The Bible and Radiocarbon Dating:
Archaeology, Text and Science
Thomas E. Levy and Thomas Higham

(The Bible and Radiocarbon Dating, Thomas E. Levy, Thomas Higham, A.J. Shortland, 2005)


During the 1970s and 1980s, there was almost a consensus concerning the dating of two main pottery assemblages that are of interest to us here:

a. The late Iron Age I assemblage, represented by the pottery from the destruction layers of Megiddo VIA, Tell Qasile X, Tel Masos II, and parallel strata at other sites, was dated to the late 11th or early 10th century BCE.

b. The Iron IIA assemblage, with northern and southern variants, represented by the destruction layers of Megiddo VA-IVB, Taanach Period IIB, Yokneam XIV, Beth Shean S-1 (in the Hebrew University excavations), Lachish V, Arad XII, and parallel strata at other sites, was dated by most scholars to the 10th century, until Shishak's raid.

The deconstruction of the Albright/Yadin `Solomonic paradigm' began during the 1970s with Benno Rothenberg's and later by Gary Pratico's demolition of Nelson Glueck's concept of `Solomon's copper mines' in the Timnah Valley in the Arabah and of his identification of Ezion Geber with Tell el-Kheleifeh. Twenty years later, in the late 1990s, questions were raised con-cerning the core of the paradigm by J. Wightman (1990) and especially by David Jamieson-Drake (1991) whose influential book included a frontal attack from an archaeological standpoint on the very concept of the United Monarchy of ancient Israel. This work fell like a ripe apple into the hands of historians and biblical scholars of the European 'revisionist school' who were inclined to minimize or reject altogether the historicity of the entire or parts of the biblical narrative. This book also inspired the work of Israel Finkelstein.

Finkelstein's Low Chronology (LC)

Since 1996, Finkelstein (1995, 1996) went one step further by suggesting the wholesale lowering by 50-80 years of archaeological assemblages traditionally attributed to the 12th-10th centuries BCE. His first point was the date of the appearance of the local Mycenaean IIIC or 'Philistine Monochrome' pottery. Following Ussishkin (1985), he suggested lowering the appearance of this pottery by 50 years until after the end of the Egyptian presence in Canaan. This subject is beyond the scope of the present discussion, but it should be mentioned that several recent studies and discoveries, such as those at Ashkelon, negate this approach; in fact, none of the excavators of Philistia find this suggestion acceptable. It also creates unsolvable problems in correlating the archaeology of Philistia with that of Cyprus (Dothan and Zukerman 2003; Mazar 1985 and forthcoming; Sherratt and Master [Chapters 9 and 20, this volume;). "C dates for this period are not of much help, due to the many wiggles and complicated shape of the calibration curve for the 11th and 12th centuries BCE. Consequently, Finkelstein suggested lowering the dates of late Iron Age I assemblages from the late 11th century to the 10th century BCE and the lowering of tradi-tional 10th century BCE assemblages to the 9th century BCE. His view became known as the 'Low Chronology' for the Iron Age of Israel. This suggestion empties the 10th century BCE of its tradi-tional contents. Sites and strata that were traditionally dated to the late 11th century BCE, such as Megiddo VIA, are dated to the 10th century BCE, until Shishak's campaign (Finkelstein 1998a, 1998h, 1999b, 2002a, 2002b, 2004, and Chapters 3 and 17, this volume).

In a separate study based on '4C dates from Tel Dor, Ayelet Gilboa and flan Sharon suggest an even lower chronology from that suggested by Finkelstein (see below).

Review of Finkelstein's Low Chronology

Since it was initially suggested in 1996, the LC and the historical perspectives that followed it have become the subject of continued controversy. I started the debate with a comprehensive review of this theory from an archaeological perspective (Mazar 1997) and others followed (Ben-Shlomo,


Shai, and Maeir 2004: 2; Ben-Tor 2000; Ben-Tor and Ben-Ami 1998; Bunirnovitz and Faust 2001; Byrne 2002; Dever 2001: 131-38; Harrison 2003; Herzog and Singer-Avitz 2004; Kletter 2004 [the most comprehensive review of many aspects of the LC and its historical implications]; Singer-Avitz 2002;2 Zarzecki-Peleg 1997). Today, most archaeologists in Israel still hold with the conventional chronology, while a few support the LC (Gilboa and Sharon 2001, 2003; Gilboa, Sharon, and Zorn 2004; Herzog 2002 but no more in Franklin [Chapter 18, this volume]; Herzog and Singer-Avitz 2004 [see footnote 2]; Knauf 2002).

The period under debate is framed by upper and lower anchors. The upper anchor is in the 12th century BCE, represented by the well-established correlations between the Egyptian Twentieth Dynasty and several sites in the Levant, in particular Beth-Shean (Level VI of the University of Pennsylvania excavations and Strata S-4 and S-3 of the Hebrew University excavations), Megiddo (Stratum VIIA), Lachish (Level VI), Tel Sera' (Stratum IX), and the Timnah Valley copper mines (Mazar 1990: 295-300; 1993; 2002: 264-72). The end of these strata, in several cases as a result of violent destruction, occurred towards the end of the Egyptian presence in Canaan, during the reigns of Ramesses IV to VI, until ca. 1140/1130 BCE. The lower anchor is related to the Assyrian conquests between 732 and 701 BCE. Destruction layers related to these conquests were identified at many sites such as at Dan, Hazor, Tel Kinneret, En-Gev, Beth-Shean, Tel Rehov, Megiddo, Yogne`am, Samaria, Tell el-Farah, Khirbet Marjameh, Timnah (Tel Batash Stratum III), Lachish (Stratum III), Beth Shemesh and Tell Beit Mirsim. Related destructions occurred also at Tel Beer-Sheba (Stratum II) and Arad (Stratum VIII). There is a consensus concerning the dates of these destructions and thus they can be taken as a datum line for further discussion. Between these two secure anchors is a period of about 400 years, which leaves us with enough room for a continuous debate.

A major point in this debate is the question of whether we are able to establish secondary chronological anchors between the two main ones mentioned above.

There are two such minor anchors on which all the sides in this debate agree. The first is represented by the site of Jezreel excavated by Ussishkin and John Woodhead (1997). The history of Jezreel is known only from the Hebrew Bible, yet all scholars agree that Jezreel was indeed the second residence of the Omride Dynasty and that it was destroyed soon after the end of that dynasty, ca. 840-830 BCE. Pottery assemblages from the destruction of Jezreel can thus safely be dated to this time. Orna Zimhoni, who published the pottery from Jezreel, pointed out its similar-ity to the pottery from the `Solomonic' Stratum VA-IVB at Megiddo, and this was one of Finkel-stein's main arguments for lowering Megiddo Stratum VA-IVB to the time of the Omride Dynasty in the 9th century BCE. However, Zimhoni has also shown that similar pottery was found in the construction fills below the royal enclosure of Jezreel, probably originating from a dismantled earlier village that could date to the 10th century BCE (Zimhoni 1997: 29-56). This suggests that the same pottery assemblage continued throughout much of the 10th and 9th centuries. Such a long duration of the same assemblage was also observed at Hazor, Tel Rehov, and other sites, and this, in my view, is the key to the resolution of our debate.

Another secondary chronological anchor is related to Arad and the Negev Highlands sites. All agree that at Arad, either Stratum XII or Stratum XI must be identified as the 10th-century settlement mentioned in Shishak's inscription (Aharoni 1981: 182-91). Many of us agree that it should be the Stratum XII village rather than the Stratum XI citadel (thus Finkelstein 2002b;

2. I included Singe-Avitz in this category since her conclusions concerning Arad and Lachish in the cited paper fit the conventional chronology. In Herzog and Singer-Avitz (2004) both authors accept the long duration for the Iron Age IIA (most of both the 10th and 9th centuries BCE) as suggested by me since 1997.


Herzog and Singer-Avitz 2004; Mazar 1990: 373; Singer-Avitz 2002).3 Arad thus provides an agreed reference point for the pottery of the Northern Negev in the second half of the 10th century. Yet this agreement works against Finkelstein's LC, since the pottery assemblage from Arad XII is identical to that found at other sites that have been dated to the 10th century BCE according to the conventional chronology, like Lachish Stratum V and various parallel levels (Mazar and Panitz-Cohen 2001: 277-79; Singer-Avitz 2002: 114). Concerning the Negev Highland settle-ments, Finkelstein (1984) dated them to the 11th century BCE, forcing an historical interpretation which would fit this period; more recently he lowered this date to the 10th century BCE, in accord with his LC (Finkelstein 2002h) and thus he now agrees to the mainstream conventional dating of these sites (Cohen 1980; Cohen and Cohen-Amin 2004; Haiman 2003; Herzog and Singer-Avitz 2004: 225-26; Mazar 1990: 390-96). He also accepts the view, long ago suggested by Cohen, Meshel, and others, that these sites should be identified with the dozens of Negev sites mentioned by Shishak. The interpretation of these sites as related to the United Monarchy as suggested by Cohen and others remains in my view the most feasible one. The pottery from the Negev Highland o sites is typical Iron IIA pottery. If Arad XII and the Negev Highland sites can be dated to the 10th

century BCE according to both the conventional and the LCs, so can be Lachish V, Beer Sheba VI

V, Tell Beit Mirsim B3, Tel Batash IV, Tell Qasile IX-VII, Gezer VIII, Beth-Shemesh IIa, and so

on, all with the same pottery. This conclusion (based, as mentioned above, on agreements between

all sides in this debate) makes the LC impossible, at least in Judah, Northern Negev, and the

southern coastal plain. Consequently, the picture emerging concerning the status of Judah in the

10th century BCE must differ from that described by Finkelstein (1999a).

Shishak's raid has been considered by many as a benchmark for the late 10th century BCE, yet there are diverse views concerning the question of whether Shishak indeed destroyed cities, and if so, which archaeological levels can be identified as having been destroyed by him. Our suggestion that a vast destruction layer at Tel Rehov should be attributed to Shishak since the place is men-tioned in his inscription and since this destruction can he dated to the second half of the 10th century (Bruins, van der Plicht, and Mazar 2003; Mazar 1998) was severely attacked by Finkelstein and Piasetzky (2003). Yet at the same time Finkelstein (2002b) suggested that a series of other destructions should be attributed to Shishak, even at sites not mentioned in his list, such as Tel Miqne-Ekron Stratum IV. This dual approach remains mysterious to me. We should either believe that Shishak simply moved through the country without causing destructions (thus Na'aman 1998) or leave open the possibility that indeed he destroyed cities and settlements (perhaps only partly), and in such a case the search for such destruction layers remains a legitimate one, particularly in a place like Rehov which is mentioned in his list. This latter approach is more feasible in my view. Lawrence Stager brought up the case of Taanach as an example of a city mentioned in Shishak's list where only one destruction level—that of Period IIB—can be identified as the city destroyed by him. The pottery from this level is identical to that of Megiddo VA-IVB, and thus another benchmark for the 10th century BCE may he suggested. Finkelstein tried to resolve the Taanach case by lowering Period II to the 9th century BCE and Periods IA and IB to the 10th century BCE, yet this claim must he rejected on the basis of pottery analysis: as shown by Rast (1978) the pottery from Period IA-B is close to that of the Late Bronze Age and fits the 12th century BCE (Mazar 2002: 278-79).4

3. A variation is Herzog's view (2002: 92-93) that Arad XII was constructed during the 10th century Bch, but continued to survive well into the 9th century. This suggestion is not repeated in Herzog and Singer-Avitz (2004), where Arad XII is dated to the 10th century alone, before Shishak's raid.

4. Finkelstein's discussion of Taanach (1998h) is based on dismissing the pottery from Period I as irrelevant due to the fact that most of it included only sherds. Yet such a dismissal stands against the principles of archaeological investiga-tion. Rast's conclusions were carefully crafted and should not be rejected.


Thus, Jezreel, Arad, the Negev Highlands, and Taanach may be taken as 'mini-anchors' in the problematic 400-year time-span described above. Evaluation of these points of reference negates Finkelstein's LC.

The Modified Conventional Chronology and Iron HA Material Culture

The results of the archaeological work of the 1990s and renewed analysis of various sites led me to change my previous view and accept Aharoni and Amiran's scheme from 1958 with some modifications. Recognizing the long duration of the Iron IIA pottery period, I suggested that the boundary between Iron I and Iron II be placed somewhere in the first quarter of the 10th century BCE (an estimated date is ca. 980 BCE) and that the end of the Iron IIA period should be some 150 years later, after the end of the Omride Dynasty and the destruction of Jezreel, ca. 840/830 BCE (Mazar in Coldstream and Mazar 2003: 40-44; Mazar 1997: 164; Mazar in Mazar and Carmi 2001: 1340). This scheme enables the definition of three major pottery periods in the 450 years between ca. 1150 and 700 BCE: Iron IB, Iron IIA, and Iron IIB, each with regional variations and each lasting about 150 years (Table 2.1). These scheme is supported by "C dates from Tel Rehov (Mazar et al. [Chapter 13, this volume]). Such a scheme may be defined as 'Modified Conventional Chronology' (MCC) for the Iron Age in Israel. It was recently accepted by several scholars as the best resolution for Iron Age chronology (Ben-Shlomo, Shai, and Maeir 2004: 2; Herzog and Singer-Avitz 2004).

