An Early Iron Age Moat in Jerusalem between the Ophel and the Southeastern Ridge/City of David

Abstract

Excavations on Jerusalem’s Southeastern Ridge, in the GivꜤati Parking Lot excavations, have exposed a man-made ditch, some 30 m wide and at least 6 m in depth, close to the hill’s summit. This paper presents the technical features of the ditch, including the rock scarps that delineate it from the north and south, as well as an enigmatic installation composed of a series of intertwined channels. Through comparison with data from previous excavations farther east, it is clear that this ditch traversed the entire width of the ridge, creating a disconnect between the Southeastern Ridge and the Ophel. Although the initial cutting of the ditch cannot be determined, it is clear that by the Late Iron IIA, the ditch served as a moat between the acropolis—possibly including the Temple Mount—and the city. This continued until the Late Hellenistic period, when construction activities backfilled the ditch, leading to its disappearance from the cityscape.

Keywords:

• Jerusalem
• Iron Age
• Moat
• City of David

Introduction

Ancient Jerusalem holds many secrets and unresolved mysteries regarding its location, topography and fortifications (Reich 2011; 2021; Finkelstein, Koch and Lipschits 2011; Geva and De Groot 2017). It seems that the least understood aspect of the city is the nature of the physical relationship between the Southeastern Ridge—the ancient core of Jerusalem—and the Ophel and the Temple Mount farther to the north (Fig. 1).¹ According to the biblical tradition, Solomon was the first to construct a sanctuary on the Temple Mount (2 Sam 24:18–25; 1 Kings 5:15–20). Many scholars accepted this account as historical fact, delineating the northern edge of the pre-Solomonic settlement south of the Ophel (Kenyon 1974: 90–91; E. Mazar 2007; 2009). Material remains, however, show that the Ophel was built at least as early as the Iron I (Winderbaum 2022; but cf. Finkelstein 2022). This brings to the fore the question of the city’s northern boundary and defences in the period prior to the northern expansion of the city, when the Temple Mount was incorporated into its borders.² While the steep slopes to the east, south and west form a natural divide between the Southeastern Ridge and its surroundings, the Temple Mount towers above the Ophel, rendering the city vulnerable to attacks from the north. How, then, was the city defended during the Middle Bronze, Late Bronze and Iron I Ages? Furthermore, what was the nature of the urban planning of the city after the Iron I, when the Ophel was used for the construction of massive public buildings? Was the Ophel connected to the main settlement on the Southeastern Ridge or was there some kind of a divide? Kenyon addressed these questions (1974: 91), stating that:

Fig. 1: Map of ancient Jerusalem marking the location of the GivꜤati Parking Lot excavations, the Ophel and other features mentioned in the article (drawing by Itamar Ben-Ezra)

At all periods, the position of the north wall of Jerusalem was arbitrary, in the sense that it was not dictated by physical features. The ascent of the eastern ridge, Ophel, towards the main ridge that forms the backbone of Palestine, is gradual, and no feature has been found that would help the defense of the southern part of the ridge, which was vital to the early town because of the water-supply.

This paper presents the recent discovery of an intentional rock-cut ditch, the first clear evidence of a divide separating the Southeastern Ridge from the Ophel and the Temple Mount to its north. The ditch, uncovered in the renewed archaeological excavations of the GivꜤati Parking Lot (Shalev et al. 2020; 2022a; 2022b), joins earlier finds in Kenyon’s excavations of Areas H, R and S, farther to the east, and Avner’s (2008) excavations, north of the modern visitors’ centre. After presenting a physical and contextual description of the ditch, we explore possible explanations for its date and function.

History of Research

Over the past 150 years, Jerusalem has undergone extensive excavations by numerous teams using a variety of methods and approaches. Nevertheless, fundamental questions regarding the location, size and nature of the city during its early history remain unresolved. One of the reasons for this is that the city was never spatially excavated in a systematic manner. While some parts of the city were extensively excavated—for example, the slopes surrounding the spring—certain areas (e.g., the Temple Mount)—have not been archaeologically accessible, while others, such as the eastern slopes of the Central Valley, have only recently been explored (Reich 2021). This has left many archaeological ‘blind spots’ that limit the accuracy of any comprehensive reconstructions of the city during the Bronze and Iron Ages.³ One such ‘blind spot’ is the northern edge of the Southeastern Ridge. This part of the site was covered by modern construction early in the 20th century CE; consequently, very few excavations were conducted here, and even fewer were published. Macalister and Duncan (1926) excavated slightly north of the Stepped Stone Structure in their Field 5. They describe a rock scarp (labelled A in their Plan 1), ca. 25 m long, stretching east–west across the western side of the ridge. They documented the height of the summit at 697.67 m asl, but unfortunately do not provide the bedrock level at the bottom of the scarp. They exposed a wall that crowned the top of the scarp, which may have functioned as a defence towards the north. Nevertheless, they admit that this reconstruction is problematic, as the scarp itself is south, rather than north, of the wall.

