The current archaeological investigations of the city of Elusa began within the city boundaries (2015–2019; Pickartz, Tezkan and Heinzelmann 2015; Heinzelmann and Erickson-Gini 2015; Di Segni 2018; Heinzelmann et al. 2017; Schöne et al. 2018; 2019; 2022) and then extended to its immediate periphery (2018–2019) and the broadening agrarian landscape (2019–2020). Work at these different locations combined, in addition to excavations, surveys, remote sensing and analysis of recent and historical aerial photographs and satellite images.

Investigations of the extra-urban area of Elusa aimed at obtaining a better understanding of the development of the city and its hinterland, from the establishment of the settlement in the third century BCE to its end in the eighth century CE. The extra-urban survey focused on four belts (1–4; for Belts 1–3 see Fig. 1) surrounding the city. The interior belt (Belt 1; width 50–200 m) comprises mainly large waste mounds arranged in a distinct ring configuration. Investigations then progressed farther from the city, to a belt of mainly horticultural land use of a less well-defined aerial shape (Belt 2; width 80–400 m). Floodplain horticulture was practiced in suitable locations, such as the banks of Nahal Besor (Wadi el-Khalaṣa), west and south of the city, and its subsidiary, Nahal Atadim (Wadi el-‘Awsaja), north of the city. These cultivated areas were in turn surrounded by a belt of cemeteries (Belt 3; width 500–800 m), containing thousands of individual burials of a simple expedient type and a few more elaborate built tombs, all still visible on the surface. Beyond the cemeteries, a densely utilized agricultural landscape—the city’s hinterland, comprising dozens of farmsteads and villages—extends over many kilometers into the surrounding countryside (Belt 4).

 
Belt 1: Waste Mounds
Distinct waste mounds on the edge of settlement, sometimes of considerable size, are a characteristic of all Negev settlements of the Byzantine period. Few, however, are as impressive as those of Elusa, where they reach a width of 140 m and rise to a height of 12 m above their surroundings (Fig. 2). These mounds were systematically surveyed and mapped using GPS. They were shown to cover an area of 14 hectares and form an almost closed ring around the built area; the accumulations documented to the north and east were much more substantial and continuous than those on the south and west, near and along Nahal Besor. The dumps begin immediately at the exterior borders of the settlement, and their proximity to the city may reflect an attempt to minimize transport of the waste. Gaps at irregular intervals between the waste mounds indicate that the mounds were positioned so as to leave room for the city’s arterial roads to pass through. In some of these locations, the streets appear to widen into large, roughly square-shaped open spaces that may have served as trading areas and for setting up animal enclosures.
The sediment composition of the mound surfaces, largely sand and ash, appears very uniform (Fig. 3). These surface sediments contain mainly small fragments of pottery, animal bones and other organic remains, as well as residues of industrial production, such as slags and misfires. Coins of late antiquity are found on the surface in unusually high numbers. Test excavations by the University of Haifa in 2015 showed that the mounds built up in numerous easily distinguishable layers of (thickness 5–10 cm), apparently originating from repeated individual events of dumping (Bar-Oz et al. 2019). This indicates regular, perhaps even annual, deposition episodes, probably as a result of systematic cleaning operations in the urban area. The high sand content of the mounds seems to indicate that the frequent sandstorms in the region necessitated such regular cleaning operations. Undoubtedly, the large waste dumps of Elusa demonstrate the existence of a highly developed and labor-intensive form of communal waste management.
The chronology of the Elusa waste mounds has only been partially clarified: while the end of their use is dated to the mid-sixth century CE (Bar-Oz et al. 2019:8241), it is yet unclear when this form of systematic waste disposal began. Nevertheless, this can be deduced from indirect evidence, for example, the observation that the mounds encircle the city at its greatest areal extent, which was reached by the late second or the early third century CE. The city’s extent at that time is indicated in part by several Roman-period tombs that probably lay just outside the city limits. In several places, these tombs are overlaid by the waste mounds, attesting to their earlier formation (Heinzelmann and Erickson-Gini 2015; Heinzelmann et al. 2017; Schöne et al. 2019). It is noteworthy that by the mid-fifth century CE the streets in the city center had been paved with high quality stone material (Heinzelmann and Erickson-Gini 2015), presumably making them much more amenable to regular cleaning. Excavation of stratigraphic sequences of several streets within the city showed that regular accumulation of waste ceased when the pavements were laid down but had resumed around the mid-sixth century CE, when dumping of waste in the exterior mounds finally came to an end. It could be suggested, therefore, that the waste mounds formed in most part over a relatively short period of about 100 years (Heinzelmann and Erickson-Gini 2015; Heinzelmann et al. 2017; Schöne et al. 2018; Heinzelmann et al. 2022).
 