In contrast to Finkelstein's view, I suggest that during the first half of the 10th century BCE a major change took place in the material culture throughout the country; this change brought to an end the Canaanite Second Millennium culture as is best demonstrated by Stratum VIA at Megiddo and related strata in the northern valleys, such as Yogi-team, Dor and Tell Keisan. The new material culture is characterized by various aspects—from new modes of pottery production (domi-nance of red slip and hand burnish, disappearance of the Canaanite painted pottery tradition) to settlement patterns, architecture and religious art.

It should though be acknowledged that a definition of the material culture of the United Monarchy is strewn with difficulties. Since archaeology supplies the only first-hand evidence for this period, apart from Shishak's inscription at Karnak, it is essential to define properly which archaeological remains can be dated to the time of this kingdom. The MCC with its long duration of the Iron Age HA suggested above makes this goal hard to achieve, since we cannot say categorically whether a certain Iron IIA archaeological context belonged to either the 10th or the 9th centuries BCE. In my view both options are open in many cases, while Finkelstein's view does not leave such an option and according to him all Iron Age IIA contexts should be dated to the 9th century BCE alone (2004: 185). I claim that the archaeological picture is far from being 'crystal clear', and that the traditional paradigm of 'the archaeology of the United Monarchy' remains a legitimate possibility, though not mandatory (for summaries see Dever 1990; Mazar 1990: 368-402). Thus, I see no difficulty in retaining the `Solomonic' date of the monumental palaces 6000 and 1723 at Megiddo. Their dating to the 9th century BCE (a main point in Finklestein's theory; see also Franklin [Chapter 18, this volume]) would leave for the entire 100 years between ca. 980 and 880-860 BCE, a poor ephemeral occupation level at Megiddo (Stratum VB). This is not impossible, but not very feasible, especially when taking into consideration the tight stratigraphy and pottery developments at sites like Hazor and Tel Rehov, and the clear 10th-century BCE date of two Iron IIA levels at Tel Rehov. The strongest point in favor of a 9th-century BCE date of the Megiddo palaces is their building technique and masons marks which resemble those at Samaria (Finkelstein 2004:185; Franklin [Chapter 18, this volume]). Yet this resemblance can be explained if we assume that both kings—Solomon and Ahab—used Phoenician masons. Builders families or


4 Shishak, King of Egypt

The challenges of Egyptian calendrical chronology

A.J. Shortland


In reconstructing ancient historical chronologies, much use has been made of chronological pins between neighbouring states, linking their chronologies together. This chapter examines one such pin, the attack of Shishak, King of Egypt on the Levant in the early first millennium BCE. Due to the danger of circular arguments, it works entirely from Egyptian records, rather than combining these with biblical or Assyrian dates as is normal. It assesses the way the Egyptian chronology is put together and its strengths and weaknesses and goes on to examine in detail the Third Intermediate Period, specifically the 22nd and 25th Dynasties. In doing this it draws extensively on Kitchen (1986), a standard reference work for this period, but one that may not be totally accessible to those not specializing in Egyptian archaeology. The chapter concludes that the most likely minimum recon-struction of the date of the accession of Shishak/Sheshonq I is 941 BCE, with dates in the mid-940s BCE being the most likely overall. This supports biblical dates for the attack well, which would conventionally place the accession of Shishak/Sheshong I in 945 BCE. It emphasizes that, while not perfect, the Egyptian chronology is very robust and internally consistent, even without reference to external events.


The reconstruction of ancient chronologies is always a difficult issue, and often a contentious one. This is especially so when, as is usually the case, the chronology is constructed from many different types of evidence: textural, archaeological, astronomical, scientific, and so on, each bringing with it its attendant specialist who may have little or no appreciation of the complexities, strengths and weaknesses of the contributions to the subject of the other disciplines. Add to this the interconnect-edness of ancient nations and therefore the necessity to take into account the histories of several neighbouring states when considering one, and the situation is ripe for confusion and dispute.

The aim of this chapter is to look again at one incident where two of these ancient nations are apparently interconnected. The textural reference is shown below:

`In the fifth year of King Rehoboam, Shishak king of Egypt attacked Jerusalem. He carried off the treasures of the Temple of the Lord and the treasures of the royal palace. He took everything, including all the gold shields Solomon had made'. (1 Kings 14.25-26)

Here there is apparently a clear reference to an Egyptian king appearing in the history of Israel. This is important since Egypt has one of the best calendrical chronologies of all the ancient states



Dating the accession to the throne of Sheshonq I is therefore a matter of adding up the reigns of the intervening kings and then applying this to the fixed point of 664 BCE. The order of the kings is fairly well established, and proceeds mostly in a sensible father-to-son pattern. The gap between the two dynasties can be bridged by reference to the extremely useful Apis 24, 1, which also gives a sensible reign length for the shadowy Osorkon IV. In the simplest form, the date of the accession of Sheshonq I can be taken as the total of the highest regnal dates of all the kings of the 22nd and 25th Dynasties. This would be 253 years, giving a date of 917 BCE. However, this is not the best fit to all the data, and ignores hard evidence from the Apis and Pasenhor genealogies. Using these, Osorkon IV's reign must be extended from the nonsensical zero to 15 years and Shabitqo from 3 to about 12 years. This lengthens the chronology by 24 years, and takes it back to 941 BCE. Thus from entirely internal Egyptian evidence, a minimum date of 941 BCE and a probable date in the mid-940s BCE must be postulated as the most likely date for the accession of Sheshonq I. This is remarkably close to the date derived from the use of external evidence (945 BCE), strengthening the assessment that the chronologies here are coherent and reasonable.

The actual date of the campaign of Sheshonq is based on the fact that the reliefs in the Bubastite Portal are unfinished and therefore the campaign and the reliefs are interpreted as falling late in his reign. Since he reigned for 21 years, year 20 is usually cited as the year of the campaigns, a date of 925 BCE. Speculation has been made that the campaigns represented in the Bubastite Portal may be just one of several campaigns made by Sheshonq I in the Levant, and attacks may have been made earlier in his campaign, leaving destruction layers in the Levantine cities perhaps dating to the 940s and 930s BCE. This is possible, but no evidence exists from Egyptian records for such attacks. While there are inconsistencies with tying the Bubastite Portal campaign in with the damage on the ground seen in archaeological excavation, from an entirely Egyptian point of view, it still remains the best fit.

As can be seen, the Egyptian chronology, like that of all other ancient chronologies, requires contradictory evidence to be weighed and assessed before a most likely chronology can be drawn up. It is not perfect, not free of error and not 'set in stone', but is subject to new findings and new interpretations. It does, however, stand up remarkably well to such findings, and the arguments now usually revolve around one or two years on the end of reigns and the affiliations of individual kings rather than wholesale changes in the length or nature of the chronology. As such we can be very confident of ascribing the accession of Sheshonq I to the middle of the 940s BCE.



Casperson, L.W. (1988) The Lunar Dates of Ramesses II. JNES 47: 181-84.

Censorinus (1983 edn) Censorini De die natali liber ad Q. Caerellium : accedit anonymi cuiusdam ept oma disciplinarum (fragmentum Censorini) (ed. N. Sallmann; Leipzig: Teubner).

Dodson, A. (2000) Towards a Minimum Chronology for the Third Intermediate Period. Bulletin of the Egyptological Seminar 14: 7-18.

Epigraphic Survey (1936-86) Reliefs and Inscriptions at Karnak (Chicago: Oriental Institute).

Hornung, E. (1965) Die Sonnenfinsternis nach dem Tode Psammetichs I. Zeitschrift fur Aegypt sche Sprache 92: 38-39.

Kitchen, K.A. (1986) The Third Intermediate Period in Egypt (Warminster: Aris & Phillips, 2nd edn). —(1991) The Chronology of Ancient Egypt. World Archaeology 23: 201-208.

Mackey, D.F. (1993) The Sothic Star Theory of the Egyptian Calendar (unpublished thesis, University of Sydney).

Manetho (1940 edn) Aegyptiaca (trans. W.G. Waddell; London: Loeb Classical Library).

typical half of )rthern ..en the ive this house (Strata

and in differ-lly but exca-lower n A4 is ictivity :arbon 15354, 25321 ions of

GrA-I with tion of


10 Lowland Edom and the High and Low Chronologies

Edomite state formation, the Bible and recent archaeological research in southern Jordan

Thomas E. Levy, Mohammad Najjar, Johannes van der Plicht, Neil Smith, Hendrik J. Bruins, and Thomas Higham


This study explores the chronological assumptions that underlie the past 40 years of Iron Age archaeological investigations in southern Jordan and offers an alternative framework based on the application of high precision radiocarbon dating. The 2002 University of California, San Diego—Department of Antiquities of Jordan (UCSD—DOAJ) archaeological excavations at the copper pro-duction center of Khirbat en-Nahas (KEN) demonstrate monumental building and industrial scale copper production in two major phases dating to the 12th-11th and 10th-9th centuries BCE. Strati-graphic excavations, new high precision radiocarbon dating using short-life samples, and small finds such as ceramics, scarabs, and arrowheads from the site show the centrality of the Iron Age landscape in the copper ore-rich lowlands of Edom for the formation of complex societies in this part of the southern Levant. The new data presented here challenge previous assumptions about the Iron Age in Jordan, such as (a) the formation of the Iron Age kingdom of Edom only took place in the 7th and 6th centuries BCE and (b) no monumental building activities took place in Transjordan during the 10th century BCE. Bayesian statistical analyses of the radiocarbon dates from KEN are presented by Higham et al. (Chapter 11, this volume).


This study discusses some of the archaeological and historical implications of the latest suite of high precision radiocarbon dates obtained from the Oxford and Groningen radiocarbon laborato-ries from the recent excavations at the Iron Age metal production center at Khirbat en-Nahas in Jordan. To appreciate the impact of these new radiometric dates on the Iron Age archaeology of southern Jordan, and radiometric dating on historical archaeology in general, some discussion of the role of text and archaeology must be discussed in order to attain some of the goals of a 'New Biblical Archaeology' outlined at the beginning of this volume (see Chapter 1). In the 19th century, systematic archaeological research in the southern Levant—the Holy Land—was born with the aim of exploring the relationship between text (the Hebrew Bible) and the newly understood field of archaeology. In 1865, the Palestine Exploration Fund—the first research organization devoted to the scientific investigation of the history of the land—was founded in London by a group of distinguished scholars and clergymen, with the express purpose of providing 'for the accurate and


systematic investigation of the archaeology, topography, geology and physical geography, natural history, manner, and customs of the Holy Land, for biblical illustration' (cf. Moorey 1991). The unique historical relationship between the Hebrew Bible (Old Testament) and the landscape of Palestine created what might be called the 'tyranny of the text'. Accordingly, in approaching the archaeological record of the southern Levant, from its 19th-century beginnings until the mid-1970s, archaeologists consistently approached the archaeological record of the Holy Land by first examining biblical text and then searching for material culture proof to support the text as historical fact (Albright 1971; Glueck 1940a; Wright 1965). Following the discovery of incon-sistencies between text and the archaeological record at key sites such as Jericho, which was supposed to have been destroyed by Joshua and the Israelite tribes at the end of the Late Bronze Age, cracks developed in the paradigm known as 'Biblical Archaeology'. By the 1970s, a growing number of researchers accepted that there were limitations on the role of archaeology in establish-ing the historicity of the Hebrew Bible along the lines that Albright (1932) and others had pro-posed. William Dever called for a more 'secular' archaeology in the Holy Land (Israel, Palestine, Jordan) that should redefine itself as secular `Syro-Palestinian' Archaeology. This was an effort to shed the weight of the tyranny of the biblical text (Dever 1974, 1982) on the archaeological record of the southern Levant. Dever (1988) argued that freedom from the biblical text could be achieved by adopting the rapidly developing paradigm spearheaded by Louis Binford (1968) and known as the 'New Archaeology' with its emphasis on culture process—an approach that was specifically `anti-historical.' While the achievements of the New Archaeology are many and include the adop-tion of the scientific method, quantification, investigative optimism, the importance of research design over simple data collection and other features that have become a mainstay of world archae-ology today, by the early 1980s critics such as Ian Hodder (1982, 1987) showed many of the fail-ings of the New or Processual Archaeology. Self-appointing themselves as the new Tost-Processual' paradigm, Hodder, and others (Preucel and Hodder [eds.] 1996) pointed out that there was no single way to undertake archaeological inference as argued by the Processual archaeologists, that all interpretations were driven by the subjective views of the researchers, that even data is 'theory laden'—that is, many 'readings' are possible. The most significant Post-Processual critique, and most applicable to Levantine Archaeology, was the fact that Processual archaeology had an anti-historical bias that assumed a kind of 'universal humanism' making it possible to construct 'laws' of human behavior.

