Archaeomagnetic Dating of the Outer Revetment Wall at Tel Lachish

Abstract

The fortifications of Lachish, a key site in archaeology of the Iron Age Southern Levant, are the focus of ongoing debate. The Outer Revetment Wall, encircling nearly the entire site, was traditionally associated with Levels IV–III and was thought to have been in use during the Assyrian campaign in 701 BCE. It has recently been suggested that it was built a millennium earlier. Here we present archaeomagnetic dating of a mudbrick tower incorporated in this wall, indicating that it was burnt during the Iron Age and was most likely built during this period. Combining archaeological, historical and archaeomagnetic data reveals the intense fire that occurred during the 701 BCE Assyrian siege. This fire could have been set by the people of Lachish, in a desperate attempt to damage the Assyrian siege engines or siege ramp, as depicted in the well-known Lachish relief, or by the Assyrians as part of their siege tactics.

Keywords:

• Archaeomagnetic dating
• Tel Lachish
• Outer Revetment Wall
• Mid-Slope Wall
• Siege ramp

Introduction

This paper focuses on the archaeomagnetic dating of the ‘Outer Revetment Wall’ at Tel Lachish, exposed by the British expedition in the 1930s. Tufnell (1953: 87) described this wall as ‘a stone and brick revetment, approximately 4 m. wide, encircling the mound at a distance of about 16.5 m. from the top scarp’. The British expedition dated the wall to the Iron Age and identified it with a wall depicted in the middle of the slope of Lachish in the famous relief from Sennacherib’s palace in Nineveh (Layard 1853: 125–129; Ussishkin 1982: 67–117). Their conclusions were supported by Ussishkin (2004c: 79) but were later challenged by Garfinkel et al. (2021b: 426–429; naming it ‘the Mid-Slope Wall’), who argued for a Middle Bronze Age date. Our study aims to resolve this debate through archaeomagnetic dating of burnt mudbricks of the ‘tower-buttress’ that had been incorporated into the Outer Revetment Wall in the southwestern corner of the mound.

The Outer Revetment Wall: history of research

Tel Lachish, located in the Shephelah, on the bank of Naḥal Lachish, has been excavated since the 1930s by several expeditions (for a recent summary, see Garfinkel et al. 2021b: 419–423). Starkey, who headed the British expedition, the first to excavate Lachish, mentioned a wall in the middle of the slope in a letter sent about a month before the excavations began (Garfinkel 2019: 294). He identified this wall with the lower fortification depicted in the relief and thus assumed that it had been in use during the 701 BCE Assyrian campaign. In order to establish the date of the wall, the British expedition began to excavate it on the northeastern side of the mound. After the first season of excavation, Starkey (1933: 197–198) reiterated the view that the wall had been in use in 701 BCE, but did not relate to its construction date. Tufnell (1953: 60), who published the finds of the British expedition, concurred, but raised the possibility that earlier wall segments could have been incorporated into the Outer Revetment Wall during its construction. The Iron Age dating of the wall was maintained, even though some finds in the northeastern side of the mound might have supported an earlier date (ibid.: 88). Eventually, the British expedition exposed the remaining top courses of the Outer Revetment Wall along almost the entire perimeter of the mound (Figs. 1–2) and attributed this wall to Levels IV–III (ibid.: 88–92). In the southwestern corner of the mound they partially unearthed a mudbrick tower incorporated within the wall (ibid.: 90).

Fig. 1: Aerial view of Tel Lachish from the north, photographed on July 18, 1934 (Ussishkin 2004b: 24 Fig. 2.3); note the excavated area along the Outer Revetment Wall on the northern and western slopes of the mound

Fig. 2: The plan of Tel Lachish according to the TAU expedition (based on Ussishkin 2004b: 34, Fig. 2.9), showing the location of the mudbrick wall of the tower-buttress sampled for this study (LC08); 1) outer gate; 2) Level IV–III inner gate; 3) the Outer Revetment Wall marked all around the mound, excluding the gate area and a gap in the northeastern corner of the mound; 4) the main city wall; 5) the siege ramp