Belt 2: Horticulture Areas
A special form of agriculture—floodplain horticulture—seems to have been practiced along the periphery of Elusa, just beyond the ring of waste mounds. The wadis provided both a suitable soil—relatively deep loess soils found on their banks—and the possibility of convenient irrigation, either by floodwater during wintertime or by water from wells sunk into the wadi banks during the summer. An additional source of irrigation water may have been the wastewater collected by large sewers from the urban area (Heinzelmann et al. 2022; Schöne et al. 2022).
Several areas of horticulture were identified surrounding the city on all sides and situated along the course of both Nahal Besor and the Nahal Atadim. These areas were apparently bordered by low dry-stone walls, which subdivided them into cultivation plots, possibly owned by different families. The present-day surface of these areas is strikingly flat and strewn with especially high quantities of small pottery fragments, very likely the result of manuring with household waste (see below). Such waste, which was rich in organic material, could have been obtained directly from the city or the nearby waste mounds. Alternatively, the high density of surface pottery could be due to post-depositional compaction of the cultivated soil by wind erosion, although this is less likely.
The relatively limited size of the horticultural areas indicates that they were utilized for small-scale vegetable gardens that served to supply the day-to-day needs of the city inhabitants, rather than for large-scale crop cultivation for commercial purposes. Remains of plant foods consumed in the city, including barley, wheat, lentil, grape, date, fig, olive, walnut, almond and hazelnut, were identified both in material from the waste mounds and in fills from within the city (Bar-Oz, Weissbrod and Erickson-Gini 2016; Heinzelmann and Erickson-Gini 2015; Heinzelmann et al. 2017; Schöne et al. 2019).
The horticultural areas northwest of the city are pierced by narrow extensions branching off from the waste mounds in a number of places, perhaps intentionally to provide some protection from wind erosion.
 