While Dever argued repeatedly for a 'secular' archaeology for the Holy Land that could be brought forth through the New/Processual Archaeology, this paradigm never really took off in Levantine Archaeology except in research embedded in the prehistoric and protohistoric periods (Levy 1996). Perhaps the notion that the Bible represents a kind of 'tyranny of the text' for archaeologists is simply inappropriate for the archaeology of the southern Levant where so much of the Hebrew Bible takes place. The leading historical archaeologists working in the southern Levant were primarily secular (Ben-Tor [ed.] 1992; Finkelstein 1988; Mazar 1990; Stager 1988); however, they could not ignore the centrality of the Hebrew Bible as a foundation—an ethno-historical source—for examining the archaeological record of the region. Historical archaeology (i.e. Middle Bronze—Iron Age) in the southern Levant did adopt many of the methodologies proposed by the New Archaeology, such as interdisciplinary research and a real interest in the application of new technologies for archaeological research. However, the question remained—how best to bring together text and archaeology. The emergence of the so-called Biblical Minimal-ist paradigm (cf. Davies 1992; Thompson 1999; Whitelam 1996) argued that the Hebrew Bible lacks any historical data whatsoever so it is a totally unreliable source. As discussed earlier (Levy and Higham [Chapter 1, this volume]), given the large number of interconnections between biblical and extra-biblical ancient sources (cf. Dever 2001, 2003), the Biblical Minimalist paradigm is untenable today. In some cases, adherents of this approach (Lemche 1998; Van Seters 1997) argue


for the centrality of ancient texts for historical reconstruction—but give precedence to any written text outside of the Hebrew Bible. When researchers grasp on to any historical piece of data uncriti-cally, whether it is the Hebrew Bible or extra-biblical textual data from media such as monumental inscriptions, ostraca (ink on pottery), engraved silver, inscribed stone seals or a seal impression, to interpret the archaeological record they run the risk of simplification and finding what their preconceived views want to find (Schniedewind 2004). For the past ca. 30 years, this is precisely what has characterized the Iron Age archaeology of southern Jordan, and in particular the region known from biblical and other sources (Bartlett 1989, 1992) as Edom. In what follows, we will illustrate how an over-reliance on extra-biblical textual data for ancient Edom has led to major chronological problems and consequently, problems with historical and anthropological interpreta-tion. We argue that only with the enthusiastic adoption of radiocarbon dating for the Iron Age archaeology of the southern Levant will it be possible to objectively investigate the relationship between the historical texts and archaeology for this period.

Research Area: Highland-Lowland Dichotomy

The region of Edom in southern Jordan extends roughly from the Wadi al-Hasa in the north to the Wadi Hisma and Jabal Ram in the south, the Wadi Arabah on the west and Transjordanian desert plateau to the east (Bartlett 1992; Glueck 1940a). The two most important physiographic attri-butes of Edom include: (a) the presence of one of the richest copper ore deposits in the southern Levant (Hauptmann 2000) and (b) the marked geographic and environmental diversity between the 'lowlands' and 'highlands' of Edom. The differences between these two geomorphic zones are pronounced. For example, the edge of the highlands, overlooking the Wadi Arabah that separates modern Israel and Jordan, is characterized by elevations that reach over 1500 masl, a semi-arid landscape and pockets of Mediterranean rainfall zones with over 600 mm of average annual rainfall (Centre 2001). In contrast, the lowlands of Edom, with elevations reaching ca. -80 masl, is typical of the Saharo-Arabian desert phytogeographic zone with pockets of Sudanian flora (Danin 1983), with mean annual rainfall at less than 70 mm. This contrast in rainfall patterns between the highlands and the lowlands has made rainfed agriculture possible in the highlands and more limited agriculture (primarily with the aid of irrigation technologies) possible in the lowlands. More important, this environmental dichotomy had a profound effect on the need for herd ani-mals such as sheep and goats and their annual movements in search of grazing land. For the most part, since at least the Early Bronze Age (Adams 2003; Levy et al. 2002) human occupation in Edom has been characterized by nomadic or semi-nomadic populations who have searched for ways to integrate the exploitation of seasonal resources available in both the highland and lowland regions. Thus, it is impossible to understand human settlement and the history of Edom without conceptualizing and integrating these two physiographic regions. However, over the past three decades, archaeologists interested in the Iron Age of Edom have overlooked the significance of the `lowland'-'highland' dichotomy. Prior to the Jabal Hamrat Fidan Project (Levy 2002; Levy, Adams, and Najjar 1999, 2001), all the major Iron Age excavations in Edom took place in the highland zone (Bennett 1966b, 1977; Bennett and Bienkowski 1995a; Bienkowski 1990; Bienkowski and Adams 1999; Bienkowski and Bennett 2003). The lack of systematic archaeological exploration in the lowland zone also meant that the role of Iron Age copper production that took place in the lowlands was not fully investigated.

As part of the deep-time study of early metallurgy and ore procurement from the Neolithic to the Iron Age (Levy et al. 2001a), one of the goals of the Jabal Hamrat Fidan UHF) Project, made up of a team of international researchers under the auspices of the UCSD—DOAJ, has been to help fill in the Iron Age research gaps that have developed due to the 'highland bias' in the Iron Age


archaeology of Edom. The Jabal Hamrat Fidan is a narrow mountain range made up of Mon-zogranite (Rabb'a 1994) that stretches for ca. 8 km north/south along the eastern edge of the Wadi Araba and represents the 'gateway' to the copper ore rich district of Faynan. The research area includes some 280 km' and is west of the main Faynan valley where various archaeology teams from the Council for British Research in the Levant have carried out mostly surveys and some excavations (Barker et al. 1997, 1999, 2000). The main seasonal drainages in the JHF research area that have been intensively and systematically surveyed for archaeological sites include the Wadi Fidan (Levy et al. 2001a), Wadi al-Jariyeh, and Wadi al-Guwayb (Levy et al. 2003). In this study, we discuss the ramifications of the stratigraphic excavations and high precision radiocarbon dating for the largest Iron Age site in the Jabal Hamrat Fidan area, Khirbat en-Nahas. First, however, it is necessary to briefly review the chronological bias in the Iron Age archaeology of Edom and how that has affected interpretation of the evolution and history of the emergence of Edomite kingdom known primarily from the Hebrew Bible and some extra-biblical texts.

The Chronological Bias in the Iron Age Archaeology of Edom

Until quite recently, the Iron Age chronology of Edom rested on the discovery of a single clay seal impression discovered at the highland site of Umm el-Biyara during Crystal Bennett's excavations in the 1960s (Bennett 1966a, 1966b). The seal contains the name of Qos-Gabr and is known from the 7th-century BCE Assyrian annals of Esarhaddon (Prism B, ca. 673-672 BCE; Pritchard 1969: 291) and in the first campaign of Ashurbanipal (Cylinder C, ca. 667 BCE; Bienkowski 1992b; Pritchard 1969: 294). Using the concept of relative dating, scholars have taken the discovery of this extra-biblical text fragment to date the Iron Age pottery found in association with it at the Iron Age site of Umm el-Biyara. As Bienkowski (1992b: 99) pointed out some years ago, the seal impression of Qos-Gabr provides a terminus post quern' for dating the Iron Age pottery at Umm al-Biyara but did not indicate just how early the Iron Age pottery found in that assemblage dated back to in time. In fact, Bienkowski (1992b: 110) also alerted readers that unpublished radiocarbon dates from the German Mining Museum's soundings at Khirbat en-Nahas and radiocarbon dates that indicated much earlier dates for the Iron Age in Edom (ca. calibrated dates of ca. 1200-900 BCE with 'Midianite' pottery; see Levy et al. 2004). However, Bienkowski's caution and the later publication of the report of the soundings at Khirbat en-Nahas in German which included radio-carbon dates (Engel 1993; Fritz 1996) fell on deaf ears. Bennett's dating of the Iron Age in Edom to the 7th and 6th centuries BCE became the accepted standard for the Iron Age archaeology of this part of Jordan. A host of studies concerning Iron Age Edom were produced based on the assump-tions established by the relative dating of Umm el-Biyara (Bennett and Bienkowski [eds.] 1995b; Bienkowski 1995; Hart 1989; Oakshott 1978, 1983; Pratico 1985, 1993b) and even more recent studies continue to work under the late 7th-6th-centuries BCE assumption for the emergence of the Edomite kingdom (Bienkowski and Bennett 2003; Crowell 2004; Porter 2004).

The enthusiasm that Bennett's late dating of Iron Age Edom received from scholars in the late 1970s to the 1990s, was in part against the views of the American archaeologist Nelson Glueck who pioneered archaeological surveys in Jordan and Iron Age excavations in Edom (Glueck 1938, 1939a, 1940a). Glueck took a more traditional view of Levantine archaeology and tended to accept extensive texts in the Hebrew Bible as historical fact in a way that many researchers believed to he biased (Dever 2000). Working in Edom, Glueck firmly believed that the majority of Iron Age mining activities in the Faynan district that he documented could he dated to the 10th century BCE

l'crminus post quern—refers to the notion that a datable object provides only the date on or after which the archaeological sediment layer that contains it was deposited.


(1940a: 69) and 9th century BCE (1940a: 86) and directly related to biblical texts such as 2 Samuel 8.13-15, 1 Kings 22.45, 48-50, 2 Chronicles 20.1ff., and many more. In the early 1990s, working with published Iron Age ceramic drawings, Israel Finkelstein (1992a, 1992b) suggested that indeed there was ceramic evidence (collared rim jars) of an early Iron Age occupation in Edom that pushed back this occupation considerably earlier than the view of Bienkowski (Bienkowski 1992a) and others. To help solve this chronological debate, which has profound implications for under-standing the history and socio-economic processes that led to the rise of the Edomite kingdom—such as core-periphery relationships between Edom and the Assyrian empire on the one hand and Edom and neighboring small polities such as Israel and Judah—it was decided that as part of the JHF Project, large scale stratigraphic excavations would be carried out at the Iron Age copper production site of Khirbat en-Nahas.

Previous Fieldwork at Khirbat en-Nahas

Nelson Glueck's (1939a, 1940a) surveys in Edom were the first systematic investigation of the network of Iron Age metal production sites in the lowlands of Edom and recognized the centrality of the site of Khirbat en-Nahas (KEN) in that system.

Figure 10.1. Aerial view of Khirbat en-Nahas, Jordan

(courtesy of ROHR publications, Nicosia)


While the Czech Orientalist Alois Musil (1907) was the first to sketch the site in 1898, and the site was subsequently visited by Kirkbride, Horsfield, Head, and Fritz Frank (Frank 1934) before Glueck, it was Glueck who photographed and made detailed sketch maps of a full range of Iron Age metal production sites around KEN. Glueck assumed that the most important periods of metal production at Khirbat en-Nahas were during and after the reign of King Solomon (1940a: 60-61). The site was later cursorily surveyed by Burton MacDonald (1992). In the early 1990s, the German Mining Museum, under Andreas Hauptmann (2000), carried out technological studies at KEN (Engel 1993) and preliminary soundings at one of the buildings visible on the site surface (Fritz 1996). In addition to Building 200 (n 1 radiocarbon sample), three slag mounds were sampled around the perimeter of the site providing a total of 8 Iron Age dates: East—near Fritz's Building 200 (n = 4 samples); North—HD 10991 (n = 1 sample); and West—near the fortress gate (n = 3 samples).

The dates from the German Mining Museum (GMM) work at KEN (Hauptmann 2000: 66) have been re-calibrated using Ox-Cal v3.6 in Figure 10.2 here. Published in the 1990s and 2000 (Hauptmann 2000) the dates clearly indicate two major phases of metal production at KEN during 12th-11th centuries BCE and 10th-9th centuries BCE and highlight a much earlier Iron Age occupa-tion in Edom than suggested by many current researchers who focus on the highlands (Bienkowski 2001a, 2001b; Crowell 2004; Porter 2004). However, part of the problem with the GMM date sequence is that they are not tied to well-defined archaeological stratigraphy at KEN. Conse-quently, with the exception of the single date from Building 200 (HD 13978), all the dates come from industrial deposits that lack cultural material such as pottery, scarabs, ground stone, casting molds, and so on. This lack of association with cultural material may also have contributed to archaeologists paying little attention to the GMM KEN dates in constructing models of settlement and history for Iron Age Edom. The recent excavations at KEN by the JHF team (Levy et al. 2004) have resolved this problem by carrying out large-scale excavations in three different cultural contexts at the site: the fortress gate (Area A), a building devoted to metallurgical processing, and an industrial slag mound similar to those sampled by the GMM team.