The Tel Aviv University (TAU) expedition (1973–1994) further exposed this tower, which they termed ‘tower-buttress’, in Squares N/5–7 in Area R (Figs. 3–5; Ussishkin 2004a: 701–707). Ussishkin argued that the Outer Revetment Wall and the tower-buttress adjoined the main city wall at the southwestern corner of the mound and were thus part of a single large fortification system encompassing the city. In his view, this system was the strongest Iron Age fortification ever to be found in Israel and its construction marked the beginning of Level IV (Ussishkin 2004c: 78–80). The TAU expedition also examined small segments of the Outer Revetment Wall in Areas GE and S and concluded that this system had been in use during the destruction of Level III, which they dated to 701 BCE (and see the reconstruction of the town accordingly, Fig. 6; Ussishkin 2004c: 83–85). This absolute dating is based on the correlation between the historical sources regarding the 701 BCE campaign and finds from Level III. In Area R alone these finds include the Assyrian siege ramp (Ussishkin 2004a: 707–723), 859 arrowheads, an iron chain, iron scales and perforated stones (Fig. 7a; ibid.: 734–739). Among the arrowheads, 104 were found at the foot of the tower-buttress and 248 were found on its parapet (Gottlieb 2004: 1954–1955).

Fig. 3: The Assyrian siege ramp and Area R in 1983, view from the southwest (based on Ussishkin 2004a: 705, Fig. 13.10); the location of the Outer Revetment Wall is marked by a dashed white line

Fig. 4: The southwestern corner of Tel Lachish and the excavations at Area R of the TAU expedition (based on Ussishkin 2004a: 702, Fig. 13.6); the location of LC08 is marked; note the Outer Revetment Wall which abuts the tower-buttress on both sides

Fig. 5: A longitudinal section through the southwestern corner of the mound, extending along the northwestern sides of Squares A–Z/6 in Area R of the TAU expedition (based on Ussishkin 2004a: 703, Fig. 13.7); 1) the siege ramp with two alternative reconstructions; 2) the tower-buttress in the Outer Revetment Wall; 3) the siege ramp, second stage (covering the tower-buttress); 4) the Main City Wall; 5) the counter-ramp; 6) the additional rampart and the Level II city wall

Fig. 6: Suggested reconstruction of Lachish Level III according to the TAU expedition (Ussishkin 2004c: 85, Fig. 3.1; drawing by Judith Dekel); note the Outer Revetment Wall encircling the mound at mid-slope

Fig. 7: The location of LC08 on the outer face of the tower-buttress; a) the tower-buttress exposed in 1985 (Ussishkin 2004a: 710, Fig. 13.14); note that one person is standing at the foot of the tower and another is sitting on the balcony on top of the tower; the estimated location of LC08 is marked; b) the location of LC08 during our archaeomagnetic sampling

The accepted Iron Age date for the Outer Revetment Wall and its stratigraphic attribution to Levels IV–III were challenged by the Hebrew University of Jerusalem (HUJI) expedition (2013–2017). The excavators (Garfinkel et al. 2021b: 426–429) suggested that it had been built in the MB II, roughly a millennium earlier than the accepted date. The dramatic change in the suggested date of the Outer Revetment Wall is based on the finds excavated in Area BB, located in the northeastern corner of the mound. Specifically, it was claimed that a mudbrick fortress was built adjacent to the Outer Revetment Wall after the wall had been built. This fortress is dated to the MB III by pottery, scarabs and bullae, and its destruction is dated by radiocarbon to the 16th–15th centuries BCE (Garfinkel et al. 2019: 706–707). The dating of the tower-buttress is not discussed in the publications of the HUJI expedition but since it is part of the Outer Revetment Wall, their conclusions may suggest that the tower-buttress too was constructed during the MB II.

Ussishkin has recently rejected the MB II dating of the Outer Revetment Wall (Ussishkin 2019: 305–307; 2023: 97–102). His rejection is based mainly on the connection between the Outer Revetment Wall and the main city wall in the southwestern corner of the mound, on Iron Age pottery found beside the wall by the British and TAU expeditions and on the stratigraphic relation between the wall and the MB II glacis (Ussishkin 2023: 98–99).

Archaeomagnetism: methods and materials

Archaeomagnetism in the Levant: history of research

The field of palaeomagnetism focuses on reconstruction of Earth’s magnetic field—the intensity and direction of which are constantly changing—in periods that predate direct measurement. Palaeomagnetism reconstructs the magnetic field as preserved in various materials, mainly geological, such as volcanic rocks and marine sediments. Archaeomagnetism, along similar lines, focuses on reconstructing the ancient field as recorded in burnt archaeological materials, such as pottery vessels and ovens.

As is evident from direct measurements of the field in modern times, the geomagnetic field is sometimes characterised by local anomalies. Therefore, a reliable reconstruction of the ancient geomagnetic field must be achieved in different regions on earth separately (we suggest using a 500–600 km radius; see Shaar et al. 2022: 9–10). Once a large and well-dated archaeomagnetic database for a certain region and period is achieved (by means of ceramic typology, historical data and radiocarbon measurements), it can be used as a reliable dating tool.