Belt 3: Cemeteries
Surprisingly, an especially large number of mostly well-preserved burials was documented in the survey, although these burials could have been easily covered at any time in the past with mobile sand dunes, which are common in this part of the Negev (Heinzelmann et al. 2017; Schöne et al. 2018). The survey revealed no signs of grave robbery, and the only damage to graves was caused by erosion, mainly at wadi edges (e.g., see Fig. 5). Due to these favorable conditions, it was possible to determine the approximate extent of the cemeteries and gain insight into their spatial organization. It appears that much of the city was surrounded by a belt of burial grounds (at least 114 hectares) of varying grave densities. Fewer graves were found on the western bank of Nahal Besor, probably because this area, located in the lee of the hills farther to the west, is mostly covered by thick sand deposits. The possibility that cemeteries also extend to the northwest of the city is reinforced by the presence of graves in a few small sand-free enclaves in this area. It may be extrapolated, therefore, that the maximum extent of the Elusa cemeteries was c. 150 hectares.
A detailed investigation of the graves was conducted in several areas, covering a total of c. 24 hectares and containing 5490 individual burials (Fig. 4). The location of all these graves was recorded by GPS, and about a quarter of them were sufficiently well preserved to record both the size and orientation. The grave orientation was uniformly northwest–southeast (-30 to -45 degrees north). Observations of the interior of several of the graves—mostly graves that were exposed by erosion, but also two intact graves that were opened and excavated—indicated that they were simple cist burials, dug to a depth of c. 1.5 m and mostly constructed of local limestone slabs (Figs. 5–7). The graves were covered aboveground by heaps of densely packed pebbles from the nearby wadis (Figs. 6, 8). Dozens of gravestones, many of them inscribed, were removed from the site by explorers and archaeologists in the late nineteenth and early twentieth centuries (Jaussen, Savignac and Vincent 1905; Alt 1921:26–31; Kirk and Gignoux 1996). Only one fragment of a gravestone with remains of a cross was observed in the 2019 survey; it was left in situ. The burials are generally evenly spread across the terrain, although small clusters of 2–6 burials can often be observed, presumably representing family burial plots.
While inspecting the cemeteries, mainly in the northern and eastern parts of the cemetery belt (Fig. 4), close to 1600 fragments of pottery wasters and misfires—clear-cut evidence of pottery production—were noted. The misfires generally comprised well-fired clay lumps, some of which contain recognizable sherds of malformed vessels made of the greenish material typical of the Byzantine period in Elusa. Such misfires were often incorporated in the pebble heaps covering the burials. It is possible that the digging of the graves in this part of the cemetery belt disturbed the remains of earlier sites of pottery production.
Systematic collection and analysis of surface finds was conducted in ten areas (A–I, K; Figs. 1, 4), each divided into two sub-areas (5 × 5 m each) situated diagonally to each other, to evaluate the chronology of the cemeteries. The collected surface pottery ranges in date from the Hellenistic period to the Early Islamic period, corresponding to the overall duration of settlement activity at Elusa. Nonetheless, the pottery’s chronological span clearly differed between the ten study sites, pointing to different phases of burial in the various parts of the cemetery belt. The earliest of the cemeteries, where the earliest surface pottery was found, seems to lie in an area east of Nahal Atadim (Areas A and K). A typological analysis of the pottery revealed that the proportion of thin-walled drinking vessels in these earlier cemeteries was higher than anywhere else. Cemeteries used at a later time (Areas B–F) yielded mostly coarse wares and cooking vessels. It is possible that these typological differences reflect in part temporal changes in mortuary customs that affected the composition of grave goods.
The two intact graves that were excavated (Tombs 5137, 5485) revealed further details of the interior construction of graves. Tomb 5137 (Fig. 6; Schöne et al. 2018), located in the southeastern part of the cemetery belt, comprised a shaft (depth 1.3 m) covered by a pile of densely packed wadi pebbles. The shaft cut through the remains of an earlier burial, suggesting that the earlier grave was no longer visible aboveground at the time that Tomb 5137 was constructed. A relatively small stone cist (0.63 × 1.27 m; orientation -38 degrees north), its side walls and cover constructed of soft white limestone slabs, was uncovered at the bottom of the shaft. The limestone was likely obtained in one of several local quarries found along Nahal Atadim and Nahal Besor. The burial cist contained the skeleton of a child, of which the skull and long bones were still preserved, although in a very fragile condition. The head was in the northwest. The body posture—with arms tightly cramped—indicates that the body was wrapped in a cloth at the time of burial. The finds included two rings, one made of iron and the other of bronze, which were uncovered at the location of the child’s left hand. Also found were several beads, dark blue ‘melon’ (ridged) beads (diam. c. 0.5 cm) and dark blue and greenish elongated beads (length c. 1.3 cm), located in connection with the child’s hip; the beads may have decorated a garment worn by the child. No chronological diagnostic artifacts were recovered from this burial.
Tomb 5485 (Fig. 7), located in the northeastern part of the cemetery belt—on the northern bank of Nahal Atadim—was found exposed down to the cover slabs of the cist due to weathering, but was otherwise undisturbed. The cist (0.70 × 1.71 m; orientation -42 degrees north) was made of local limestone slabs. The skeleton was somewhat better preserved than that of Tomb 5137, comprising the skull, the long bones and the hip and rib bones, although it too was in a very fragile state. The head was in the northwest, and the flexed posture indicated that the corpse was wrapped in a cloth at the time of burial, as was the case with Tomb 5137. No grave goods were found. A cursory anthropological examination of the skeleton revealed that it belonged to a young female. Both skeletons remained in situ and were reburied after documentation.
 