The 2002 Field Work at Khirbat en-Nahas and New Radiocarbon Dates

Recently, the JHF team reported on a series of 10 high precision radiocarbon dates from the 2002 excavations at KEN and processed at the Oxford Radiocarbon Accelerator Unit (Levy et al. 2004). These included 4 dates from the Area A fortress, 4 dates from the Area S building complex and 2 dates from the Area M slag mound. These data and their analyses demonstrated how some of the chronological biases in the Iron Age archaeology of Edom could be surmounted with the aid of high precision radiocarbon dating. That study also showed that occupation began at KEN at least as early as the 11th century BCE and that the monumental fortress was built in the 10th century BCE. It also showed that complex societies existed in Edom that where heavily involved in the extraction of copper ore and production of copper long before the influence of Assyrian imperial-ism was felt in the region from the 8th-6th centuries BCE. To holster the study of KEN and its chronological position in the Iron Age of Edom, an additional 27 carbon samples were processed from KEN for dating at the Centre for Isotope Research, Groningen, the Netherlands and are reported on here (Table 10.1).


Figure 10.2. Calibrated radiocarbon dates from slag mounds and Room 200,

Khirbat en Nahas, Jordan (after Hauptmann 2000). HD 10991 = North slag mound, calibrated

910-820 BCE; HD 13978 = Building 200, calibrated 900-805 BCE.

In this section, we focus on presenting a snapshot of the archaeological deposits and the expanded number of radiocarbon samples processed at Groningen since the publication of our original suite of dates processed at the Oxford Radiocarbon Accelerator Unit (Levy et al. 2004). A detailed study of these dates is presented by Higham et al. (Chapter 11, this volume). Here we spotlight the cultural and historical implications of the dates.

The Iron Age Fortress at KEN (Area A)

The ca. 73 x 73 m= fortress at KEN had never been excavated before the 2002 field season and its dating was only speculative. To obtain an archaeological 'signature' of the fortress complex, we decided to focus our work on sampling what appeared to be the gate located on the western perimeter of the fortress. Although covered mostly in rough, possibly hewn, blocks of Burj Dolomite shale, the outlines of what appeared to be room cells or guard rooms could be detected protruding on the surface of rubble surrounding the gate (Figs. 10.3, 4). Broadly, 7 strata could be defined in Area A: A4, A3, Alb, A2a,N1b, and Ala. As A4 represents bedrock, we begin with the earliest construction phase—Stratum A3.


Figure 10.5. The two northern guard rooms exposed in the KEN gate, 2002.

After clearing much of the collapse from the gate house, it was clear that we had found a typical Four-Chamber Iron Age Gate (Fig. 10.5). In the interests of preservation we left more than half of the gate house unexcavated for the future and focused our excavations on the two most northern rooms in the four-room gate house. As will be shown below, the passageway or street between the two sets of guard rooms had been intentionally sealed in antiquity, so it was decided to leave this blockage unexcavated. A summary of the different strata found in and around the gate house follows from the earliest building layer (Stratum A3) through the main occupations phases (Strata Ala and A2b). Some of the problems and potentials of these new dates are discussed below.

Stratum A3, Gate, Founding Phase

The Four-Chamber Gate was founded in Stratum A3 above the sterile bedrock stratum found in the earlier Stratum A4. One of the problems in excavating in and around the gate is that the differ-ent excavation areas inside and outside this structure were not connected stratigraphically but linked together based on the relative similarity of the depositional sequence in this single exca-vation area. Thus, on some occasions the possibility that later deposits infiltrated into lower stratum can not be ignored. Consequently, the date GrA-25320 (calBc 895-825) for Stratum A4 is too late for the basal layer at the site and conies from the main 9th-century BCE industrial activity that took place in this area when the gate was already abandoned. The four additional carbon samples selected for radiocarbon dating from Stratum A3 (Table 10.1: GrA-25318, GrA-25354, GrA-25321, GrA-25322) are also problematic but for the same and opposite reasons. GrA-25321 (calBc 835-795) and GrA-25322 (calBc 895-875, 835-800) seem to be from the basal portions of the 9th century BCE A2b industrial layer that were ascribed in the field to A3. Alternatively GrA-25318 (calBC 1210-1045) and GrA-25354 (calBc 1185-1180, 1125-945) are associated with 12th-11th-century BCE metalworking activities that took place in Area A before the construction of


the gate (see Figs. 10.6, 8-9). When these dates are coupled with the Stratum A3 date from the Oxford lab (OxA-12366 [calBC 1000-9851), the main construction phase of the four chamber gate falls within the early 10th century BCE. Clearly more stratigraphic excavations are needed in and around the Area A gate to clarify the construction date of this structure on a more definitive basis.

Based on the radiocarbon determinations from the Oxford suite (Levy et al. 2004), those from Groningen published here, and the stylistic similarities of the gate at KEN with those from other Iron Age south Levantine fortresses, it seems safe at this point to date the KEN example to the Iron IIA period. As seen in Table 10.2, the KEN gate is most similar to the well-known examples of Megiddo IVA, Beersheva V and III, Tel Dan, Ashdod 10, and Tell en-Nasbeh (Early)), The KEN gate has a facade of ca. 16.8 m and is smaller than the major Iron II settlement sites such as Megiddo IVA, Tel Dan, and Beersheva V, but on par with Beersheva III, Ashdod 10, and slightly larger than Tell en-Nasbeh (Early). The highly specialized nature of the KEN fortress which serviced the metal production activities of 10th-9th-century BCE Edom is thus enigmatic. The gate can be considered a mid-size example of the typical Iron IIA examples found in the southern Levant that was linked to a single role—helping to ensure the operation of copper production at KEN at the beginning of the Iron IIA period. If the perimeter of the KEN fortress is compared with other 10th-9th-century BCE fortresses in southern Israel, Jordan, and the Sinai Peninsula (Table 10.3), at ca. 73 x 73 m2, the KEN fortress is one of the largest fortifications from this period in the southern Levant. The closest parallel to the KEN gate and fortress complex is the one excavated by Glueck (Glueck 1938, 1939b, 1940b) at Tell el-Kheleifeh near Aqaba on the Red Sea which he dated to the 10th (Period I), 9th (Period II), and 8th (Period III) centuries BCE (Glueck 1993). Although Pratico (1993a, 1993b) has gone to great lengths to re-date Tell el-Kheleifeh to mostly the 8th-6th centuries BCE, the similarities between the KEN gate and fortress and Tell el-Kheleifeh, as well as the ceramic assemblages, are so striking that we suggest that in light of the corpus of '4C dates from KEN, the dating of Tell el-Kheleifeh needs to be reassessed once again in conjunction with future radiocarbon dates. Thus, a working hypothesis can be constructed that suggests that during the 10th and possibly the 9th centuries BCE, the KEN fortress played a pivotal role in the exploitation of copper ore and metal in the Faynan district and that it was part of an Iron IIA trade network that incorporated the early Hazevah fortress (Cohen and Yisrael 1995) on the western side of the Wadi Arabah with land trade routes leading to Israel and the Mediterranean and Tell El-Kheleifeh controlling seaborne trade to the south. As will be seen in the discussion of the following strata (A2b-A2a), considering the considerable energy that went into the construction of the KEN fortress, it is extremely puzzling that the KEN fortress seems to have had a relatively short-lived use. Additional excavations are needed in the gate to clarify the changing function of the gate and fortress.

Table 10.2. Characteristics of south Levantine Iron Age Four-Chamber Gates (Sources: Herzog

1992; Levy et al. 2004; A. Mazar [personal communication])

Site Facade

(m) Depth

(m) Passage

width (m) Depth of

Chambers (m) Width of

Chambers (m) Date of Construction

Megiddo IVA 25 15.5 4.2 3 8.2 Late 9th-8th century BCE

Beersheva V 20.8 12.6 4.2 3 6 End of 10th or 9th

century BCE

Beersheva III 16.6 13.6 3.6 3 5 Early 8th century BCE

Tel Dan 29.5 17.8 3.7 4.5 9 9th century BCE


Ashdod 10 16.5 13.75 4.2 2.4 3.8 End of 11th or Early

10th century BCE

Tell en-Nasbeh (Early) 15 12 4 1.8 4.4 No hard data

Khirbat en-Nahas 16.8 10.6 3.63 2.9 13 10th century BCE


Strata A2b and A2a, Gate

Stratum A2b. It is remarkable that shortly after the construction of the Khirbat en-Nahas gate, it seems to have gone out of use as defensive facility. There are two lines of evidence to suggest this. First, the passageway leading between the two sets of gate rooms was carefully filled with closing walls on each end of the passage, and a rock fill. As seen on the plan in Figure 10.7, both the western and eastern ends of the passage were closed in Stratum A2b with a stone wall. These closing walls, like the walls which indicate the location of the guard rooms of the gate house, were seen protruding out of the rubble fill covering the gate complex and easily mapped. Secondly, the eastern Stratum A2b closing wall was fully exposed during the excavations around this part of the gate revealing the careful work done to ensure the closure of the passage. As seen in Figures 10.11 and 10.12, this blockage was carefully constructed with tons of rock fill placed behind it to fill in the passage and render it useless. In addition to closing the passageway through the gate, the Stratum A2b activities included using the former guard rooms for smelting and other metal processing activities. Three new Groningen dates (GrA-25314 [calBC 895-825], GrA-25315 [calBC 895-825], and GrA-25316 [Table 10.1; calBC 1005-905), along with the Oxford date (OxA-12367 [calBC 900-875]) show that during the early 9th century BCE the gate and fortress ceased to have a military function.

Strata A2a. There may have been a brief abandonment phase in Area A (Fig. 10.7) at KEN following the heavy Stratum A2b metalworking activities at the site. The stratigraphic development of the site shows that a second metal processing layer is superimposed on Stratum A2b that has been labeled Stratum A2a. This stratum represents more scanty evidence of metal production with a series of rather ephemeral stone built installations attached around the perimeter of the gate including those areas directly in front of the passageway. Three new dates were processed at Groningen (Table 10.1): GrA-25311 (calBC 895-825), GrA-25312 (calBC 890-885, 835-800), and GrA-25334 (calBC 1210-1010). When these dates are considered along with the single date processed from this stratum in the Oxford lab (OxA-12368 [calBC 900-805]), it is clear that GrA-25334 is too old and represents an earlier piece of charcoal that was mixed in with material from this later stratum. As this stratum is characterized by many pit installations, the mixing hypothesis seems most accurate. Thus, Stratum A2a represents a late 9th-century BCE phase of metal produc-tion.

In summary, the expanded sample of radiocarbon dates for Area A at KEN indicates that in the 12th-11th centuries BCE, prior to the construction of the gate, Iron Age metal production activities had already begun at KEN. During the 10th century BCE, the Four-Chamber Gate was constructed in Stratum A3 and used for less than a century. In the following Stratum A2b, in the early 9th century BCE, the four-room gate ceased to function as a gate, was filled in and used for metal processing activities. By the end of the 9th century BCE, a second, more ephemeral phase (Stratum A2a) of metal production took place in and around the abandoned gate complex.


Metallurgical Processing Building (Area S)

In our recent overview of the 2002 excavations and surveys in the Faynan district, we (Levy et al. 2003) described the Iron Age settlement along the Wadi Guwayb where Khirbat en-Nahas is located and one of its tributaries, the Wadi al-Jariyeh, as an 'Iron Age landscape' that reflected the organization and power of copper production in the Faynan district at this time. When examining the distribution of building remains across the site surface at KEN as seen in Figures 10.1 and 10.13-14, it can similarly be described as an Iron Age setting reflecting the power, organization, and fabrication of copper during the Iron I and Iron IIa-Ilb periods writ large at a single locale. KEN is an extraordinary site in that virtually 100 per cent of the building remains visible on the site surface are associated with Iron Age ceramic remains. Based on the excavations by both the German Mining Museum (Fritz 1996) and those of the UCSD—DOAJ (Levy et al. 2004) teams, there seems little doubt that these buildings which extend over an area of ca. 10 ha all date to the Iron Age and almost reflect the very rare 'Pompeii principle' described by Michael Schiffer (1987) which relates to an archaeological site with an unusual degree of preservation, little evidence of post-depositional disturbance, and the documenting of 'frozen moments in time'. While KEN has certainly suffered from the vicissitudes of time, the later Iron Age strata at the site (ca. 9th century BCE) seem to represent abandonment where many artifacts were left in situ. Taken together, KEN represents a unique record of a Near Eastern Iron Age metal production factory town. The excava-tions in Area S represent the first systematic stratigraphic excavation of a building complex at KEN based on digital recording methods (Levy et al. 2001b) to ensure the utmost accuracy in data acquisition and analysis.

Stratum S4. This stratum represents the basal layer in Area S where several small (possible) cooking installations and isolated pockets of metal processing were found. Three new radiocarbon dates were processed from Stratum S4 (Table 10.1: GrA-25348 [calBC 970-835], GrA-25349 [calBC 1000-865], GrA-25352 [calBC 1005-900]). The first date processed from this stratum in the Oxford lab (OxA-12169 [calBC 1130-1015]) indicated a 12th-11th century BCE date. However, the new determinations point to the 10th-9th centuries BCE. Given the presence of Iron I scarabs from Area S (Levy et al. 2004), we suggest that the carbon samples collected from the small exposures in this area came from later contexts and that Stratum S4 should be dated to the Iron I period. This problem can only be resolved through more controlled excavations in other building complexes at KEN (see Figs. 10.13, 14).