The growing archaeomagnetic database of the Levant, particularly during the Iron Age (Shaar et al. 2011; Shaar et al. 2016; Ben-Yosef et al. 2017; Shaar et al. 2018; Shaar et al. 2020; Vaknin et al. 2020; Shaar et al. 2022; Vaknin et al. 2022), has already been used for archaeomagnetic dating (e.g., Ben-Yosef, Tauxe and Levy 2010; Peters, Tauxe and Ben-Yosef 2017; Shahack-Gross et al. 2018; Vaknin et al. 2022). In addition to its chronological application, archaeomagnetism is a useful tool for the reconstruction of site formation processes related to burnt materials. For instance, it can identify burnt archaeological materials, determine whether they cooled down in situ and shed light on their firing temperature (Shahack-Gross et al. 2018; Vaknin et al. 2020; Vaknin et al. 2022; Vaknin et al. 2023).

Archaeomagnetic sampling

We used the name LC08¹ for all the burnt bricks from the tower-buttress that were sampled for all archaeomagnetic experiments (Fig. 7). Three bricks were sampled without field orientation² for archaeointensity experiments only (LC08A–C). Ten additional samples (LC08D–M) were sampled as oriented samples (Fig. 8). Each of the 13 samples represents the outmost several centimetres of a different brick, with the exception of samples D and E, which were taken from two sides of the same brick (Figs. 8, 9a). Before sampling the oriented samples, we polished the outer part of the bricks in order to create flat surfaces. Due to cavities left by chaff which had been added to the mudbrick matrix during construction, we applied a very thin layer of plaster of Paris in order to create flat surfaces (Figs. 8–9). We then marked horizontal lines on these surfaces (Fig. 9) and measured their field orientation, using the method described in Vaknin et al. (2022: 6).

Fig. 8: The location of the different samples for palaeointensity and archaeomagnetic directions (LC08); Samples A–C are unoriented samples taken for palaeointensity only; all others were sampled as oriented samples and thus the flat surfaces and the use of plaster of Paris can be seen

Fig. 9: Representative examples of sampling and sample preparation; a) Samples D, E and G; note that D and E were sampled from two sides of the same brick; b) Samples J and L; c) closeup of Sample I; d–f ) Sample K after it was initially cut in the lab for archaeomagnetic direction experiments (d–f at 1:4 scale)

Archaeomagnetic experiments

All archaeomagnetic experiments were carried out at the magnetically shielded palaeomagnetic laboratory at the Institute of Earth Sciences, The Hebrew University of Jerusalem. We cut the ten oriented samples in the laboratory into smaller specimens (Fig. 9d–f) and glued every specimen in a non-magnetic palaeomagnetic sampling box, as described in Vaknin et al. (2022: 6). All the oriented specimens underwent Alternating Field (AF) demagnetisation, and the mean direction of every sample was calculated according to the protocol described in Vaknin et al. (2020).

For archaeointensity we measured 24 unoriented specimens, three from each of the following bricks: LC08A, B, C, D, F, J, K and L. We carried out the archaeointensity experiments, cooling rate and anisotropy corrections following the protocol described in Shaar et al. (2020: 4–7). We analysed the archaeointensity results and used the acceptance criteria as in the recently published Levantine archaeomagnetic curve, LAC.v.1.0 (Shaar et al. 2022; Vaknin et al. 2022). The archaeomagnetic age of LC08 was obtained using the AH-RJMCMC algorithm (Livermore et al. 2018), which employs all data in the LAC.v.1.0 from Israel and Syria (Shaar et al. 2022: 4–5; Vaknin et al. 2022: 6) combined with new data from LAC.v.1.1 (Hassul et al., forthcoming) as its prior information. Every datum consists of the archaeointensity results and an age range based on the available archaeological dating methods. Archaeomagnetic results are not considered part of the prior age ranges in order to prevent circular reasoning. Due to the debate regarding the Outer Revetment Wall, we assigned a very wide age range for LC08: 1800–680 BCE. The output of the AH-RJMCMC technique is an age posterior distribution from which we extract the 68.2% and 95.4% credible intervals.

Thermomagnetic curves

In order to shed light on the ancient firing temperatures of the sampled bricks LC08, we measured thermomagnetic curves of magnetic susceptibility in repeated heating and cooling cycles at progressively elevated peak temperatures from 100⁰ C to 700⁰ C, in 100⁰ C steps (as described in Vaknin et al. 2022: 6).