Kenyon’s excavations marked a true opportunity to understand the northern edge of the ridge, with three excavation areas spread along a north–south axis: Area S in the Ophel, Area H on the summit of the Southeastern Ridge, and Area R located between the two (Fig. 2). Not surprisingly, the excavations in these areas were utilised to discuss the issue of a defensive wall to the north of the Southeastern Ridge and its function in light of the higher topography of the Ophel and Temple Mount. Kenyon reconstructed the northern fortifications as a wall, positioned on the southern edge of a natural valley that protected it from the north (Kenyon 1964: 7–17; 1966: 76–77, 82–83; 1967: 66–67). According to her measurements, the valley dipped from the level of 701 m asl to the south, in Area H, to 695 m asl in Area R1, rising back to 702 m asl in Area S.

Fig. 2: The Southeastern Ridge, indicating all the excavation areas discussed (drawing by Vadim Assman, courtesy of the GivꜤati Parking Lot Expedition)

These figures were reevaluated by both Steiner (2001) and Prag (2017), each of whom reached a different conclusion. Steiner (2001: 27) published several architectural remains that she argues date to the Middle Bronze Age—most prominently Wall 51 and Structure J found in the northern squares of Area H. She suggested interpreting these walls as part of a possible tower and even a gate standing at the northern edge of the city. Steiner noted that the elevation on which these walls stood is 700.35–698.40 m asl. At the time of Steiner’s publication, Area R remained unpublished, and she therefore relied upon the preliminary notes by Kenyon, who stated that the rock level in Area R dropped to 695.50 m asl in Square RII, and even to 692.0 m asl in Square R1 farther to the east. Steiner concluded that during the Middle Bronze Age, natural gullies, created by water running eastward to the Kidron and westward to the Central Valley, served to form a valley that was used as a defence for the walls of the city (Fig. 3). In her reconstruction, she suggested that Wall 3, the city’s eastern fortification wall, climbed westward towards Wall 51, thus forming the northern fortification wall of the city during the Middle Bronze Age, and that the natural gully to the north turned the wall into a more prominent feature and was probably the reason for its exact positioning.

Fig. 3: Steiner’s reconstruction of the northern defensive valley (after Steiner 2001: Fig. 3.14; courtesy of Margreet Steiner)

The later publication of Area R also addressed Kenyon’s suggested northern valley (Prag 2017). Prag concluded that while the rock surface in Area H does appear to be at 698/700 m asl, the distance of ca. 35 m between Areas H and R is quite large and therefore a drop of 3–5 m over this distance is not that dramatic. Furthermore, she questioned the possibility of reconstructing the natural landscape, as extensive quarrying of the hill occurred in later periods (ibid.: 211–212). Finally, she pointed out that the height of 692 m asl documented in the east of Area R could very well be related to the fall of the slope eastward into the Kidron Valley, which is steep and dramatic.

Later excavations revealed additional data concerning Kenyon’s finds. Although Area R was covered by modern construction and was therefore unapproachable (Fig. 2), Area H was re-excavated by Eilat Mazar, who provided a terminus post quem for Wall 51 in the Iron Age, based on the pottery found in the soil layer beneath it. She therefore included the wall in the reconstruction of her Large Stone Structure (Mazar 2009). In 2003, Rina Avner excavated a trench, ca. 13.5 m long and 3.5 m wide, north of Kenyon’s Area H (Fig. 2). In her final report, she describes finds dating to the Iron II, the Early Roman and the Early Islamic periods, providing measurements for the bedrock, sloping from 697 m asl in the south to 696 m asl, some 7 m to the north (Avner 2008). While Avner did not link her finds to those from Kenyon’s and Mazar’s excavations, she did publish the dig promptly, enabling the integration of her finds into a broader picture. In the following section, we present the finds from the excavations in the eastern part of the GivꜤati Parking Lot (Area 70; Figs. 1, 2 and 4) and then return to the excavations described above, in order to reevaluate the possibility for a natural or a man-made feature bordering the Southeastern Ridge from the north.

Fig. 4: Aerial view of the GivꜤati Parking Lot excavations, with the location of the four cross-sections indicated (courtesy of the GivꜤati Parking Lot Expedition and the City of David Archive)

The new finds from the GivꜤati Parking Lot excavations

Over the past two decades, extensive excavations along the western slope of the Southeastern Ridge have contributed greatly to our understanding of the city’s development on the slopes of the Central Valley. Most relevant to this article are the excavations in the GivꜤati Parking Lot, exposing the summit of the Southeastern Ridge at the point where it begins to slope westward towards the Central Valley and merges with the Ophel to its north (Ben-Ami 2013; Ben-Ami and Tchekhanovets 2019). In 2017, Tel Aviv University and the Israel Antiquities Authority renewed the excavations in the GivꜤati Parking Lot, focusing on remains dating to the Iron Age and the Persian and Hellenistic periods (Shalev et al. 2020; 2022a). Ten areas were selected for excavation, two of which (Areas 70 and 10) create a cross-section extending more than 30 m and exposing the bedrock as it slopes down from east to west (Fig. 5, Section 2–2; Shalev et al. 2020: Fig. 5). This cross-section joins another cross-section to its south, excavated by Ben-Ami and Tchekhanovets in their Areas M1 and M4 (Fig. 5, Section 3–3; and see Ben-Ami 2013: 7). Recently, a northern section was excavated (Fig. 5, Section 1–1). In the southern and northern sections, the level of the natural bedrock on the easternmost edge, i.e., the hill summit, is 700–701 m asl. Surprisingly, in Area 70, the bedrock plummeted. Only after removal of several metres of occupation remains dating to the Byzantine, Roman and Hellenistic periods was bedrock reached at a height of ca. 692.30 m asl—approximately 9 m lower than expected.