Stratum S3. The most notable aspect of Stratum S3 at KEN is that it represents the first large scale evidence of metal smelting and processing in this part of the site. A thick layer of slag varying from 20-40 cm in thickness was found running under the entire S2b building structure in Area S (Fig. 10.16). This clearly shows that massive metal production occurred at KEN before the construction of many of the smaller 9th-century BCE buildings at the site such as the one in Area S and Building 200 (Fritz 1996). Two new radiocarbon dates were processed in Groningen from Stratum S3: GrA-25353 (calBC 1040-900) and GrA-25347 (calBC 1045-915) calibrated to the 11th and 10th centu-ries BCE. The Oxford date (OxA-12342 [calBC 1005-965]) is similar (see Higham et al. [Chapter 11, this volume]). The presence of a metal leaf-shaped arrowhead (Fig. 10.17) typical of the Iron I period (Mazar 1990) may add additional evidence to placing this industrial layer in the very late 11th-early 10th centuries BCE. Thus, it is possible that the widespread metal processing that occurred in Stratum S3 (Area S) was contemporary with the main occupation and use of the KEN fortress (Stratum A3) before it went out of use in Stratum A2b.


Figure 10.16. Stratum S3 slag layer (L. 351) found below the building in Area S.

Figure 10.17. Leaf-shape metal arrowhead found in L. 344, Stratum S3.

Stratum S2b. Stratum S2b represents the main construction and use phase of the building in Area S. The building is a relatively small structure measuring ca. 6 x 10.20 m divided into four rooms (Fig. 10.18) and was primarily used to process slag from the nearby smelting operations. Based on over 350 ground stone artifacts including grinding slabs, mortars, pestles, and other objects found in association with thick deposits of crushed slag around the perimeter of this building, we assume slag was intensively crushed here to retrieve as much residual copper embedded in the slag as possible. The original Oxford date (OxA-12168 905-830]) and two recently processed dates from Groningen (Table 10.1: GrA-25344 [caltw 970-835]; GrA-25345 [calBC 995-840]) suggest that the building was constructed and used primarily during the mid-9th century BCE. As seen in Figure 10.2, the single radiocarbon determination from Building 200 (HD 13978 [calEC 900-805]) also falls in the mid-9th century BCE. As a working hypothesis we suggest that after the fortress went out of use, there were widespread building activities at KEN when many of the structures visible on the site surface were constructed (see Figs. 10.1, 13).

Figure 10.7. Area A, Stratum Ala and ATh.


Stratum S2a. This stratum represents a relatively short period of construction to enlarge the initial building plan of Stratum S2b. As seen in Figure 10.19, courtyards were added to the northeast of the four-room building in Area S and a second phase of surfaces was found inside the structure reflecting a re-use of the building. The original Oxford date (OxA-12274 [calBc 895-875]) has been augmented with four new Groningen dates (Fig. 10.19, Table 10.1; GrA-25343 [calBC 900-825], GrA-25332 [calBC 895-830], GrA-25331 [calBC 1005-920], GrA-25329 [calBC 895-825]). Given that one (GrA-25331) out of the five dates from this stratum is a somewhat earlier anomaly, we assume that it may represent an old wood problem. The Bayesian analysis of all these dates is presented by Higham et al. (Chapter 11, this volume). However, a general assessment of the suite of S2a dates suggests a mid-9th-century BCE occupation for this stratum very close in time to S2b.

Stratum S / . In this chapter, we present the first series of radiocarbon dates for Stratum S I at KEN (Table 10.1). As this stratum was somewhat disturbed by post-depositional formation processes, we originally shied away from attempting to use radiocarbon to date this horizon (Levy et al. 2004). However, given the rich quantity of artifacts found in Stratum S1, we felt compelled to produce a relatively large sample of dates (GrA-25342 [calBC 1000-895]; GrA-25328 [calBC 890-885, 835-800]; GrA-25326 [calBC 900-835]; GrA-25325 [calBc 895-810]; GrA-25324 [calBC 895-830]). Stratum S1 activities occurred after an abandonment of the S2b-S2a occupation when there was an intentional in-filling of the four-room building to make a large enclosure. A scarab depicting a hunting scene was found in a fill from S1 (Basket: 6438, Locus 316; Levy et al. 2004: 875) and cannot be used to definitively date this specific stratum at KEN. Thus, at first glace these dates indicate a mid- to late 9th-century BCE date for Stratum S1. A clearer picture of this suite of dates is presented in the Bayesian analysis by Higham et al. (Chapter 11, this volume).

Area M—Slag Mound

Perhaps the most ubiquitous feature at KEN is the more than 30 extensive mounds of slag that mostly frame the perimeter of the site (Fig. 10.1; Levy et al. 2004: 869). The German Mining Museum team made soundings in three slag mounds at KEN and sampled examples on the west, north, and eastern boundaries of the site (Engel 1993; Hauptmann 2000). As noted above, these soundings produced an important series of dates (Fig. 10.2) that point to two major phases of metal production at KEN during the 12th-11th and 10th-9th centuries BCE. These soundings were made quickly by roughly sectioning the mounds so that carbon samples could be procured and an immediate picture of the phases of smelting revealed. While Engel (1993) suggested four major phases of smelting took place at KEN, our new data suggests more caution in making such an assessment. Using careful stratigraphic excavation methods, a ca. 5 m x 2.5 m sondage (Fig. 10.21) was dug through a slag mound located ca. 30 m south of Area S. This 6-week excavation was able to penetrate and accurately record only 1.2 m of the ca. 5 m deep slag mound. Within this expo-sure a minimum of 7 phases of smelting episodes was defined on the basis of identifying layers of large flat tap slag running across the section (Fig. 10.21). The radiocarbon dates from the top of this section (OxA-12436 [calBC 829-801]) and the base of this section (OxA-12437 [calBC 910-886]) suggest that in this relatively shallow sample of the slag mound there were two phases of smelting from the early 9th century BCE and early 10th-late 9th centuries BCE. This comes as no surprise given the findings of the Germans (Fig. 10.2). However, the excavations in Area M highlight the need carefully to excavate this (or another) slag mound to virgin soil so as to identify the full sequence of copper ore smelting activities during the Iron Age in Faynan.

Perhaps the biggest surprise from the Area M excavation was the discovery of part of a large stone built building in close association with the slag mound (Figs. 10.21-22). As only a corner of the rectilinear building could be exposed (Fig. 10.22), it was impossible to trace a floor interior


that might inform about the function of the structure and its relationship to the industrial activities at the site. However, a similar relationship has been documented for a later period site in the Faynan district, less than 2 km from KEN. Long ago, Glueck (1940: 66) described the nearby Medieval Islamic metal processing site of Khirbet Neqeib Aseimer as consisting of a large rectangular building with large deposits of slag abutting and surrounding it. Like Khirbet Neqeib Aseimer, the Area M rectilinear building and other similar occurrences at KEN were no doubt closely connected to the actual smelting process. This discovery adds another dimension to the high degree of Iron Age industrial specialization that took place at KEN.

Figure 10.21. Overview of excavations in the Area M slag mound at Khirbat en-Nahas, 2002. Note corner of building behind survey rod.



Figure 10.22. Detail of interior of the corner of large building found imbedded

in and around the Area M slag mound. The foundations of the building are

preserved to more than 2 m in depth.

Conclusions and Some Considerations of Iron Age History

For those of us working on the Iron Age archaeology of the southern Levant outside of Israel-Palestine, in neighboring regions such as Edom, the application of radiocarbon dating as an essential element in the tool box of archaeologists is now essential. As discussed here, the previous assumptions and dependence on the relative dating of ceramics linked to a paucity of extra-biblical textual discoveries is no longer tenable for Edom. Scholars working in other regions in Jordan, such as Moab (Harrison [Chapter 12, this volume) have also acknowledged that we have passed the point of no return and must employ high precision radiocarbon dating to critically test the relationship between history and archaeology in our region. In many respects, the bar has been raised for Iron Age archaeology by the recent publication by A. Mazar's team (Bruins, van der Plicht, and Mazar 2003a) of the radiocarbon dating project of Iron Age levels at Tel Rehov in the journal Science, which attempts to document evidence for the destruction of Iron Age towns in Palestine by the Egyptian pharaoh Shoshenq/Shishak I during the Iron IIA period. Whether Mazar's team is correct or not (Bruins, van der Plicht, and Mazar 2003b; Finkelstein and Piasetzky 2003; and in this volume, see Mazar et al. [Chapter 13], Sharon et al. [Chapter 6]; Bruins et al. [Chapter 19]) is immaterial—it is now impossible to carryout Iron Age historical archaeology without reliance on the object framework offered by `4C dating methods.


The recent excavations at Khirbat en-Nahas show conclusively that Iron Age social complexity, and perhaps the emergence of the kingdom of Edom known from biblical texts began some 200-300 years earlier than previously assumed (Bennett 1992; Bennett and Bienkowski [eds.] 1995b; Bienkowski 2001a, 2001b). It is not necessary to look to a core civilization (Assyria or Egypt) to explain the rise of the Edomite kingdom (Porter 2004)—we should look for local processes of change, especially the relationship of the small neighboring polities such as Israel and Judah with Edom at the end of the Late Bronze Age, Iron I, and early Iron II periods. For ancient Edom, the key to the emergence of social complexity is in what happened in the lowlands—in the Faynan district, close to the rich copper ore resources. With the recent large scale excavations at KEN, there is now evidence that control of copper production and trade in copper was probably the main catalyst for the rise of social complexity in Iron Age Edom. While many researchers (Bienkowski 1992a; Bienkowski and van der Steen 2001; Finkelstein 1992a; Knauf-Belleri 1995) have argued that large scale trade in other goods, especially from Arabia, was the key factor in the rise of Edom as a state, this assertion has not been demonstrated with archaeological evidence—certainly not on the scale of the metallurgical evidence discussed here.

The architecture, in situ excavations of copper industrial remains and imports confirm two major phases of production in the 12-11th centuries BCE and 10-9th centuries BCE at Khirbat en-Nahas. We are now at the beginning of being able to engage the Hebrew Bible and extra-biblical sources for gleanings of historical fact and historical processes. However, the current suite of 37 radiocarbon dates from KEN are not without problems (see Higham et al. [Chapter 11, this volume]) and it is clear that many more samples must be tested from sealed archaeological deposits associated with 'cleaner' assemblages of ceramics, scarabs, seals and other archaeological evidence. While the current dates push the occupational history of Edom back to the 12th-9th centuries BCE, the sample size is too small to confront the arguments concerning the High and Low Chronologies for the Iron Age in Israel/Palestine. These dates do bring the Iron Age archaeology of Edom back, to a certain degree, to historical questions raised long ago by Nelson Glueck (1940) concerning the Iron I and Iron Ila. While lack of space prevents a detailed discussion here, the fact that Edom is mentioned no less than 99 times in the Hebrew Bible justifies a re-examination of some historical issues in relation to the new archaeological excavations in the lowland region to establish some working hypotheses for the Iron Age history of Edom. For example,

ntt2 1^1t1 "11171 '.B1L7tt 7'171.t 1C,ii"11 1S'N1 WTT Genesis 36.21

These are the kings who reigned in the land of Edom, before any king reigned over the Israelites. (RSV)

Leaving aside the problem of the dating of the Hebrew Bible and the documentary hypothesis (Friedman 1988), Genesis 36.21 may be a minor footnote in the biblical text; however it does not exhibit an ideological stance, it is a neutral statement. While so-called Biblical minimalist scholars (Davies 1992; Thompson 1999; Van Seters 1997; Whitelam 1996) argue that the Hebrew Bible is pure myth lacking evidence of historicity, it is precisely in these 'footnotes' in the Hebrew Bible, which have no propaganda value or theological message, that some elements of history may be found. Baruch Halpern (Halpern 2001: 124-32) refers to the role of 'minimal text' in ancient his-torical documents for revealing historical events as 'the Tiglath-Pileser principle' and he presents a kind of historiographical method for how actual events can be gleaned from a critical reading of the ancient documents (in this case, both Assyrian and biblical sources). Thus, for the first time in biblical Edom, archaeological investigations at the lowland site of KEN provide radiometric data, scarab, arrowhead, Midianite ceramics, and other archaeological data to suggest a major industrial phase in the Iron I period. As this 12th-11th-centuries BCE metal production could only have been organized by a complex polity. While the RSV translation of the Hebrew 'Z1L7ti is given as 'kings' it may better be translated as 'chieftains', perhaps along the lines of the complex chiefdoms referred to by Sahlins (1968: 24-25) as chiefdoms organized along conical clan lines such as among the nomads of Central Asia, the island societies of Polynesia and Micronesia, in Circum-Caribbean America societies and the Southwest African Bantu. Whether we call the early Iron Age society that inhabited the lowlands of Edom 'chiefs' or 'kings' is immaterial; the point is that Genesis 36.31


probably refers to the 'hereditary leaders who reigned in the land of Edom, before any hereditary leaders reigned over the Israelites'—a seemingly insignificant footnote in the biblical text that may help contextualize the socio-economic dynamics that existed in Edom during the 12th-11th centuries BCE. Another example is:

crwL,,= '77'1 =4:^M Mr: C1Itt17:: GrINZ nr1W11 2 Samuel 8.14

:1L777 117--nr 777, 71t7

And he put garrisons in Edom; throughout all Edom he put garrisons, and all the Edomites became David's servants. And the LORD gave victory to David wherever he went. (Rsv)

As shown above, the earliest monumental building activity documented to date at KEN is the Iron Ha four chamber gate and fortress complex. Given the tapestry of different ethnic groups who occupied southern Canaan at the end of the Late Bronze-early Iron Age, which group may have been responsible for the construction of the KEN fortress? According to Halpern (2001: 4), the historical David appears in the books of 1 and 2 Samuel, and dies in 1 Kings 2. 2 Samuel 8.13-15 suggests that David's troops subjugated Edom and as illustrated above (2 Samuel 8.14), established garrisons all over Edom. Glueck (1940: 84-85) used these passages to suggest that David controlled the mines in the Faynan district and that this exploitation continued under Solomon. According to the biblical text (see below, 1 Kings 22.47) Israel ruled Edom through a deputy administrator (Na'aman 2004) whose place of residence is not known.