Results

Archaeomagnetic directions

In total, 73 specimens from ten samples (six to nine specimens from every sample) underwent demagnetisation experiments. Each of the specimens yielded a strong and unified magnetic signal, indicating that it had been recorded during one heating event (Fig. 10). All 73 specimens met the criteria described in Vaknin et al. (2020: 7) and yielded vectors scattered close to the average direction of the normal geomagnetic field in Israel. Each of the ten samples showed a very tight cluster with a precision parameter k > 640 and a 95% confidence cone α₉₅ ≤ 2.6° (Table 1). Unlike the tight clusters of the results from every sample, there is a significant scatter between the average direction results of the various samples (Fig. 11), which is indicative of slight movements that occurred after the bricks had cooled down.

Fig. 10: Representative result from the Alternating Field (AF) demagnetisation experiment of one specimen displayed as a Zijderveld diagram (Zijderveld 1967); since the magnetic signal is a three-dimensional vector, it is displayed by two lines (red and blue), each representing the projection of the vector on a different plane; the original magnetisation is gradually erased by an increasingly strong magnetic field (4.0mT, 8.0mT, etc.); it displays a strong magnetic vector and straight lines converging to the origin indicating stable magnetisation

Fig. 11: Direction results; mean directions of the different samples of LC08 are plotted as coloured symbols on an equal area projection; their α₉₅ confidence cones are plotted as coloured circles; the average direction and the α₉₅ cone of the samples from the gate of Lachish Level III are plotted in black, representing the direction of the geomagnetic field in the Levant in 701 BCE, as published in Vaknin et al. 2022: 2

Table 1: Direction results

Archaeointensity and archaeomagnetic dating

All 24 archaeointensity specimens prepared from eight different bricks in LC08 met the acceptance criteria, which is an exceptional success rate (Fig. 12). The mean archaeointensity and the standard deviation of LC08 were calculated using the method described in Shaar et al. (2020) and are marked in blue in Fig. 13a (Table 2). These are presented along with previously published data from Israel and Syria which were used for the LAC.v.1.0 and the LAC.v.1.1. The estimated archaeointensity curve (grey line) and its error envelope (light grey area) based on the LAC.v.1.1 are shown in Fig. 13a, along with the intensity of the field in 1980 CE. The result of the archaeomagnetic dating of LC08 is presented in Fig. 13b.

Fig. 12: Representative results of paleointensity experiments all carried out on a specific specimen: a) a representative Arai plot, where blue circles, red circles and triangles represent IZ steps, ZI steps and pTRM checks, respectively; the inset displays the Zijderveld (1967) plot of all the demagnetisation steps; the nearly ideal behaviour is characterised by a straight Zijderveld plot converging to the origin (inset), a nearly linear Arai plot and pTRM checks (triangles) which overlap the infield data points (red circles); the Zijderveld plot (inset) represents the projection of the vector on two different planes; b) the magnetisation after the different steps of the paleointensity experiment; the Y-axis represents the magnetisation normalised to the initial magnetisation of the specimen; the X-axis represents the temperature steps; the blue graph represents the initial magnetisation recorded in the past; it starts at 1.0 by definition and decreases gradually; note that the magnetisation is nearly entirely erased only at 600 °C; the red graph represents the magnetisation recorded in the lab, starting at 0.0 by definition and rising with the rising temperature steps; c) a representative result of a cooling rate correction experiment

Fig. 13: Archaeomagnetic dating of LC08; a) the archaeomagnetic curve in the Levant based on all the data of LAC.v.1.0 and LAC.v.1.1 (see text for details); the blue line and light blue area represent the mean PI of LC08 and its standard deviation, accordingly; they are presented along all the prior age range of LC08 (1800–680 BCE); the intensity of the geomagnetic field at Lachish in 1980 CE (83 ZAm^2) is marked by a dashed green line; b) the result of the archaeomagnetic dating of LC08: the prior age range of LC08 which was used by the AH-RJMCMC algorithm (Livermore et al. 2018) is represented as a uniform probability density function (grey background); the posterior age probability distribution for LC08 is displayed in blue

Table 2: Palaeointensity results

Thermomagnetic curves

We produced thermomagnetic curves for three specimens (Fig. 14) from different bricks (LC08D, J and L).

