to the northwest, part of the farming plot was exposed at the foot of the berm on which the eastern area was opened.
Architectural Remains. The survey conducted prior to the excavation identified on the surface in the eastern area remains of a north–south wall and clusters of stones to the north and southwest of the wall. Initially, a stone layer was removed in a controlled excavation, revealing the remains of a small structure built on a layer of yellowish gray sand (Fig. 19). The structure included the wall that was observed on the surface, which was abutted on the northeast by a curved wall. A short wall partitioned the space between the two walls into two cells—a small square cell on the south and a northern cell, which apparently was originally oval in shape; their floors seem to have been a layer of compact earth (sand), as no trace of a more substantial floor was preserved. At the southern end of the structure was a square construction consisting of three closely placed east–west rows of stones, perhaps the base of an installation. Immediately to the south of the structure was a small, irregular cluster of stones—perhaps stones that collapsed from a structure—which extended southward and eastward beyond the excavation square. The stone layer that covered the structure represents either a destruction, possibly due to stone robbery, or some deliberate covering that may have occurred as part of the abandonment process.
Sections cut to the south and west of the western wall revealed three light gray sediment layers (anthrosediment; thickness 0.1 m each), separated by grayish sand of a lighter color. The anthrosediment layers do not abut the wall of the structure, nor do they pass under it; the construction of the structure may have severed them. Most of the pottery retrieved from the soil fills that sealed the structure and from the fills deposited under it dates from the Roman period (first–second centuries CE), along with a smaller amount of pottery from the Early Islamic period (ninth century CE onward).
The structure was most probably part of the local agricultural system. It was constructed on top of the meeting point of berms bounding farming plots, and it may have served for the storage of equipment or produce as well as for short-term lodging for workers who cultivated the nearby plots. The dominance of Roman-period pottery in the fills seems to point to activity at that time in the area. It is possible that the builders of the agricultural system in the Early Islamic period made use of accumulations rich in Roman-period finds that were exposed on or near the surface, and they mixed them with refuse that they brought from Caesarea.
Farming Plot. In the farming plot in the western subarea, the aeolian sand layer was removed from the anthrosol over an area of about 4 × 6 m (Fig. 20). The anthrosol layer was thicker (thickness c. 0.8 m) here than the anthrosol layer exposed in Areas B and D. It was composed of at least two sublayers: an upper sandy one (thickness c. 0.6 m), light gray and rather moist, and a lower, hard, dark gray and dry one (thickness c. 0.2 m). Additionally, a trench at the southern edge of the area was excavated down to the shallow groundwater level, about 1.5 m below the base of the anthrosol layer. Two probes sunk from the surface of the anthrosol layer reached the sterile sand layer on which the anthrosol rested. The two anthrosol layers in the plot contained fragments of double-slip glazed bowls from Beirut, dating no earlier than the eleventh century CE.
The area was opened at the top of a mound (height c. 4 m) with a commanding view of its surroundings, where remains of a structure were identified on the surface (Fig. 21). The excavation here soon made clear two facts: (1) The mound is an independent feature on a farming plot surrounded by berms, close to the northern enclosing berm; and (2) the structure on top of the mound is one of three small structures (Structure C) exposed by Porath during 1973–1974 in the agricultural system (Porath 1975). Porath’s excavation revealed the outline of the structure and its floor, on which were found—in situ—pottery vessels, an iron knife and a Fatimid silver coin. That excavation did not expand outside of the structure, nor below the top of the walls (for a description of the finds, see Taxel and Roskin 2022).
Since Porath’s excavation, most of the floor was destroyed, and some stones were removed from the walls, probably as the combined result of natural causes and human agency. The current excavation re-exposed the structure, including the foundations below floor level of both inner and outer walls. Additionally, two stratigraphic sections were excavated: one that cut the northern slope of the mound from top to bottom, and a smaller at the southern foot of the mound.