Fig. 5: Sections 1–1, 2–2 and 3–3, showing the artificially shaped rock steps as the bedrock slopes from east to west (courtesy of the GivꜤati Parking Lot Expedition)

Two upright rock-cut scarps delineate the unusual drop in the bedrock from the north and south, with 35 m between them. The scarps therefore define a 35 m wide ditch, likely continuing to the east beyond the borders of the current excavation. While the two rock-cut scarps at the sides of the ditch are clearly man-made, they differ somewhat in nature: the southern scarp is cut vertically while the northern one is stepped (Figs. 6–7).

Fig. 6: a) The southern scarp, looking west; b) the northern scarp and Installation 7744, looking north (photos by Erik Marmor, courtesy of the City of David Archive)

Fig. 7: Section 4–4 (drawn by Vadim Assman, courtesy of the GivꜤati Parking Lot Expedition)

The upper part of the southern scarp, reaching an upper elevation of ca. 700 m asl, was first exposed by Ben-Ami and Tchekhanovets in their Area M4, where they documented the walls of the so-called ‘Akra’ (Ben-Ami and Tchekhanovets 2016; Zilberstein 2021). The renewed excavations continued work against the southern face of the scarp, exposing another roughly 7 m of the vertical rock, reaching the level of the rock at 691.25 m asl (Fig. 6a). This means that the scarp is ca. 9.5 m high. The scarp is clearly man-made, as such steep vertical steps do not develop naturally in the Meleke rock formations typical of Jerusalem (Gill 1996).

The northern rock scarp, located 30 m to the north, was revealed by Ben-Ami and Tchekhanovets at a height of 701 m asl in Area M3. They suggested that it was a quarry dating to the Roman period. The height of the scarp and its true nature only became clear farther to the south in the excavations of Area 70 (Fig. 6b). This scarp, also clearly a manmade feature, is carved in steps and is not as steep as the southern one. A clear scarp is seen at the northeastern edge of Area 70, where it stands ca. 3 m steep (from 700 m asl at the top to 697 m asl at the bottom). A 3 m wide rock tread is located below the scarp. This tread was used for the cutting of Installation 7744 (Figs. 7 and 8), a unique feature composed of five rock-cut channels (see below). The rock level drops here to the south and west. To the south, the step is 3 m high. The rock surface slopes farther to the south as it extends beyond the excavation boundary, reaching the level of 693 m asl at the lowest exposed point (Fig. 5, Section 2–2; Fig. 7). To the west, we exposed a 2 m high step that leads to another tread at a height of 694 m asl. To the southwest, the rock seems to drop once again, although the ditch is difficult to follow due to the Late Roman construction in this area.

Fig. 8: Installation 7744 (photo by Erik Marmor, courtesy of the City of David Archive)

The dating of the cutting of the ditch can be determined using relative stratigraphic data—namely, architectural features that relate to the scarp or were cut in the rock/built into the ditch. As the ditch was excavated in two distinct sections (one on the southern end and one on the northern edge, with the middle left unexposed), we have chosen to present data from the two areas separately.

Facing the southern scarp, numerous features dating to many periods may be discerned—most notably, massive walls dating to the Late Hellenistic and Early Roman periods (Shalev et al. 2019). The most prominent feature encountered is Wall 7009, a massive stone wall oriented north–south, built into the deep space created as a result of the quarrying and approaching the face of the scarp from the north (Fig. 6a). Ben-Ami and Tchekhanovets identified this wall as the inner face of the wall of the Akra and dated it to the Early Hellenistic period (2nd century BCE; Ben-Ami and Tchekhanovets 2016). Our excavations east of the wall clearly show that it was originally built as an independent wall based on the bedrock (after it had been lowered). In the upper parts, accumulation layers dating to the Persian period approached the wall (early and late; Freud and Shalev 2023). Under these layers, we exposed the foundation trench of the wall; it is now clear that Wall 7009 was in use during the Persian period, but its construction date is still undetermined. Accordingly, the cutting of the scarp must predate the Persian period.