Fig. 14: Thermomagnetic curves; we measured the magnetic susceptibility in repeated heating cycles at progressively elevated peak temperatures in the range of 100–700 °C (each colour represents one cycle); the solid lines represent the heating process whereas the dashed lines represent the cooling process; the thermomagnetic curve of LC08D is nearly reversible (a), demonstrating little alteration, which is indicative of stability of the magnetic minerals up to 700 °C; the thermomagnetic curves of LC08J (b) and LC08L (c) are nearly reversible up to 600 °C, but significant alteration was observed after heating these specimens to 700 °C (red dashed lines); only in the case of LC08D was a small drop in the susceptibility observed at ca. 350 °C; the main drop in the susceptibility in all three specimens occurred between ca. 580 °C to 620 °C, probably indicating the presence of magnetite

Discussion

Site formation: the firing of the bricks of the tower-buttress

All the magnetic signals measured by us within every specimen and within every sample were unified and pointed roughly to the north and down, generally in the direction of the average normal field in Israel. This indicates that these magnetic signals were recorded when these bricks were fired in situ, probably during a single heating and cooling event. This observation strongly supports the possibility that the tower-buttress was built of sun-dried bricks that were later fired in situ (Ussishkin 2004a: 732), and not from pre-fired bricks.³

Theoretically, the direction results do not rule out the possibility that these bricks were pre-fired and then fired again within the wall. This possibility, however, seems very unlikely: if the bricks had been pre-fired in a kiln and then heated again to lower temperatures, we would expect at least some of them to still have a partial magnetic signal recorded in the kiln. This theoretical situation would result in two different components of the magnetic vector within specimens and a large scatter within samples. This is not the case in any of the 73 demagnetised specimens (Fig. 10) or the ten samples (Table 1). In any case, the magnetic signal we measured was recorded in situ, an observation that bears major chronological implications. This observation indicates that the archaeomagnetic signal represents the time of a conflagration that took place after the wall had been built and rules out the possibility that it predated the construction of the wall. Our results further support the recent suggestion (Vaknin et al. 2023) to re-evaluate previous hypotheses regarding the use of pre-fired bricks in the Southern Levant before the Roman period (e.g., Namdar et al. 2011: 3477; Faust et al. 2017: 146).

Unlike the unified magnetic signals within every sample, the results of the different samples are somewhat scattered, suggesting that some of these bricks moved slightly after cooling. Indeed, in the field the bricks did not seem to be exactly in their original position (Figs. 8, 9a–b). The relatively low inclination yielded by most of the samples can be explained by a slight tilt of the tower-buttress to the south, which seems very likely since the tower was built on the very steep southwestern slope of the mound (Fig. 5). Therefore, the direction results are very useful for site formation research, but they do not enable reconstruction of the direction of the geomagnetic field and cannot be used as a chronological marker.

In our archaeointensity experiments the magnetisation was nearly entirely removed only at ca. 600⁰ C (Fig. 12). The thermomagnetic curves were reversible up to 600–700⁰ C (Fig. 14). These results indicate that these bricks had been heated to at least these temperatures. These firing temperatures or even higher temperatures, as previously suggested (Shimron 2004: 2648–2649), are indicative of an intense conflagration that took place adjacent to the tower-buttress.

Archaeomagnetic dating of the firing of the bricks in the tower-buttress: Iron Age after all

In addition to the possible ancient dates for the firing of the tower-buttress bricks, we considered yet another possibility: that these bricks were burnt after they had been exposed by the British expedition in the 1930s. The area surrounding LC08 had probably been covered with seasonal vegetation until the TAU excavations in Area R began in 1983. The firing of tower-buttress bricks could have been caused by accidental fires, as occurred in the 1973 season, or by fires lit intentionally by the TAU expedition before each of their following seasons, which burnt the seasonal vegetation on the entire mound (Ussishkin 2014: 69–71 and Fig. 4.2). The possibility that LC08 recorded the magnetic field during one of these fires was ruled out by our archaeointensity results of LC08, which are approximately twice as strong as the magnetic field during the excavations of the TAU expedition (Fig. 13a). These results show that the recurring fires did not affect the ancient magnetic signal (see below) of LC08. This suggests that brush fires do not necessarily reach sufficiently high temperatures to affect the magnetic signal of burnt mudbricks.