:t7n :4: :'17N: t7?-;1 w171 Kings 22.48

There was no king in Edom; a deputy was king. (Rsv)

According to the traditional High Chronology (Rogerson 1999), the rule of these two kings would be from ca. 1000-931 BCE. According to these data, and the suite of radiocarbon dates now available from KEN, several working hypotheses may be suggested for the possible builders and controllers of the Stratum A3 gate and fortress complex: (a) David, (b) Solomon, (c) David and Solomon; or (d) the local Edomite population. Clearly, more data and analyses are needed to clarify this working hypothesis. According to the biblical text, following the death of Solomon and the emergence of the divided monarchy of Israel and Judah, it is inferred that the Edomites finally gained their independence from Judah during the reign of Jehoram which according to the traditional High Chronology dates to 848-841 BCE (Rogerson 1999) or the mid-9th century BCE.

:t77.; 11"7nsi 77-71'-7 nrcnt; C1-12 wirl- 2 Kings 8.20

In his days Edom revolted from the rule of Judah, and set up a king of their own. (Rsv)

How to link the expansion in mid-9th century BCE metal production observed at KEN in the flurry of building activities seen in the Stratum S2b building (Fig. 10.18), the slag mounds (Fig. 10.10 and Building 200; Fritz 1996) to the biblical text? One working hypothesis is that following the Edomite revolt against Jehoram, the local Edomite population took over metal production at KEN, had no need for the garrison/fortress at the site and changed the organization of production at the site from one that was based on coercion (via the fortress) to an as yet undefined alternative form of organization.

While these are untested working hypotheses relating ancient Near Eastern texts to the archaeological record of Edom, they will 'only be tested adequately through larger scale excava-tions at Khirbat en-Nahas, in-depth studies of the full array of material culture represented at the site, and a much larger compendium of radiocarbon dates. How exactly does the copper ore-rich 12th-9th-centuries BCE metal producing region of Edom relate to the highland sites such as Busayra, Umm al-Biyara, Sela, and others?


The excavations at KEN and the radiocarbon dates originally published in the journal Antiquity (Levy et al. 2004) have sparked a great deal of welcome scholarly controversy on the Internet (<>) and most recently in an article by Israel Finkelstein (2005) published in the journal Tel Aviv. However, as shown in this chapter with the publication of an additional 27 high precision radiocarbon dates accompanied by a more detailed discussion of the archaeology at the site, the data run contrary to Finkelstein's assertions. In short, we can close by concluding with a number of points that are contrary to Finkelstein. Our data indicate: a) the fortress at KEN is 'sandwiched' stratigraphically between two metal production horizons at the site, with the latest production layer dating to the mid-9th century BCE and do not indicate the suggested 8th century BCE domination of Edom; b) The fort was not cut into piles of copper industrial waste—these slag deposits, based on our excavations, accumulated around it; c) Copper production was especially active at KEN throughout the 10th to 9th centuries BCE—it did not shift to other neighboring sites. Radiocarbon dates from other secondary centers like Khirbat al-Jariyeh (Hauptmann 2000:66) show contemporaniety between KEN and its satellite sites. This is seen in the statistical analysis presented above, that is, if the Strata A3 and S3 are indicative of increased copper production, then this dates to after 900 BCE at A3 in the fortress area and after 950 BCE at S3 in the industrial building complex. In fact, the new KEN data, in conjunction with Hauptmann's work (2000) support the recent interpretations of the Iron Age settlement pattern data from the Wadi al-Guwayb and Wadi al-Jariyeh for a complex network of copper ore extraction and processing (Levy et al. 2003); d) the fortress at KEN did not exist during the 8th century BCE so it was not contemporary with the Assyrian palace compound at Busayra. Rather than trying to make KEN conform to preconceived models that posit Assyrian domination of Edom in the 8th and 7th centuries BCE, the new data show a much more complex situation between the lowlands and highlands of Edom and its relationship with neighboring regions throughout the Iron Age. Thus, more archaeological research is required before definitive historical interpretations can be made.


We are grateful to Dr Fawwaz al-Khraysheh, Director General of the Department of Antiquities of Jordan for his sage advice and support of the excavations at Khirbat en-Nahas. Thanks also to the Society for the Conservation of Nature in Jordan for permission to work in the Dana Nature Reserve where KEN is located. Thomas Levy is grateful to the C. Paul Johnson Family Charitable Foundation (Napa and Chicago) and the University of California, San Diego for providing him with the grants and other funding that made the excavations at KEN possible. We also appreciate the help of Dr Russ Adams (co-PI and ceramicist of the JHF project), Dr Jim Anderson (chief sur-veyor), supervisors Yoav Arbel, Lisa Soderbaum, Elizabeth Monroe, the entire JHF team and Sheik Abu Shushi and the Bedouin villagers at Qurayqira for all their support.


Adams, R.B. (2003) External Influences at Faynan During the Early Bronze Age: A Re-analysis of Building 1 at Barqa el-Hetiye, Jordan. PEQ 135: 6-21.

Albright, W.F. (1932) The Israelite Conquest of Canaan in the Light of Archaeology. BASOR 74:11-23.

—(1971) The Archaeology of Palestine (repr., Gloucester, MA: Peter Smith).

Barker, G.W., et al. (1997) The Wadi Faynan Project, Southern Jordan: A Preliminary Report on Geo-morphology and Landscape Archaeology. Levant 29: 19-40.

—(1999) Environment and Land Use in the Wadi Faynan, Southern Jordan: The Third Season of Geoarchaeology and Landscape Archaeology (1998). Levant 31: 255-92.

P' •



essentially do in fact

Conclusions and Summary

The dating work undertaken at KEN thus far in the gate complex in Area A and the slag processing building in Area S have demonstrated incipient copper production from at least the early 10th century BCE in the Faynan area. It is clear that the areas of KEN that have been excavated do not date to the 8th-6th centuries BCE. We obtained close agreement with shorter-lived specimens dated at the radiocarbon facilities at ORAU and Center for Isotope Research—Groningen. However, problems in obtaining universally short-lived species resulted in some samples, comprising probable old wood, being dated. The Bayesian analysis of the radiocarbon dates supported this, and showed that there were more outliers in the data compared with what one might expect were the variation purely derived from statistical uncertainty alone.

The results of the modeling show that there is an expansion in copper production evident at the site from about 950 BCE. The impressive gate structure in the fortress at KEN appears to have been utilised for its intended purpose for a brief period (ca. 5-10 years) after which it too was devoted to the processing of copper. In both areas, we determined an overall span of time of about 100-150 years. Towards the end of the 9th century BCE, activity in both areas ceased but further excavation and dating is required to determine whether this occurred across the site as a whole, or was more circumscribed.

Further work is planned. Many more dating samples are needed to more adequately test ques-tions rooted in historical problems such as those connected with archaeology and the Hebrew Bible. We are particularly interested in the identification of other botanical specimens for AMS dating and improving calibrated ranges in the models developed so far by selecting further material for analysis. The results presented here suggest that the adoption of Bayesian modeling for investigating the dating of sites in this and other regions is most profitable and enables higher levels of dating resolution to be obtained in the study of archaeological chronologies.


We are grateful to members of the ORAU and Groningen AMS laboratories for their careful labora-tory work on preparing the samples dated in this project.


Anderson, A.J. (1991) The Chronology of Colonisation in New Zealand. Antiquity 65: 767-95.

Bronk Ramsey, C. (1995) Radiocarbon Calibration and Analysis of Stratigraphy: The OxCal Program. Radiocarbon 37(2): 425-30.

—(2001) Development of the Radiocarbon Calibration Program OxCal. Radiocarbon 43: 355-63.

Head, M.J. (1987) Categorisation of Organic Sediments from Archaeological Sites. In Archaeometry: Further

Australian Studies, edited by W.R. Ambrose and J.M. J. Mummery (Canberra: Department of Prehistory,

Research School of Pacific Studies, ANU): 143-59.

Hedges, R.E.M., et al. (1989) The Oxford Accelerator Mass Spectrometry Facility: Technical Developments in Routine Dating. Archaeometry 31: 99-113.

Higham, T.F.G. (1994) Radiocarbon Dating New Zealand Prehistory with Moa Eggshell: Some Preliminary Results. Quaternary Geochronology (Quaternary Science Reviews) 13: 163-69.

Higham, T.F.G., and M.D. Jones (2004) Settlement and Chronology. In Change through Time: 50 Years of

New Zealand Archaeology, edited by L. Furey and S. Holdaway (Auckland: Publishing Press): 215-34.

Levy, T.E., et al. (2004) Reassessing the Iron Age chronology of Biblical Edom: New Excavations and 14C

Dates from Khirbat en-Nahas (Jordan). Antiquity 863-76.

McFadgen, B.G. (1982) Dating New Zealand Archaeology by Radiocarbon. New Zealand Journal of Science 25: 379-92.


21 Desert Settlement through the Iron Age Radiocarbon dates from Sinai and the Negev Highlands

Hendrik J. Bruins and Johannes van der Plicht


Iron Age desert settlements in the Negev Highlands and the adjacent area of north-eastern Sinai are still enigmatic. Various theories have been developed to explain these settlements, particularly concerning the majority of the fortresses that are built in an elliptical or irregular shape. Chronology is obviously a crucial factor in archaeological theory-building. The time factor in Levantine Iron Age archaeology used to be like pottery clay that could be moulded to suit various theories. Radiocarbon dating, notwithstanding its limitations, provides an independent and scientific basis for chronology, though quality control is essential. Radiocarbon dates are presented from Iron Age strata at Tell el-Qudeirat in north-eastern Sinai, and from Nahal Ha'Elah and Horvat Haluqim in the Negev Highlands. Our main conclusion is that the establishment of the elliptical fortresses and related settlements appears to predate the Solomonic period.


There are many remains of Iron Age settlements in the hilly desert of the Central Negev and adja-cent area of north-eastern Sinai. Detailed archaeological surveys in part of the region have so far uncovered about 350 Iron Age sites, containing 58 fortresses, 1195 dwelling structures, 360 animal pens, many cisterns, 30 threshing floors and 80 silos dug into the ground, as reported by Haiman (1994). The climate of the region is arid; the average annual rainfall ranges from about 125 mm in the north to 75 mm in the south (Bruins 1986).

Geoarchaeological excavations at the site of Horvat Haluqim proved beyond doubt the existence of rainwater-harvesting agriculture in the Iron Age (Bruins and van der Plicht 2004). This was to be expected, as the region is too arid for normal rainfed farming. The many threshing floors and silos (Haiman 1994), if indeed dating to the Iron Age, are evidence of ancient farming that could only have been conducted successfully through irrigation by runoff rainwater, received from the surrounding catchments, arrested on each field by built terrace walls across the wadi. Wheat and barley require some 300 mm of rainfall to produce a reasonable yield and fruit trees even more. Only runoff water supply from local catchments could add sufficient water in addition to the low amounts (ca. 100 mm) of direct rainfall on the fields, in order to reach moisture levels in the soil comparable to 300-500 mm a year (Bruins 2003).