The date of the carving of the northern scarp may be established through the relationship between the ditch and Installation 7744, located on the rock tread below the scarp (Shalev et al. 2023; Fig. 8). Three main components form the installation: the rock scarp, which appears to have been smoothened by running liquid; a shallow channel (7831) running east–west, which apparently gathered the liquid that flowed down the scarp; and four channels, oriented north–south, which apparently distributed the liquid that accumulated in Channel 7831. It should be noted that only three of these channels clearly flow from north to south, leading the liquid into a lower rock tread. The westernmost channel is carved in such a way that may have led liquid from the south towards Channel 7831; alternatively, it may be unfinished. The other three channels vanish after they drop to the lowest exposed tread. These perplexing measurements suggest that the installation may never have been completed. It is also unclear where the liquid concentrated, and above all, the purpose and manner of this function remains enigmatic. One possibility would be to suggest that these channels are in fact cut-off channels used in a quarry. This possibility, however, is unlikely, given the finding of a second set of channels (Installation 7702), 12 m to the north and outside the ditch (Fig. 9). All these channels differ from the cut-off channel typical of the quarries: they are rounded and smoothened. Furthermore, the treads between the channels in Installation 7702 are too thin to be building stones. The two sets may be part of one larger installation or two adjacent installations with a similar function, and we assume that they were used for soaking as part of the process of production of some unknown product (Shalev et al. 2023).

Fig. 9: Installation 7702 (photo by Oscar Bejarano, courtesy of the GivꜤati Parking Lot Expedition)

The three eastern channels had a whitish surface (L7724), made of thick crushed lime, sealing them at their upper rims and extending to the rock scarp to the north (Fig. 10). The sealed stratified accumulation (L7741, L7742) includes a small assemblage of pottery sherds (Fig. 11). It includes types characteristic of the Early Iron IIA, such as the open cooking pot (Fig. 11:7); of the Late Iron IIA, such as the carinated bowls with inverted rim (Fig. 11:1–2); and of the Iron IIA–B transition, such as the cylindrical jar with thickened rim (Fig. 11:9). The sherd presented in Fig. 11:5 is a small fragment of a bowl with an elongated and outturned rim. Bowls with outfolded rims first appear in Jerusalem in the early Iron IIB (Uziel, Dan-Goor and Szanton 2019: 65 and Fig. 4:3:1–3) and continue to appear throughout the Iron IIB and mainly in the Iron IIC (Zimhoni 2004b: Fig. 26.3:16–18). The small size of the fragment makes it difficult to pinpoint the date of this bowl. The latest sherds that can be dated in the fill belong to the Late Iron IIA or the early Iron IIB, thus providing a terminus ante quem for the carving of Installation 7744.

Fig. 10: Section showing soil stratified accumulation L7742 and L7742 placed inside the channels and sealed by surface L7724, looking south (photo by Oscar Bejarano, courtesy of the GivꜤati Parking Lot Expedition)

Fig. 11: Pottery found in the soil fill sealing the channels and sealed by surface L7724 (drawing by Irina Lidsky-Reznikov; plate prepared by Yulia Gottlieb)

It is difficult to determine the stratigraphic relationship between Installation 7744 and the ditch: it is possible that the installation was carved at the same time as the ditch was cut or, alternatively, that it used the existing scarp and ditch. Since it appears that the installation was never finished and did not become operational, we prefer the second possibility.

Several additional remains may be useful in the dating of the ditch, although these are somewhat circumstantial in nature as they are located to its west. W7094 is a short wall on the western edge of the ditch; it is oriented north–south and built of large, flat hewn stones. The wall dates to the Iron IIA–B transition (late 9th–early 8th centuries BCE; Shalev et al. 2019: Fig. 5). It is unlikely to have been built prior to the cutting of the ditch, as it seems to intentionally block the ditch. Further northwest, Building 100 was built in a location that, like W7094, blocks the ditch, possibly putting it out of its original use.

The northern part of the ditch remained exposed until at least the Early Hellenistic period. Floor 7809 and Wall 7786, both dating to the Early Hellenistic period, were built on the bedrock on the northwestern side of the ditch. Somewhat later, during the second half of the 2nd century BCE, the ditch was sealed: its northern part was intentionally filled with earth, while its southern part was blocked by massive construction (Shalev et al. 2019).⁴

In light of the stratigraphic data presented above, it may be safely determined that the cutting of the ditch occurred no later than the Late Iron IIA. This is primarily evidenced by the fill sealing the channel installation, along with secondary data originating from the building activity that took place in the late 9th or early 8th century BCE in front of the ditch.

The cutting of the ditch may be related to other major landscaping projects that reshaped the natural slope towards the west. Sections 1–1 and 3–3 in Fig. 5 show a steep scarp, 5 m high, on the eastern slope of the summit. This man-made scarp separates the top of the ridge from the lower part of the slope to its west. The same scarp can also be seen to the south of the GivꜤati Parking Lot (Crowfoot and Fitzgerald 1929: 19–20), although it is missing in the location where the ditch crosses its path (Fig. 5, Section 2–2). If this is indeed a continuous scarp along the slope, it may be interpreted as an intentionally built defensive feature—an effort to form a vertical obstacle protecting the settlement from the west. The date of this scarp is still unclear, although it is abutted by Early Hellenistic walls, providing a terminus ante quem for the cutting of the scarp. However, the relationship between the ditch and the scarp remains enigmatic—they may have been cut together or separately, and if the latter, we do not have a clear understanding of which came first.