The archaeeointensity measured from LC08 is more than twice as strong as the field during the MB II (Fig. 13a). This rules out the possibility that this wall was fired only during the MB II. In fact, such a strong field in the Levant is indicative of only four episodes, termed ‘spikes’, all of which occurred during the Iron Age (Fig. 13a). The strong magnetic field during the Iron Age in the Levant, including these spikes, is termed the Levantine Iron Age anomaly (Shaar et al. 2016; Ben-Yosef et al. 2017; Shaar et al. 2017; Shaar et al. 2018). Relying on the magnetic results alone we may conclude that the bricks of the tower-buttress were fired during one of the Iron Age ‘spikes’.

That said, when dating a wall, it is important to differentiate between the construction date, the period of its use and its stratigraphic attribution. Since our results indicate that the tower-buttress was burnt during the Iron Age, it must have been in use during this period. Since our results also imply that the tower-buttress was built of sun-dried bricks, it is unlikely that such a structure was in use continuously from the MB II into the Iron Age. Therefore, it was most likely constructed during the Iron Age. Based on the connection between the Outer Revetment Wall, the tower-buttress and the main city wall on the southwestern side of the mound, we support the attribution of these fortifications to Levels IV–III (Ussishkin 2004c: 79–80, 85). However, this stratigraphic observation and our results do not rule out the possibility that earlier stone-built fortifications were incorporated into the Outer Revetment Wall when it was constructed, as previously suggested by Tufnell (1953: 60). For instance, it seems possible that the stone wall in the northeastern part of the mound was constructed during the MB II, as suggested by the HUJI expedition (Garfinkel et al. 2021b: 426–429), and was still in use during the Iron Age when it was integrated into the Outer Revetment Wall. This possibility has already been implied by Ussishkin (2023: 98–102).

In order to further pinpoint when the tower-buttress bricks were fired, we can rule out the latest spike, which occurred in ca. 600 BCE (Vaknin et al. 2022: 5), since the Assyrian siege ramp had completely covered the tower-buttress since 701 BCE (Ussishkin 2004a: 723, 726). Therefore, the magnetic data must have been recorded during one of the three earlier spikes (Fig. 13a), as indicated by our archaeomagnetic dating (Fig. 13b).

Combining our archaeomagnetic results with the archaeological data from the outer part of the tower-buttress leads us to the conclusion that the tower was burnt in 701 BCE during the Assyrian siege. Our archaeointensity results of LC08 overlap with those from the gate area of Lachish III (Fig. 13a), which serve as the chronological anchor for 701 BCE (Vaknin et al. 2022: 4). A date of 701 BCE is indeed possible for the firing of LC08 according to our archaeomagnetic dating (Fig. 13b). In the southwestern corner of the mound, the British expedition exposed the meeting point of the tower-buttress and the Outer Revetment Wall, as well as charred branches and olive pits, stone slingshots and ‘innumerable arrowheads’ (Tufnell 1953: 90). It seems that these traces of fire may explain the firing of the tower-buttress bricks and the other finds support its connection to the Assyrian siege. The olive pits mentioned in the report may suggest that branches of olive trees, still bearing olives, were used as fuel for the fire that occurred near the tower-buttress.

The very clear remains of the siege and battle, which were also found by the TAU expedition in the immediate vicinity of the tower-buttress, as mentioned above, further support a 701 BCE date for the firing of the bricks. Indeed, Ussishkin (2004a: 732) explains the presence of fired bricks as the result of fires ignited during the battle. The construction of the siege ramp took at least several weeks and according to Ussishkin (ibid.: 707–723) it was constructed in two stages.⁴ The study of the distribution of arrowheads and slingshots in Area R demonstrated that the construction of the siege ramp was accompanied by extensive counter-measures undertaken by the defenders of the city in order to prevent its progress (Gottlieb 2004: 1951–1956, Fig. 27.21). These measures, as well as the Assyrian offensive actions, became more intense as the construction of the ramp approached the defence lines, which is evident in particular at the foot of the tower-buttress (ibid.; Ussishkin 2004a: 732–739). Perhaps the mudbrick debris found in front of the tower-buttress, on top of the first stage of the siege ramp and below its second stage (Ussishkin 2004a: 723), had fallen from its walls at this time. We conclude that the tower-buttress, which was clearly in use during the 701 BCE siege, was subsequently burnt, perhaps when the construction of the siege ramp approached it. Since the tower-buttress was incorporated into the Outer Revetment Wall (Figs. 1, 3–4), it seems very likely that the entire Outer Revetment Wall was in use in 701 BCE as well. This shows that the assumption made by Starkey, based on the Lachish relief before his excavations had begun, was most likely correct.