Belt 4: The Hinterland
The survey of the city's broader hinterland—an ongoing project—aims at evaluating the amount of land that was under cultivation and the irrigation technology used. It is understood that farming based on run-off irrigation and extensive terracing—techniques which were widely practiced in the Negev Highlands (Evenari, Shanan and Tadmor 1971:95–119; Bruins 1986:33–51; 2012)—was ill-suited for the relatively flat topography of the surroundings of Haluza. No evidence of terracing has been found along the part of Nahal Besor near the city, where this form of farming was likely hindered by the wide wadi channel prone to destructive flash floods. Only a few subsidiary wadis suitable for terrace farming are found near Haluza, such as Wadi Umm Ghasuma, a subsidiary of Nahal Besor. Other features documented in the survey so far are farmsteads, boundary walls, cisterns, winepresses and dense concentrations of surface ceramics.
The survey carried out in 2019—2020 was conducted within the area bounded by Road 222 and Nahal Besor, beginning at Haluza and extending c. 3.5 km to the southeast, up to the modern cultivated fields of Kibbutz Revivim (Heinzelmann et al. 2019/20). Some work was also conducted west of Nahal Besor, although the survey in this area has progressed slowly due to military restrictions, allowing access only on weekends. In preparation for the survey, available aerial and satellite images covering a radius of about 4 km around the city were analyzed. These images included early aerial photographs taken by the British Royal Air Force and the Israeli Air Force from the 1940s to the 1960s, which proved to be especially useful for detecting features of interest (Figs. 9, 10). More than 450 points of interest (POI), representing suspected ancient remains other than terraces or boundary walls, were identified (Fig. 11). Ninety-nine of the POIs were subsequently surveyed in 2019 and 2020.
To date, a total of c. 10 sq km of land (1012 hectares; Fig. 12) have been surveyed. Favorable conditions for the survey included the mostly flat, unvegetated terrain, offering unhindered visibility of the ground surface and built features that were often visible above ground. However, it was also apparent that over extensive parts of the surveyed area, ancient remains were buried due to the movement of sand. A comparison of present-day conditions with those observed in aerial photographs from the 1950s shows that in some cases even in recent decades entire structures, even quite large ones, were buried under the shifting sand. It is estimated that 70–80 percent of the area under investigation is now covered by sand accumulations (e.g., the area around Farmstead SI-001 [see below]; Fig. 13). In some places these may reach considerable depth, such as on the windward side of natural promontories. To account for the distorting effect that variable topography and sand accumulation across the area may have on the survey, a digital terrain model (DTM) with a resolution of c. 20 cm per pixel and a high-resolution orthophoto mosaic were created for the investigated area using a drone (Figs. 14, 15). The model both revealed the topography of the area and provided additional information on archaeological features.
The ground survey was restricted to the sand-free areas within the surveyed quadrant (30%). For the first 90 hectares of the survey, the surveyors were first spaced at intervals of 10 m, but the intervals were increased to 50 m once it was realized that only one new POI which was not previously identified by remote sensing was found. All find spots were registered by GPS, photographed both from the ground and with the aid of a drone and described in detail. During the ground survey, 81 of the 99 remotely sensed POIs in the area were confirmed as archaeological features. Twenty-eight additional features were registered during the survey, most of which were too small to be identified in the aerial and satellite images, such as cistern openings. Altogether, 109 ancient sites were registered in the sand-free areas. Assuming that the density of find spots in the sand-covered areas is similar to that in the sand-free area (one site per every 3.5 hectares), it is estimated that more than 400 such sites exist within the surveyed quadrant as a whole. The survey also registered over 200 wall remnants, such as terraces and field boundary walls; most of the walls were previously identified as lineaments in the aerial photographs and were thus verified on site. These walls are widely distributed across the surveyed areas, indicating the extent of recent landscape modification and possibly the importance and organization of private land ownership.
A systematic collection of surface finds was conducted at all sites, and 54 of them provided enough material to permit dating. Selected sites were also investigated through geophysical prospection by means of magnetometry and electrical resistivity (e.g., the area around Farmstead SI-001 [see below]; Fig. 13), and in some cases a georadar. The sites registered in the survey varied widely in nature, and several different types were identified: settlement sites, farmsteads, winepresses, cisterns and wells, as well as terrace walls and boundary walls.
 
Settlement Sites. The largest site uncovered in the survey was a village of about 4.1 hectares (SI-006; Fig. 16), situated c. 2.1 km south of Haluza on a natural terrace between Nahal Besor and one of its tributaries, Wadi Umm Ghasuma. Preliminary evaluation of surface pottery from this site indicates a date between the Roman and the late Byzantine periods. Two nearby cemeteries, one to its south (NP-008) and the other to its north (NP-009), undoubtedly belonged to the village. This settlement may be one of nine Negev settlements mentioned by name in the Nessana Papyrii, of which the archaeological remains are yet to be identified (Kraemer 1958:228). A smaller village (SI-018), also associated with a series of tombs, and having several deep wells, which like SI-006 overlooks Wadi Umm Ghasuma.
 
Farmsteads. These are single collapsed structures, often occurring as whitish mounds of building debris (rarely larger than 15 × 15 m, height 1–2 m; Fig. 17). Wall remains were often recognizable on the surface of these mounds. These small to medium-sized buildings—typically found along the banks of wadis and in intervals of less than 100 m from each other—appear to represent farmsteads. The distribution of the farms along the wadi indicates that each managed, and perhaps privately owned, a section of the wadi system, which was likely utilized for farming based on a form of run-off irrigation. Isolated farmsteads were also found in flatter terrain, away from the wadis, where a different form of farming was likely practiced (see below). Some of the farmsteads comprised two adjacent buildings (e.g., SI-001; Fig. 13), indicating that they were occupied by two separate households or that the buildings were used for different functions. A small number of simple burials or built tomb structures were found in connection with some of the farm building (e.g., SI-065; Fig. 16). A preliminary evaluation of the surface pottery from these farmsteads indicates that they were generally in use from the Early/Middle Roman period to the late Byzantine period, with much of the material—indicating a peak in occupation—dating from the early and middle Byzantine periods. So far, only a few scattered pottery sherds from the Early Islamic period have been found in both the small villages and the farmsteads, suggesting that most were no longer in use by that period.
 