In sum, a man-made ditch, ca. 30 m in width and up to 9 m in depth, was uncovered in the current excavations (Figs. 7 and 12). It is delineated in the south by a steep scarp scaling the height of the ditch, while in the north, the scarp is stepped. The ditch continues eastward beyond the boundaries of the excavated area. To the west, the rock levels merge with the naturally lowered bedrock levels of the slope. To the south and north of the ditch, the bedrock was cut so that it forms a high step protecting the settlement from possible advances from the west.

Fig. 11: Pottery found in the soil fill sealing the channels and sealed by surface L7724

Evidence for continuation of the ditch eastward

Excavations in Area 70 have exposed what appears to be the western opening of the ditch extending to the eastern boundary of the excavations. It is difficult to determine the extent of the ditch to the east, beyond the current excavation area. The squares excavated by Avner (2008: Fig. 1) are located ca. 40 m east of the southern scarp. Avner recorded the bedrock level on the southern edge of the excavations as 697.50 m asl (Avner 2008: Sections 3–3 and 4–4). Further north, the level of the bedrock is ca. 695.50 m asl (ibid.: Section 1–1) and 696.30 m asl (ibid.: Section 2–2).

According to Steiner, the level of the bedrock in Area H, just south of Avner’s trench, is 700 m asl, indicating a 2.5 m decline, followed by an additional 2 m decline. This drop in the bedrock does not conform to the natural topography and is similar to the levels exposed on the southern side of the ditch in the current excavations. Even more striking are the undated rock-cut channels exposed in Avner’s excavation (Avner 2008: Fig. 2, Sections 1–1, 3–3 and 4–4), which seem to resemble the channel installation found within the ditch, below its northern scarp. The walls built above them date to the Early Roman period, providing a terminus ante quem for the cutting of the rock and the channels.⁵

The unexpectedly low bedrock levels also appear in Kenyon’s Area R, with one square located ca. 10 m northeast of the edge of Avner’s trench. Here the bedrock was recorded at 695 m asl, a level similar to that recorded by Avner. As one proceeds further east into what appears to be the slope towards Kidron Valley, the rock level drops to 692 m asl. In both these squares, the carved bedrock was built upon, but the dating of the earliest architectural features is not clear (Prag 2017). Ending the sequence is Area S, 30 m farther to the north, where the natural rock level was recorded at 702.00 m asl.

Considered in its entirety, the data from Kenyon’s and Avner’s excavations point to the continued presence of a ditch crossing their areas of excavation, along the same lines and displaying the same features as the ditch exposed in the GivꜤati Parking Lot (Fig. 2). From all the above, it seems that both Avner and Kenyon had been excavating within the ditch the western edge of which we have exposed in Area 70 (Fig. 12).

Fig. 12: Rock levels at the GivꜤati Parking Lot excavations and the western edge of the moat (courtesy of the GivꜤati Parking Lot Expedition and the City of David Archive)

Discussion

As noted above, Kenyon had already proposed that a valley separated the Southeastern Ridge and the Ophel, based on the bedrock levels in Areas H, R and S, but the evidence was inconclusive, as no rock scarp was recorded and no pictures of hewn rock were presented. Faced with these circumstances, Steiner reconstructed a valley based on very few patches of exposed rock surfaces, which allowed only a very basic outline. However, as detailed above, the results from Kenyon’s excavations, particularly Area R, are isolated and do not provide a broad and coherent picture, making the claim for such a valley easily refutable (e.g., Prag 2017). Similarly, the rock-cut ditch revealed in Area 70 could be interpreted as a local phenomenon. Only through the analysis of the results from the GivꜤati Parking Lot excavations, in conjunction with the evidence from Avner’s and Kenyon’s excavations, can conclusive evidence be provided for the presence of an east–west ditch, separating the Southeastern Ridge from the Ophel (Fig. 2).

Our calculated reconstruction is that the ditch was ca. 30 m wide and 6–9 m deep. Its northern face runs from the northern scarp in Area 70, passing just north of Kenyon’s Area R, while its southern face runs from the southern scarp toward and just north of Kenyon’s Area H. There is no doubt that this is a monumental feature that required the quarrying of ca. 13,000 m³ of Meleke rock formation⁶ and completely altered the landscape of Jerusalem for many generations. The decision to position the ditch here may have been influenced by natural gullies running eastward and westward, as suggested by Steiner, but was also likely influenced by the narrow nature of the ridge in this area, as well as the proximity to the Ophel. According to our calculations, the ditch had to be ca. 75 m long to span the entire width of the ridge.

A ditch of this nature would have functioned first and foremost as a physical obstacle, preventing direct movement along the top of the Southeastern Ridge from north to south and vice versa by artificially cutting it into two separate segments. The ditch was probably not the only physical obstacle in this part of the ridge. As noted above, the finds include a 5 m high scarp along the upper part of the western slope. It is possible that this scarp also served as a contemporaneous fortification, intended to defend the city from the west. However, its date is uncertain—it may have been cut earlier than the ditch—and the fact that only a small part of the scarp was exposed makes it impossible to determine its function with any certainty. The discussion below, therefore, refrains from integrating the scarp into a broader chrono-historical reconstruction.