The use of fire during the 701 BCE siege leading to the firing of the tower-buttress

In our view, there are two main possible scenarios regarding the circumstances that led to the firing of the bricks of the tower-buttress. One possibility is that the defenders of Lachish threw torches and other burning materials, as depicted in the Lachish relief (Fig. 15a). This could have been intended to sabotage the construction of the siege ramp. Even though no wood remains were found within the siege ramp, it is certainly possible that wood had been used but was not preserved (Ussishkin 2004a: 716–717). In the Lachish relief, the siege ramp is covered with lines that are usually interpreted as wooden logs placed by the Assyrians to facilitate the ascent of their siege engines (ibid.: 717; Garfinkel et al. 2021a: 435). The use of wood in order to build a siege ramp is mentioned in the description of the siege laid by Esarhaddon, son of Sennacherib, to the city of Uppume, as well as the attempt of the defenders of the city to burn it (Borger 1956: 104; Ephʿal 2013: 89; Siddal 2019: 46).⁵ Similar use of wood in order to build a siege ramp during the Babylonian attack on Jerusalem is implied by Jeremiah (Jer 6:6; see Ephʿal 2013: 85). If indeed wood was used during the construction of the siege ramp at Lachish, perhaps the defenders succeeded in setting it on fire. If the Assyrians raised their siege engines on the siege ramp before its construction was completed in order to attack the tower-buttress, the defenders of the city probably attempted to set these engines on fire by throwing torches, as depicted in the relief (Fig. 15a; see Ussishkin 2004a: 741). From every one of the seven siege engines seen in the relief, an Assyrian soldier is depicted pouring water (Fig. 15a) in an attempt to prevent the battering ram and the area between the siege engines and the city walls from catching fire (Ephʿal 2013: 89; Ussishkin 2004a: 741). Ussishkin (ibid.) suggested that the defenders of the city also poured boiling oil on the Assyrian soldiers from the city walls. An intense fire at the foot of the tower caused by these actions might explain the firing of the studied bricks.

Fig. 15: Two examples of the use of fire in warfare depicted in Assyrian reliefs; a) part of the Lachish relief from Sennacherib’s palace at Nineveh (Ussishkin 1982: Fig. 78; drawing by Judith Dekel); note the torches thrown from above and the Assyrian soldier pouring water in front of the siege engine; b) a relief from Ashurbanipal’s palace at Nineveh depicting the Assyrian conquest of an Egyptian city; an Assyrian soldier setting the city gate on fire is marked (photo: Osama Shukir Muhammed Amin FRCP [Glasg], CC BY-SA 4.0, via Wikimedia Commons)

An alternative explanation that should be considered for the firing of the bricks is that it was caused by the Assyrian army. The Lachish relief, which is focused on the final stages of the siege and does not depict the long period of construction of the siege ramp (Garfinkel et al. 2021a: 435), does not support this possibility. The Assyrians could have started a fire at the foot of the tower-buttress for several reasons. It is likely that on top of the tower the defenders were protected by fortifications partially made of wood. Such fortifications seem to be depicted in the Lachish relief, where they protrude beyond the line of the walls (Yadin 1963: 325–327, 434, 449), a fact that could have facilitated their burning from below. A sufficiently large fire at the foot of the tower-buttress or somewhat higher, on the partially built siege ramp, could have destroyed these flammable fortifications.

Another reason for the Assyrians to have started a fire outside the tower-buttress is that heat and smoke from such a fire would have disturbed the defenders located on the tower in their attempts to prevent the construction of the siege ramp. It is important to note that the wind in the area is from a westerly direction during the day all year long, with slight variations.⁶ Perhaps the wind direction was one of the Assyrian considerations when they chose to attack the southwestern corner of the mound, in addition to the main consideration—the topography (Ussishkin 2004a: 695–696)—and the advantage of attacking a corner of the city (Ephʿal 1984: 60–61). A trial cut carried out through the siege ramp in 1977 revealed a layer of ash and charred wood, identified as olive and terebinth, beneath the siege ramp (Ussishkin 2004a: 707, 716). Ussishkin suggested that this layer represents fires that had been lit prior to the construction of the siege ramp. Since the trial cut is 35–50 m from the tower-buttress (Fig. 4), it seems unlikely that the defenders of the city had lit these fires from within the city fortifications. We therefore suggest that the Assyrians lit these fires in order to create a smoke screen.