The age of these Iron Age settlements, particularly the date and character of the fortresses, are controversial issues. A detailed review was given by Cohen (1986), who interpreted most sites to have been short-lived within the 10th century BCE: established during the reign of King Solomon and destroyed by Pharaoh Shishak (Cohen 1980; Cohen and Cohen-Amin 2004). A similar inter-pretation was given by Haiman (1994), who dated all Iron Age sites in the surveys to Iron Age II in cultural terms, to the United Monarchy in political terms, and to the period 975-925 BCE in chronological terms. All settlements came to an end with the Shishak campaign. Other viewpoints, based on ceramic, architectural and Biblical considerations, range from about the 13th to the 7th centuries BCE. Most scholars considered the sites to be Israelite in one way or another, but Rothenberg (1967, 1972, 1988) and Finkelstein (1984, 1988) suggested a non-Israelite origin of these settlements.

An early date was proposed by Rothenberg (1967, 1972, 1988, 1999) on the basis of his exten-sive excavations at Timna. Besides Egyptian and Midianite pottery, Negev-ware (Negbite) pottery was also found in the Timna excavations, dating on Egyptian evidence from the late 14th to the 12th centuries (Ramesside): 'We are led to suggest that these local inhabitants of the Arabah and the Negev are the Amalekites mentioned in the Biblical narrative' (Rothenberg 1988: 276). Follow-ing a hiatus, represented by wind-blown loess deposition, the Egyptians returned to Timna during the 10th-9th centuries BCE (22nd Dynasty), as Timna Layer I yielded pottery vessels of the 22nd Dynasty, as well as Negbite ware, but no Midianite ceramics. This return of the Egyptians is related by Rothenberg (1999: 163) to the campaign by Pharaoh Shishak in ca. 920 BCE.

Aharoni considered these settlements of Israelite origin, ranging from the 10th to the 7th centuries BCE (Aharoni 1967). He changed his interpretation in a later publication (Aharoni 1978), suggesting that the settlements were established by Saul in relation to his wars with the Amalekites in the 11th century BCE. Herzog (1983) proposed an association of the Central Negev sites with the settlement of the tribe of Simeon, suggesting a date in the 11th century BCE. Meshel (1979) considered both the 11th and 10th centuries BCE as options, relating the fortresses possibly to Saul or David. However, both Meshel and Goren (1992) emphasize the basic problem of dating the fortresses and the lack of unambiguous criteria.

Finkelstein (1984, 1988) proposed a nomadic origin of these settlements, established by local desert tribes. These nomadic goat-sheep pastoralists became sedentary, in his view, due to sup-posed economic changes in the south, related to a revival of mining activity at Timna and prosper-ous Philistine centres along the southern coastal plain and the Shephelah. Finkelstein (1984, 1988) dated the settlements in the Negev Highlands to Iron Age I in cultural terms and to the 11th and early 10th centuries BCE in chronological terms.

It is clear that the use of ceramics and other archaeological criteria for dating Iron Age settle-ments in the Central Negev is problematic, enabling a wide range of scholarly opinion and a wide time range. The matter was summed up succinctly by Barkay (1992: 324): 'It is difficult to reach a conclusion concerning the fortresses, as their interpretation depends on their dating, and their dating involves serious difficulties. In many studies, scholars appear to be caught up in a circular argument, in which the dating is based on the general interpretation given to the fortress phenome-non and is in turn used to support the proposed interpretation of the character of the fortresses'.

What about radiocarbon dating in the region concerning the Iron Age? Bruins (1986) initiated research in the early 1980s in the Tell el-Qudeirat area in north-eastern Sinai (Fig. 21.1), but focussed mainly on geoarchaeological issues. Nevertheless, some "C results from the tell did not support a 10th-century BCE date for the Early Fortress, which appears to be older. The differences found here between radiocarbon dating and archaeological age assessment in the Iron Age con-vinced us of the necessity to establish an independent radiocarbon chronology for Near Eastern historical archaeology, particularly for the Bronze and Iron Ages (Bruins and Mook 1989). The


17th International Radiocarbon Conference held in Israel in 2000, on the initiative of Israel Carmi of the Weizmann Institute of Science, stimulated further interaction between Near Eastern archaeology and 14C dating (Bruins, Carmi and Boaretto [eds.] 2001).

Figure 21.1. Location map of Tell el-Qudeirat, Nahal Ha'Elah, Horvat Haluyim

and Khirbet en-Nahas based on a satellite image of 6 April 1998.

Ceramics undoubtedly carry the mark of time, but this mark should be established by independent dating (van der Plicht and Bruins 2001; Bruins, van der Plicht and Mazar 2003a; Mazar et al. [Chapter 13, this volume]). Moreover, some ceramic types, such as Negbite ware cannot be easily defined in time by traditional archaeological approaches. Perhaps also other pottery types may cover longer time spans than perceived.

An archaeological investigation involving radiocarbon dating on the eastern side of the Arabah Valley was conducted by Levy et al. (2004) in the ancient mining district of Faynan (Biblical Edom) at Khirbat en-Nahas, which is the largest Iron Age copper-smelting site in the southern Levant. Hitherto it was assumed that Iron Age settlement in the region, as well as the establishment of the Kingdom of Edom, occurred in the 8th-6th centuries BCE. However, the 'C dates from Khirbat en-Nahas give evidence of Iron Age occupation already in the early Iron Age (ca. 1200-1000 BCE) and also in the 10th-9th centuries BCE. It is important to compare these "C dates from neighbour-ing south-western Jordan with the dates from Sinai and the Negev, presented in the current article.


The radiocarbon dates from Iron Age strata in north-eastern Sinai and the Central Negev were measured at the University of Groningen (Table 21.1). All samples were pre-treated with the acid/alkali/acid (AAA) method (Mook and Waterbolk 1985). The purified organic matter of each sample was subsequently converted into CO,. Conventional radiometry of the CO, gas was con-ducted by Proportional Gas Counter (PGC). The samples from north-eastern Sinai were all meas-ured in the 1980s by PGC, as the Accelerator Mass Spectrometry (AMS) facility in Groningen was not established until 1994. Recent small samples from Horvat Haluqim were dated by AMS. The CO, gas derived from those samples underwent additional treatment to convert it into solid graphite to enable AMS measurement (van der Plicht et al. 2000).

Table 21.1. Iron Age radiocarbon dates from north-eastern Sinai and the Central Negev.

Tell el-Qudeirat Fortresses

Woolley and Lawrence (1914-15) suggested associating the relatively well-watered area of Tell el-Qu deirat in north-eastern Sinai with Biblical Kadesh Barnea, the main place of sojournment of the ancient Israelites in the desert following the Exodus from Egypt (Figs. 21.1, 2). Though many scholars have accepted the above suggestion, there is so far no independent evidence to confirm this viewpoint. The extensive excavations by Cohen (1980, 1981a, 1981 b, 1983, 1986, 1993a) uncovered three different Iron Age fortresses at the tell, which followed each other in time.


Geoarchaeological research by Bruins (1986) showed that the fortresses were established on a firm natural foundation: a Pleistocene deposit composed of slightly cemented coarse gravel and rounded boulders up to 40 cm in diameter. Similar deposits in the area contained Middle Palaeo-lithic discoid cores (Goldberg 1984). Prior to the building of the oldest fortress, the stony Pleisto-cene deposits probably appeared as a slightly elevated hillock rising just above the general level of the valley plain, which at this point is 164 m wide. This geomorphic surface provided a logical choice and firm foundation for the site-location of the successive fortresses in the valley of Wadi el-Qudeirat, as floods during the rainy winter season sometimes cover the entire width of the valley (Bruins 1986).

Figure 21.2. View of Tell el-Qudeirat looking SSW across the

width of the valley (photo by H.J. Bruins 1981).

The Upper Fortress

Dothan (1965) excavated a small part of the upper fortress in 1956. He dated the construction of the fortress to the 9th century BCE and its violent destruction to the Babylonian military campaign of the early 6th century BCE (Fig. 21.2). Cohen (1981a, 1983, 1993a) discovered the more complex situation that the Upper Fortress was built on the remains of the Middle Fortress. Cohen (1981a, 1983, 1993a) suggested dating the construction of the Upper Fortress to King Josiah ca. 640-609 BCE, and its destruction to the Babylonian onslaught, possibly coinciding with the destruction of Jerusalem and the First Temple in 586 BCE.

The uppermost destruction layer in the centre of the tell was exhibited clearly in the western profile of Square K-67. The black layer was situated at a depth of 50 cm below the surface of the tell. A sample of fine powdery charcoal mixed with soil was taken from this destruction layer by one of us (H.B.) in December 1981 in cooperation with Cohen, who considered this layer to represent the destruction of the Upper Fortress. The sample was measured in Groningen and yielded a radiocarbon date of 2535 ± 50 BP (GrN-12329).


The calibrated age relates to one of the most problematic sections of the calibration curve (Fig. 21.3), as the historical timescale of about 780-420 BCE (x-axis) corresponds to the same BP date on the radiocarbon timescale (y-axis), around 2500 BP. The calibrated date of GrN-12329 in the lo range is 796-758 (17.9%), 685-660 (9.9%), 647-542 (40.4%) BCE and the 2G range is 803-515 (92.8%), 464-450 (1.5%), 440-428 (1.2%) BCE. This result is quite meaningful, because the highest relative probability (40.4%) in the 1G range is in the period 647-542 BCE, which fits well with the period of the wars between Egypt (Pharaoh Necho II, 610-595 BCE) and the Babylonians (Nebuchadnezzar II, 604-562 BCE), in which Judah became involved, eventually leading to its destruction in 586 BCE. There are several correlations and synchronisms between Biblical and historical textual data concerning this period (Finegan 1979).

Figure 21.3. Dating of the uppermost destruction layer in Square K-67, related to the destruction of the Upper Fortress, based on fine charcoal.

The southern casemate room in the western side of the Upper Fortress yielded five complete storage jars in a thick ash layer. One of the jars was full of charred cereal grains (Cohen 1983). The sample received for radiocarbon dating (Square P5, Locus 523) was large enough for high-precision dating in the large PGC counter, resulting in a date for the charred cereals of 2515 ± 15 BP (GrN-15551, Fig. 21.4). Notice how well these two BP dates of destruction layers from dif-ferent parts of the Upper Fortress agree with one another: the results overlap within 1a. The charred cereal grains from a jar are 20 midpoint BP years younger than the powdery charcoal, which shows that the fine charcoal in the latter destruction layer is also rather short-lived. The la calibrated date of the cereal grains is 780-772 (4.3%), 766-760 (4.0%), 681-666 (10.9%), 634-591 (34.2%), 578-557 (14.7%) BCE and the 2G range is 788-757 (15.1%), 697-658 (16.5%), 649-542 (63.8%) BCE.

Atmospheric dab from Stover et al (1990), OxCal J3.9 Brook Ramsey (2003), cub r2 al 12 prob uspistrag


Figure 21.4. Dating of a destruction layer in Square P-5, related to the destruction

of the Upper Fortress, based on charred cereal grains found inside a jar.

The calibrated 20 results usually show less detail than the la calibrated date. The period 649-542 BCE has the highest relative probability (63.8%) in the former range. However, the high-precision BP date results in five possible calibrated periods in the la range, due to the plateau and wiggles in the calibration curve. Notice that the period 591-578 BCE is excluded in the la calibrated date, due to a small wiggle (Fig. 21.4). This result is perhaps significant, because it leaves out the date 586 BCE, during which Nebuchadnezzar destroyed Jerusalem and the First Temple. The period 634-591 BCE has the highest relative probability (34.2%) in the la range, which would favour, in fact, the destruction of the Upper Fortress at Tell el-Qudeirat in north-eastern Sinai during the earlier military campaign of the Babylonians, as they marched to Egypt in 601 or 600 BCE, according to Finegan (1979: 126).

The Middle Fortress

The outline of the Middle Fortress was similar to the Upper Fortress with a rectangular ground plan of ca. 60 x 40 m and eight protruding towers (Cohen 1983, 1993a, 1993b). The remains of the 4 m thick broad solid walls of the Middle Fortress were preserved to a height of about 1.80 m. An earthen rampart surrounded the fortress, resting on a revetment wall, 2.5 m high, which was completely excavated along the eastern side of the fortress (Cohen 1983, 1993a, 1993b).

A black ash layer was found on the eastern side in Square Q-9, sloping down from the revet-ment wall (Wall 207) at 12° towards the east. The black ash layer touches the revetment wall at a level of 18.96 m, which is about 60 cm above its base level of 18.35 m. The ash layer, ca. 10 cm thick, formed the ancient surface on top of a layer of fine yellowish-brown loessial sediment, about 60 cm thick, overlying the Pleistocene gravel deposits that also form the foundation for the revetment wall. The dark ash layer, a former living floor just east of the eastern revetment wall, seems to signify a destruction event that postdates the construction of the Middle Fortress on


stratigraphic grounds. If the revetment wall was built later than the formation of this destruction layer, a direct stratigraphic contact between the two would have been virtually impossible. It would have been necessary to remove part of the destruction layer and dig down to the gravel deposits to build the first course of the revetment wall. However, the direct stratigraphic contact is evidence that the destruction layer postdates the building of the revetment wall.