Two possible interpretations may be given for the function of the ditch: it could have been an outer defensive moat at the edge of the city (north or south; see further below) or an inner barrier separating the acropolis from the city.⁷ Choosing between these options depends upon our understanding of the city’s location in each period when the ditch was in use. Furthermore, changes in the city’s layout could have altered the function—so that what may have begun as a defensive moat could have turned into a barrier between the various parts of the city. It should be borne in mind, therefore, that due to the uncertainty regarding the date of the ditch, any interpretation remains uncertain.

The evidence for the dating of the ditch presented above indicates that the ditch was already in use during the Late Iron IIA and remained open until the end of the Early Hellenistic period. The initial date of cutting cannot be determined at this stage, but it may be earlier than the Late Iron IIA. The discussion below begins with the more secure periods, followed by the periods for which we still have insufficient evidence.

The Iron IIA–Late Hellenistic period

A survey of the architectural remains shows that during the timespan of the Iron IIA–Late Hellenistic period, the ditch served as a barrier between two monumental complexes. To the north and northeast of the barrier are the remains exposed in the Ophel (Mazar and Mazar 1989; Mazar 2015; 2018; Winderbaum 2022). The earliest architectural remains known from this area date to the late Iron I and Early Iron IIA (Winderbaum 2022; cf. Finkelstein 2022 for a different date). These include small domestic structures: Buildings Ib and Ia (following Winderbaum’s terminology), with possible evidence for metalworking. Later in the Early Iron IIA, these were buried under a massive constructional fill that supported a large edifice (Winderbaum 2022: 167). Stratigraphically, the edifice includes Lower Buildings IIIa, IIIb and Wall IV, although these were not dated by pottery (and see Finkelstein 2022, who dates them to the Late Iron IIA or early Iron IIB). A third construction phase includes Building II, which cut and reused earlier elements. Wall IV continued to be used in this phase. The fourth construction phase includes a large administrative structure, built during the Early Iron IIB (Buildings IIIa and IIIb), and its extension to the east, which expanded the saddle of the Ophel farther to the south/southeast into the slopes of the Kidron Valley—thus changing the landscape of Jerusalem.⁸

South of the barrier, the Stepped Stone Structure stood during the Iron IIA. This constitutes a massive revetment built along the slope that supported the Large Stone Structure, a largely disputed public building (e.g., E. Mazar 2007; 2009; A. Mazar 2010; Finkelstein 2011). The dating of the Stepped Stone Structure has been discussed extensively. We share the commonly accepted view that it was constructed during the very late Iron I or the Early Iron IIA and continued to function in its original form into the Late wall oriented north–south, is bonded to the Stepped Stone Structure and is a remnant of the public building supported by the latter. E. Mazar claimed that Wall 20 was the eastern wall of the Large Stone Structure. While it is unclear whether Mazar’s reconstruction is plausible, the remains of Wall 20 and the Stepped Stone Structure provide evidence for the presence of a public building south of the barrier throughout the Iron IIA.⁹

A consideration of the architectural remains both north and south of the ditch from the Early Iron IIA onward supports the notion that the ditch functioned as an internal urban barrier, separating the acropolis located on the Ophel and possibly the Temple Mount from the city to its south. The ditch formed a separation between two areas that served for public construction: the complex of the Large Stone Structure and the Stepped Stone Structure to the south, and the Ophel complex to the north.

Rock-cut ditches are known from several central Late Iron IIA sites, such as Samaria (Finkelstein 2011), Jezreel (Ussishkin and Woodhead 1997) and Hazor (Yadin et al. 1989: 53; Finkelstein 2000: 117–118) to the west of the Jordan River, as well as Khirbet Atarus, east of the Jordan (Finkelstein and Lipschits 2010; Finkelstein, Lipschits and Sergi 2013). In all of these sites, these ditches should be defined as moats that were used as a component in a fortified edifice, distinct from the environs. The only example of an inner barrier is the ditch at Hazor. According to Yadin, it was 45 m wide and 10 m deep and served to separate the Stratum X–IX edifice from the rest of the upper tell (Yadin 1972: 138–139; Yadin et al. 1989: 53). Ben-Tor’s excavations in Area A5 exposed monumental subterranean halls built at this spot, which were dated to the Late Bronze Age (Ben-Tor et al. 2017: 54). It is unclear from the publication whether these halls negate the possibility of an earlier barrier/ moat, and the fill within the halls, which dates to the Iron IIB, casts doubt on the date of the halls (ibid.: 58). According to Finkelstein (2000; 2013), the cutting of these moats and barriers is associated with the expansion of the Omride kingdom and is a fossile directeur of Omride construction (Finkelstein 2000; 2013: 103–104). The finding of the monumental rock-cut barrier in Jerusalem that was already being reused close to the end of the Iron IIA joins other evidence that these moats should not be viewed as a chronological fossile directeur but instead, as an element within the repertoire of fortification systems, in use prior to the late 9th century BCE and adopted as a divider within the urban planning of cities in both the northern and southern kingdoms (and see Finkelstein and Kleiman 2019; Sergi 2023: 77–78, regarding the possible date of the moat and gate system at Hazor).