Wind direction during a siege plays a major role in the description of Esarhaddon’s siege of the city of Uppume mentioned above. It is said that the defenders of Uppume started a fire at the foot of the city walls but that the wind direction was then changed miraculously by the god Marduk in the direction of the city wall. Consequently, the fire reached the wooden elements on the city wall and burnt them (Borger 1956: 104; Ephʿal 2013: 89). Ephʿal (ibid.) suggested that these wooden elements ‘were most probably defensive constructions such as the balcony-like embattlements atop the wall’. This example demonstrates the possibility that some of the fortifications of Iron Age cities contained wooden elements and that the Assyrians were aware of the effect of wind direction on a fire at the foot of a wall of a besieged city. The use of fire by the Assyrian army is demonstrated in Ashurbanipal’s relief depicting his conquest of an Egyptian city (Yadin 1963: 462–463), where an Assyrian soldier is portrayed holding a torch and setting the city gate on fire (Fig. 15b).⁷

The high elevation of the burnt bricks within the wall (Fig. 7a) and their high firing temperatures seem to support the possibility that their firing was indeed the result of an intense conflagration set by the Assyrians. If there were wooden elements on the balcony on the tower-buttress, it is hard to assume that the fire that burnt the studied bricks would not affect them. Therefore, even if the defenders of the city threw torches in order to set the siege engines or the wood used to construct the siege ramp on fire, they would probably have avoided throwing a large amount of fuel close to the wall, which could eventually burn their wooden fortifications. It seems more probable to us that the firing of the studied bricks was the result of an Assyrian act of war.

Conclusions

1. The tower buttress at Tel Lachish was most likely built of sun-dried bricks that were fired in situ to at least 600–700⁰ C.

2. This firing occurred in the Iron Age, most probably during the 701 BCE Assyrian siege.

3. The mudbrick tower-buttress was built during the Iron Age.

4. The Outer Revetment Wall was probably constructed during the Iron Age, but earlier fortifications could have been incorporated into it.

5. The firing of the tower-buttress could be attributed to the defenders of Lachish, but we prefer an attribution to the Assyrian army.

Acknowledgements

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 804490). We are grateful to Saar Ganor and Vladik Lipshits for their support in the fieldwork. We appreciate the very fruitful discussions regarding this research with David Ussishkin, Yosef Garfinkel, Yehuda Dagan and Igor Kreimerman. We further acknowledge three anonymous reviewers, whose constructive comments significantly improved this paper.

Disclosure statement

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

Additional information

Notes on contributors

Yoav Vaknin

Yoav Vaknin: The Sonia and Marco Nadler Institute of Archaeology, Tel Aviv University, and Institute of Earth Sciences, The Hebrew University of Jerusalem

Ron Shaar

Ron Shaar: Institute of Earth Sciences, The Hebrew University of Jerusalem; email: [email protected]

Erez Ben-Yosef

Erez Ben-Yosef: The Jakob M. Alkow Department of Archaeology and Ancient Near Eastern Cultures, Tel Aviv University; email: [email protected]

Oded Lipschits

Oded Lipschits: The Jakob M. Alkow Department of Archaeology and Ancient Near Eastern Cultures, Tel Aviv University; email: [email protected]

Notes

1 LC stands for Lachish and 08 is a running number out of all archaeomagnetic samples from Tel Lachish.

2 Field orientation is the orientation in which the bricks were unearthed.

3 There is a debate regarding fired bricks that were exposed in the MB III fortress at Tel Lachish mentioned above (Area BB). The British expedition interpreted these bricks to be ‘kiln-fired red bricks’ (Tufnell 1953: Pl. 11:12), whereas the HUJI expedition interpreted the firing of the bricks to be the result of conflagration (Garfinkel et al. 2021b: 429).

4 According to a recent study, the siege ramp was even longer than suggested by Ussishkin and its construction began far from the city and gradually approached it (Garfinkel et al. 2021a).

5 Note the significant amount of wood used during the construction of the Roman siege ramp at Masada (Liphschitz, Lev-Yadun and Waisel 1981; Lev-Yadun, Lucas and Weinstein-Evron 2010).

6 According to wind measurements in 2001–2012 in the Netiv HaLamed He meteorological station, located ca. 18 km northeast of Tel Lachish.

7 A similar practice is portrayed in the story of Abimelech, son of Gideon, who besieged the Tower of Shechem, cut down brushwood, put it against the tower and set the tower on fire (Judg 9:49). This biblical narrative ends when a woman in the besieged tower in Thebez managed to kill Abimelech when he approached the tower in an attempt to set it on fire as well (Judg 9:50–54). We thank Yosef Garfinkel for the reference to the Abimelech narrative.