The Structure. The structure was constructed at the top of the mound, overlooking large parts of the agricultural system, including the structure in Area B, near the seashore, and the limekiln in Area C. This is a square structure (c. 3.2 × 3.7 m; Fig. 22), with walls surviving three courses high—two foundation courses and one regular course above them. The walls were constructed of fieldstones and dressed stones of various sizes; some of the building stones had collapsed over the upper part of the northern slope of the mound. The dressed stones, some exhibiting traces of plaster, were in secondary use, most probably brought from Caesarea. The stones of the wall were held together with reddish brown hamra that served as a dry bonding agent; this method was also observed in the square structure in Area B. A square stone with a carved rectangular socket, possibly a hinge socket of an entrance, was set in the eastern wall, near its southern end. The excavation along the outer face of the walls reached just below the foundation courses, in a homogeneous, grayish yellowish sandy fill that contained mixed finds from the Roman and Early Islamic periods. Only a small patch of the structure’s floor of flat kurkar stones—exposed in its entirety by Porath—was preserved in the southwestern corner of the structure. The floor and walls of the structure were founded on the same homogeneous, grayish yellowish sandy fill that was found below the foundations. The structure is small but sturdy, and seems to have had a roof made of light materials, although it is possible that the stone walls supported a second floor built of wood. A patch (c. 2 × 3 m) of what seems to be a bedding—a layer of Glycymeris nummaria shells, most of which were laid on their concave side over a layer of hamra soil—was exposed east of the structure. The bedding may have carried a floor in a vestibule or small courtyard in front of the structure’s entrance.
The Sections. The section that cut through the northern slope of the mound exposed the upper, modern, aeolian sand layer and into the ancient layers that form the mound (Fig. 23). These include alternate light grayish yellow and gray anthrosediment layers, all containing meager finds from the Roman and Early Islamic periods. A thin and sparse layer of Glycymeris nummaria shells sealed the topmost layer, possibly indicating that such a layer covered the surface of the mound to protect it from erosion. At the northern end of the section was a segment of a wall running in a general north–south direction. The wall, having no uniform height or width, was poorly constructed of several courses of fieldstones and a few roughly dressed stones; its eastern end curves slightly to the southeast. It was constructed on or into a rather dark brown-gray anthrosediment layer sloping northward, which contained numerous mixed finds, mostly from the Early Islamic period, including fragments of double-slip bowls from Beirut, dating from the eleventh–early twelfth centuries CE.
The coarse and makeshift character of the wall, its location at the foot of the mound, its curving course and the fill layers covering it that appear to be deliberate—all suggest that it was built to encircle the mound, continuing to the southeast and southwest, and support its lower earthen fills. In hope of confirming this hypothesis, a smaller section was cut at the bottom of the southern slope of the mound. And indeed, the section revealed a short wall segment constructed of small and medium-sized fieldstones that was founded at least in part on a light gray anthrosediment layer. The distance between the structure and the outer face of the walls exposed in the southern and northern sections is almost identical (8.7–8.8 m), suggesting that the wall in the southern section represents another segment of the circumferential wall that bounded the mound. The distance between the two wall segments is about 20 m, and judging by the slightly curved outline of the northern wall segment it can be speculated that the base of the mound was rounded.
The mound with the structure on top represents a relatively late, probably short, phase of the agricultural system. This is indicated by the location of the mound—in the middle of a farming plot that apparently ceased to be in use at this time—and by the eleventh-century ceramic material found in the mound’s foundations, on the structure’s floor and at the top of the northern slope. The reason for the construction of the mound and the structure, surely involving considerable investment in time and resources, is not clear. It is possible that at least one of its functions was to watch over the agricultural system, or at least over a large swath of its southern part.
The findings from Porath’s excavations in the 1970s and the preliminary results from the current research project suggest that this agricultural system was created during the Fatimid period and was in use until the beginning of the Crusader period. However, it cannot be ruled out that certain parts of the system were erected already in the Abbasid period, as is hinted by some preliminary results from the OSL dating. The system comprised dozens of sunken plots of various sizes, as well as structures, walls and installations. The limekiln found in Area C, and a similar kiln apparently found by Porath, tell of an increased effort to produce lime toward the end of the use of the system, or shortly thereafter. The considerable area covered by the system, the size of the plots and the berms and the sheer scale of the anthropogenic material brought in from Caesarea for stabilizing the berms and enriching the soil—all attest to the considerable investment in resources and manpower in the erection and upkeep of the agricultural system. We believe that this agricultural enterprise, like several others that were created along the coastal plain, was initiated by the authorities, although at this stage we cannot specify the motives, or indeed the nature, of the agricultural activities undertaken or the crops grown in these systems.