The Bronze Age

Scholarly research has pointed to the difficulties in understanding the northern defensive line, which led Kenyon, followed by Steiner, to assume that there was a natural valley that dictated the construction of a fortification wall at the northern edge of the city. They dated it to the Middle Bronze Age, associated with Wall 3, which surrounded the Southeastern Ridge from all sides. As it stands, the finds from the GivꜤati Parking Lot excavations can only prove that the ditch was open during the Late Iron IIA, although an earlier, Middle Bronze Age, date cannot be ruled out.

Theoretically, if the ditch is indeed earlier than the Late Iron IIA, it may be suggested that it was initially a fortification moat, as at that time it lay beyond the boundaries of the settlement (and see further below). The magnitude of this project matches the scale and magnitude of the spring fortifications attributed by most scholars to this period (Reich and Shukron 2010; 2021: 682–683; Reich 2018; Finkelstein 2023: 35; Regev et al. 2021: MB Phase III).¹⁰

While the location of the moat seems ideal to defend the town situated to its south, two issues pose a challenge to this interpretation. First, as mentioned above, there is no evidence thus far to support such an early dating. Second, recent scholarship has challenged the notion that the Middle Bronze Age city was fortified, except for the Spring Tower that protected a critical resource (Reich and Shukron 2010; Ben-Ami 2014; Ussishkin 2016; Regev et al. 2021). Kenyon’s Wall 3 does not appear to be a fortification wall, and nor does her Wall 50 (which Steiner related to her ‘defensive valley’), which seems to be much later and in any case is located ca. 10–15 m south of the ditch, making it impossible for the two to have functioned together. Furthermore, no fortification wall was found on the western side of the ridge prior to the Hellenistic period (Ben-Ami 2014; Ussishkin 2016). In fact, very few Middle or Late Bronze Age sherds were found in the GivꜤati Parking Lot excavations in general, and it is unclear whether the western slope of the Southeastern Ridge was part of the settlement during this era.

Another issue that impacts our understanding of the ditch relates to the ongoing debate regarding the location of the settlement during the Bronze Age and the history of the Temple Mount. Most scholars claim that the Bronze Age city centred around the spring, south of the ditch. However, Finkelstein, Koch and Lipschits (2011; and see Knauf 2000; Naʾaman 2023) have suggested that the ancient mound of Jerusalem was located on the Temple Mount, north of the ditch. According to this interpretation, the Middle Bronze settlement was centred around the mountain’s peak with an extension towards the spring, while during the Late Bronze Age it nucleated around the peak, with fewer activities in the vicinity of the spring. Our limited knowledge regarding building activities in Jerusalem during these periods makes this assumption tentative. Earlier finds from the Ophel include fragments of cuneiform tablets, scarabs and bullae dating to the Middle and Late Bronze Ages, found in secondary deposition in later fills (Mazar et al. 2010; 2014; Keel 2015). While these finds are unique and many relate to elite activities (Naʾaman 2023), it is unclear whether they point to the existence of an edifice standing next to or on top of the Temple Mount or whether they were brought with the soil from elsewhere. Nevertheless, if one accepts the suggestion that Jerusalem in the Middle and Late Bronze Ages was located on the Temple Mount, the location of the ditch precludes its interpretation as a defensive moat, as a wall located on the slope would have served as a sufficient defence.

Conclusions

The current paper presents the existence of a massive, intentional rock-cut ditch oriented west–east and separating the Ophel from the Southeastern Ridge. While there is no direct evidence for dating the hewing of the ditch, it was certainly in use prior to the Late Iron IIA–early Iron IIB, at which point it was reused for a different purpose. The ditch was cut either during or before the Late Iron IIA. Because of the uncertainty regarding its date, its initial intended function remains unknown. From the Late Iron IIA onward, it functioned as a barrier, restricting movement from the city to the Ophel and the Temple Mount. Viewing it as such entails that until the end of the Early Hellenistic period, the Ophel functioned as an acropolis, physically separated from the city.

The interpretation of the ditch as a pre-Iron Age fortification moat is probable and clearly appealing; thus far, however, it is based on circumstantial evidence. We therefore refrain from reaching any conclusion until further information is at hand. With regard to the Iron Age, the above discussion shows that the Iron IIA was a time of major building activities, which went hand in hand with massive landscaping projects. The Stepped Stone Structure, the reshaping of the western slope and the construction of the Ophel are all projects that completely altered the natural topography of ancient Jerusalem. The Stepped Stone Structure was constructed to allow massive construction to the south of the ridge’s summit. The Spring Tower was also built or rebuilt during this era. At the Ophel, the external ‘Tower’ of Building IIIa–IIIb allowed for the construction of a monumental structure against the slope to the Kidron Valley. Building 100 was constructed, while the natural slope towards the Central Valley was reshaped into wide steps. All of these projects may not have taken place simultaneously, but they are part of the same royal mindset that dramatically changed the urban landscape of Jerusalem and can be placed, generally speaking, in the formative moments of Iron Age Jerusalem—i.e., the end of the Iron I until the beginning of the Iron IIB.