Winepresses. Wine-producing facilities occurred in the form of pressing basins with floors paved with stone slabs (e.g., SI-068; Figs. 16, 18). The basins were connected to collecting tanks sunk into the ground and coated with hydraulic mortar (e.g., SI-040; Fig. 19), in a manner similar to water cisterns found in the area. Interestingly, this clear evidence of wine production was found in areas of flat terrain rather than within the wadi systems, indicating that run-off irrigation was less desirable for the cultivation of vineyards. A surprising find was that most of these wine-production sites were found in isolated settings, away from the farmsteads.
 
Cisterns and Wells. Installations related to water use comprise cisterns and deep wells. Cisterns are often found at the foot of moderate slopes, which served as catchment areas. Wells dug to a depth of several meters to reach groundwater were found both within the city and its hinterland. A particularly well-preserved example with a circular shaft (BR-004; interior diam. c. 2.3 m; Fig. 20), constructed of well-cut masonry, is located on the west bank of Nahal Besor. The beginnings of arches used as a substructure for cover slabs were detected at the upper preserved edge of the well shaft. Since the aquifer is located 2–3 m below the wadi bottom, this well might have originally reached a depth of 12–14 m. Some of the cisterns and wells registered in the survey seemed to be associated with farms, while others were found in isolated settings, mostly in the areas of relatively flat terrain away from the wadis. These wells were likely employed primarily for irrigation.
 
Terrace Walls. Such walls, naturally restricted to the wadis, were found for example in the southern part of Wadi Umm Ghasuma and in the western part of another nearby unnamed wadi, dubbed Nahal Irus (Nahal Iris; for the mass occurrence of the Negev Iris in spring). They were also found along Nahal Besor along with many short segments of boundary walls, especially where the slope of its banks was less steep. These features are oriented parallel to the course of the wadi and subdivide the entire slope into multiple terraced agricultural areas, extending from the riverbed to the upper edges of the wadi (Figs. 21, 22). Several of these terraced areas were clearly associated with farm buildings (e.g., SI-025 and SI-044). Some of the terraced plots are located at an elevation of up to 10 m above the wadi bed and therefore, their irrigation cannot have been provided by flood water from the wadi. It is possible that these terraces were irrigated using well water obtained at the upper edge of the wadi slopes, taking advantage of the natural gradient. This possibility is indicated by irrigation channels documented in some of the terraced fields, with corresponding water outlets in the terrace walls.
 
Boundary Walls. Dry-built walls extending over appreciable distances were documented in the areas of relatively flat topography away from the wadis, subdividing the terrain into fields of various sizes (Figs. 12, 21, 23). These walls appear to have demarcated property boundaries and may have provided some protection from the damage caused by sandstorms and animals. The boundary walls form an extensive net-like system, extending from the outskirts of the city into the surrounding terrain across much of the survey area, with no indication of any attempt at orderly parceling. Gaps in the boundary walls appear to represent passageways and roads. These walls reinforce the possibility that large swaths of the flat terrain along the wadis were part of a system of intensive exploitation of the landscape for agriculture. Evidence of extensive manuring, based on high densities of surface pottery sherds found across the survey area, further reinforces this interpretation. Dense concentrations of small sherds were observed in nearly all areas not covered by sand, including areas located at considerable distance from the settlements. Collection of 66,000 of these surface sherds in six survey areas (TA-001–TA-006; 100 × 100 m each; Fig. 12) revealed an average density of 46 pieces per sq m, in some places reaching values close to 200 pieces per sq m, far above the typical densities found within settlements in the survey area. These sherds represent a time span from the first to the seventh centuries CE, with a clear dominance of sherds dating from the early and middle Byzantine periods. Systematic fertilization appears to have been conducted with organic urban waste also containing appreciable quantities of pottery sherds, unintentionally transported to the fields.
 
It is suggested, based on the results of this survey, that large parts of Elusa’s hinterland beyond the wadi systems were under intensive cultivation—particularly viticulture—utilizing groundwater for irrigation (for a more extensive discussion, see Heinzelmann et al. 2019/20). It should also be taken into consideration that grapevines are able to grow deep roots, to depths of over 8–10 m, especially in the local sandy soil, and could have reached groundwater, quite possibly limiting the need for artificial irrigation during the first year after planting (Amnon Bustan, pers. comm.). This survey uncovered ample evidence of the complex agricultural system extending across Elusa’s hinterland. It comprised, in addition to the well-known system of run-off farming, previously undocumented forms of agriculture, including small-scale horticulture on the city periphery and wine cultivation using deep wells and intensive manuring on the flatlands outside the wadis.