Thus, Jerusalem of the Iron IIA–B was different than has generally been reconstructed. It included an acropolis to the north and a city on the lower, southeastern, ridge with a barrier separating the two.¹¹ During the Iron IIB–C, the city also included the Western Hill and the newly built quarters. A better dating of the moat and of the surrounding building projects may help gain a better understanding of the original function and political setting of the city.

Acknowledgements

The excavations at GivꜤati Parking Lot are conducted by the Israel Antiquities Authority and Tel Aviv University (Licence Nos. G11-18, G10-19, G11-20, G3-21, G1-22 and G7-23). The excavations are financed by the El-Ad Foundation. Y. Gadot and Y. Shalev serve as the excavation directors. D. Gelman, M. Zindel, M. Hemmed and O. Bejarano were the Area 70 supervisors. Other members of the team are: S. Bailey and M. Hacohen (assistant area supervisors) and V. Assman (surveyor). A. Reʾem, N. Sapir, Y. Baruch, O. Lipschits, K. Gur-Arieh, M. Ganon, U. Dasberg and M. Baruchi provided support in the orgenisation and logistics of the GivꜤati Parking Lot project. J. Uziel edited an earlier version of the article and offered invaluable comments. Our thanks are extended to them all.

Disclosure statement

The authors report that there are no conflicting interests to declare.

Additional information

Notes on contributors

Yuval Gadot

Yuval Gadot: The Sonia and Marco Nadler Institute of Archaeology, Tel Aviv University

Efrat Bocher

Efrat Bocher: The Martin (Szusz) Department of Land of Israel Studies and Archaeology, Bar-Ilan University, Ramat Gan, and the Ancient Jerusalem Research Center, Jerusalem; email: [email protected]

Liora Freud

Liora Freud: The Sonia and Marco Nadler Institute of Archaeology, Tel Aviv University; email: [email protected]

Yiftah Shalev

Yifat Shalev: Israel Antiquities Authority; email: [email protected]

Notes

1 The geographical terms used with regard to Jerusalem are complicated and generate confusion. This paper uses the terminology proposed by Kenyon that describes the various features objectively: the Southeastern Ridge is used to denote the hill commonly referred to as the City of David/Silwan; the Central Valley is used for the feature otherwise known as the Tyropoeon Valley; the Ophel is used here as a modern geographical term in relation to the topographical saddle connecting Mount Moriah and the Southeastern Ridge, without linking it to the location of the biblical Ophel (see Franklin 2014 and references therein).

2 Knauf (2000), Finkelstein, Koch and Lipschits (2011) and Naʾaman (2023) have argued that the ancient core was located on the Temple Mount and north of the Ophel, leaving the Southeastern Ridge outside the city. As further addressed below, we believe that the Southeastern Ridge is in fact the location of the ancient settlement (see, e.g., Geva and De Groot 2017; Gadot and Uziel 2017).

3 The debate about the size of the city during the Iron Age, which took place in the mid-20th century CE, is a case in point. While some scholars claimed that Jerusalem was restricted to the Southeastern Ridge since this is the area where actual finds were reported, others claimed that the Western Hill must have been part of the city (Avi-Yonah 1956: 157–160; Avigad 1983). Only the excavations below the destroyed houses of the Jewish Quarter following the 1967 war settled this debate (and see summary in Geva 2003: 505–522).

4 The sealing of the ditch and its significance for the political history of Jerusalem will be discussed in a future article.

5 Avner (2008) assumed that these were related to stone quarrying, but their function and date should now be reconsidered, in light of Installations 7744 and 7702.

6 We wish to thank one of the anonymous readers for this calculation.

7 Another possibility is that the ditch functioned as a siege ramp, like the one surrounding Tell eṣ-Ṣafi/Gath (Maeir and Gur-Arieh 2011). This possibility, however, seems less likely. For one thing, this ditch is not dug in earth fill like the one at Gath, and for another, there would be no real need to cut such a moat if the besieging army were located on the higher ground to the north.

8 Mazar and Mazar (1989) argued that this structure was a gate. However, since so little of the structure was unearthed, we prefer a more general definition of the building as a public structure (Winderbaum 2022: 159–160).

9 Wall 50, the northern wall of the Large Stone Structure according to E. Mazar, was originally discovered by Kenyon and was published by Steiner, who interpreted it as a fortification wall dating to the Middle Bronze Age (2001; and see above).

10 Note that Regev et al. (2017) suggested an alternative scenario in which the Spring Tower was built at the end of the 9th century BCE.

11 The proposition that the lack of architecture in Iron IIA Jerusalem is an indication for the nomadic nature of its elite, as suggested by Ben Yosef (2019), now appears to be unnecessary, as there is sufficient positive evidence for the use of architecture and landscaping projects for royal propaganda.