Figure 21.5. Dating of a destruction layer in Square Q-9, probably related to

the destruction of the Middle Fortress, based on fine charcoal.

A sample for radiocarbon dating was taken by Bruins (1981) from this ash layer at a level of 18.73 m and at a distance of 80 cm east from the revetment wall. The amount of fine charcoal was comparatively small, and the AMS facility in Groningen had not yet been established. Hence, the PGC date (GrN-11948) had a rather large standard deviation: 2740 ± 110 BP (Fig. 21.5). The result of the 1n calibrated age range is 1020-798 (68.2%) BCE. The 20 age range is very wide: 1259-1232 (1.0%), 1217-758 (91.3%), 642-588 (1.9%), 581-544 (1.2%) BCE. The relevant 1n result, indicating the narrowest time range with the highest relative probability, covers both the 10th and 9th centuries BCE. Cohen suggested that the Middle Fortress was built during the time of King Uzziah, ca. 769-733 BCE, and destroyed towards the end of the reign of Manasseh (ca. 698-642 BCE). The above radiocarbon date is clearly older, by about 150 to 200 years.

The Lower Fortress: The oldest archaeological remains discovered at Tell el-Qudeirat were found at a depth of about 5 m below the surface of the mound. The Lower Fortress had an elliptical ground plan, about 27 m in diameter, with casemate rooms around a central courtyard. In addition, several buildings and silos were found to the west of the fortress. Many types of pottery vessels were found in the ash covered floors of the casemate rooms (Cohen 1983, 1993a). The excavator (Cohen 1980, 1983, 1993a) suggested that the Lower Fortress was established during the reign of Solomon and destroyed in the course of Pharaoh Shishak's campaign, all in the 10th century BCE. The western profile of Square K-67 in the centre of tell el-Qudeirat, which exhibited the upper-most destruction layer 50 cm below the surface of the tell, also exposed the lowermost destruction layer at a depth of about 5 m. A sample of fine powdery charcoal mixed with soil was taken from this destruction layer by the first author, again in 1981, in cooperation with Cohen, who considered this layer to represent the destruction of the Lower Fortress. The dark ash layer, about 10 cm thick, covered a 20 cm thick layer of loessial soil, also containing a few pieces of charcoal, indicating past human activity predating the dark ash layer. Below the loessial soil lies a 'virgin' layer of fine gravel mixed with sandy loam (Bruins 1986). The fine charcoal sample from the ash layer was measured in Groningen and yielded a radio-carbon date of 2930 ±. 30 BP (GrN-12330, Fig. 21.6). The la calibrated age ranges are 1210-1200 (5.5%), 1191-1177 (8.1%), 1162-1141 (12.7%), 1131-1107 (13.3Vo), 1103-1050 (28.6%) BCE. The 2a calibrated ages are 1258-1235 (6.5%), 1215-1016 (88.9%) BCE. The most probable calibrated age range of 1103-1050 BCE would place the destruction layer in the first half of the 11th century BCE, which is about 150 years older than the suggested destruction, according to Cohen, by Shishak around 925 BCE. Alternative "C dating options, albeit of lower relative probability, include the 12th century and even the 13th century BCE, while the 11th century BCE is the youngest possible date in the 2o range. A possible old-wood effect of the charcoal is unlikely to move the date into the first half of the 10th century BCE, as this would require a lowering of the date by about 150 BP years. It was shown from the Upper Fortress at Tell el-Qudeirat that the difference between charred seeds and fine charcoal can be quite small, that is, only 20 BP years!


Figure 21.6. Dating of the lowermost destruction layer in Square K-67, probably

related to the destruction of the Lower Fortress, based on fine charcoal.


Nahal Ha'Elah Fortress

The site lies about 13 km north-west of the modern town of Mizpe Ramon and about 10 km due north of the Makhtesh Ramon cirque (Fig. 21.1). The fortress is located on a lofty hill (685 m) west of nahal Ha'Elah (nahal is the Hebrew word for a dry stream valley or wadi). A cistern is situated eastwards below the fortress, above the western bank of nahal Ha'Elah. Hillside conduit channels carried runoff water to the cistern. Remnants of an Iron Age settlement, consisting of 10 structures, including a 4-room house and single room dwellings, are situated 1 km north of the fortress (Cohen 1986).

Excavations at the fortress were conducted in 1983 by Cohen (1986, 1993b). The fortress has an elliptical shape, being 34 m long and 20 m wide. It consists of 13 casemate rooms and a gate, surrounding a central courtyard. The walls of local hard limestone were found to be 0.60 to 0.90 m wide, based on bedrock and still standing to a height of 1.50 m. Many of the casemate rooms were excavated, often showing a thin ash layer. Cohen (1986, 1993b) dated the fortress and its destruction to the 10th century BCE. A large sample of fine charcoal (28.6 g) was given by Cohen for radiocarbon dating, derived from one of the casemate rooms (Locus 1235/64). The weight of the sample was sufficient for high-precision measurement with PGC, yielding a date of 2840 ± 15 BP (GrN-15552, Fig. 21.7). The 10 calibrated age is 1006-972 (46.2%), 957-940 (22.0%) BCE and the 2o calibrated age is 1043-1028 (4.7%), 1023-969 (57.7%), 961-923 (33.0%) BCE. The date seems too old for association with the Shishak campaign (ca. 920 BCE), as favoured by Cohen (1980). But uncertainty regarding a possible old-wood age of the charcoal requires caution.


Horvat Haluqim Agricultural Terraces

The site of Horvat Haluqim is an excellent example of an Iron Age desert village, located along three parallel dry stream valleys (wadis) at the south-eastern slopes of the Haluqim Anticline, 2 km north-west of Kibbutz Sede Boker (Figs. 21.1, 8). The site comprises 25 structures, including a fortress, seven 4-room houses, other buildings and 4 cisterns. Cohen (1976) excavated the fortress and a number of buildings at the site. The elliptical fortress (23 m long and 21 m wide) has virtually the same size as the Lower Fortress at Tell el-Qudeirat (see above).

Cohen (1976) suggested that Horvat Haluqim was established, together with most other Iron Age fortresses and sites in the region, during the reign of King Solomon and destroyed during the campaign of Pharaoh Shishak. Thus, according to the above viewpoint of Cohen, the settlements were inhabited for only 30 to 40 years within the 10th century BCE. Much later in time, during the Roman period, two buildings were constructed at the site, probably in the 2nd-3rd centuries CE, according to ceramics and a coin found at the site (Cohen 1986).

Bruins (1986) carried out a survey in the three wadis of Horvat Haluqim and found more than 70 terrace walls with adjacent fields. Geoarchaeological excavations in the 12th terraced field of the eastern wadi led to the discovery of a buried anthropogenic layer (accumulative palaeo A horizon), beginning at a depth of about 45-50 cm below the present surface (Fig. 21.9). This anthropogenic soil layer has a remarkable thickness of about 75 cm, spanning the entire terraced field (Bruins and van der Plicht 2004). It has unique significance as an archive of past human agricultural activity at the site through time. This type of information is irreplaceable within the site, as such data cannot be obtained from the building or ceramic remains.

Figure 21.8. The elliptical fortress at Horvat Halugim, looking ESE at some of

the casemate rooms (photo by H. J. Bruins 2004).


Figure 21.9. Geoarchaeological excavations in the 12th terraced field (outline field marked in white of the eastern wadi at Horvat Haluqim, showing the location of the pits in which the dark anthropogenic soil layer was found at a certain depth below the present surface. The location of the fortress is indicated by the white elliptical line. The view is WSW towards the Zin Canyon and Avdat.

The anthropogenic layer yielded small Iron Age pottery sherds, small animal bones, charcoal flecks, as well as some pieces of flint in its lower part. Radiocarbon measurement of a bone from a sheep or goat gave an AMS date of 2860 ± 40 BP (GrA-14398). The la calibrated age (Fig. 21.10) is 1124-1121 (0.9%), 1111-1099 (4.7%), 1080-1061 (7.5%), 1052-971 (45.7%), 958-938 (9.4%) BCE, and the 2a range is 1189-1179 (1.9%), 1154-1142 (1.5%), 1129-915 (92.0%) BCE. The most likely period within all these options is 1052-971 BCE with a relative probability of 45.7% in the 1a range.

The above date, derived from a human-made agricultural soil layer, is the first independent scientific evidence that rainwater-harvesting agriculture at Horvat Haluqim was carried out rela-tively early in the Iron Age (Bruins and van der Plicht 2004). The bones entered the anthropogenic soil layer due to manuring practices by the ancient farmers. They used home refuse to improve soil fertility, as established by detailed micromorphological research by Bruins and Jongmans (to be published elsewhere). Bone dates from sheep or goats are based on short-lived organic material and no 'old-wood arguments' can be used in this case to advocate a younger age. Comparing this date with the detailed Iron Age radiocarbon sequence from Tel Rehov (Bruins, van der Plicht and Mazar 2003) suggests that we are dealing here in chronological terms with the later part of the Iron Age I period.


Figure 21.10. Dating part of the anthropogenic agricultural soil layer in terraced field

12 of the eastern wadi at Horvat Haluqim, based on a sheep or goat bone.

Figure 21.11. Dating another part of the anthropogenic agricultural soil layer in terraced field 12

of the eastern wadi at Horvat Haluqim, based on a speck of powdery charcoal.


A small but distinct charcoal fleck from another part of the anthropogenic layer gave a radio-carbon date of 2590 ± 60 BP (GrA-12448). The 51'C value of -17.75 %o is higher than for wood (usually around -25%o), which indicates that the charcoal is more likely derived from short-lived shrubs, annual plants or animal dung. The lo calibrated date (Fig. 21.11) is 828-759 (41.2%), 682-666 (6.6%), 634-591 (14.1%), 578-557 (6.4%) BCE and the 2o calibrated date is 896-875 (1.8%), 842-519 (93.6%) BCE. The highest relative probability in the 1a range is for the period 828-759 BCE. The result suggests human activity at the agricultural terrace also during the later Iron Age. Settlement in the Negev-Sinai region during the later Iron Age is also known from the Middle and Upper Fortress at Tell el-Qudeirat (Fig. 21.1) and the two successive fortresses at Hazeva in the Arabah Valley (Cohen 1993c). However, notice that the above date from the human-made agricultural soil at Horvat Haluciim is older than the two dates (charcoal and charred cereal grains) for the destruction of the Upper Fortress at Tell el-Qudeirat.

Discussion and Conclusions

Cohen (1983, 1993a) suggested that the Upper Fortress at Tell el-Qudeirat was built during the reign of King Josiah (ca. 639-609 BCE). If so, then its existence was short—probably less than 40 years. In 609 BCE, Pharaoh Necho II moved the Egyptian army through Judah on his way to Mesopotamia to assist the Assyrians against the Babylonians. Josiah decided to confront the Egyptians near Megiddo, but was killed in battle (2 Kings 23; 2 Chronicles 35). Later Johoiakim (608-598 BCE) was placed on the throne by Pharaoh Necho II, instead of Jehoahaz (2 Kings 23), indicating considerable Egyptian influence over Judah. Some of the ostraca found by Cohen (1983, 1993a, 1993b) in the Upper Fortress show both ancient Egyptian hieratic numerals and Hebrew words, which might fit the above political picture. In 605 BCE the Egyptian army in Mesopotamia was decisively defeated by the Babylonians in the famous battle at Carchemish near the Euphrates River.

The commander of the Babylonian troops, Nebuchadnezzar, became king in 604 or 603 BCE. He marched his army to Egypt in 601 or 600 BCE and fierce battles took place, as the Babylonians and the defending Egyptians inflicted heavy losses on each other (Finegan 1979). Johoiakim changed allegiance twice, first to Nebuchadnezzar and then back again to Necho, and he died in about 598 BCE. It is at this point that 2 Kings 24.7 gives an interesting geopolitical state of affairs: `And the king of Egypt did not come out of his land anymore, for the king of Babylon had taken all that belonged to the king of Egypt from the Brook of Egypt to the River Euphrates'. The fortress at Tell el-Qudeirat was the closest Judean presence near Egypt, located only 25 km west of Wadi el-Arish. Perhaps the Upper Fortress was destroyed during these fierce battles between the Babyloni-ans and Egyptians in 601 or 600 BCE, maybe punishing Johoiakim for changing alliance. This historical age fits somewhat better with the calibrated radiocarbon date (charred cereal grains in a jar) for the destruction of the Upper Fortress than the year 586 BCE, the Babylonian destruction of Jerusalem.

The "C date associated with the destruction of the Middle Fortress at Tell el-Qudeirat is con-siderably older (10th-9th centuries BCE in the 1a range) than the age suggested by Cohen (1983, 1993a) in the mid-7th century BCE. One radiocarbon date of a destruction layer outside the eastern revetment wall is certainly a reason to regard the result as preliminary with regard to the Middle Fortress. Yet the four radiocarbon dates of the three fortresses are internally coherent in terms of stratigraphy and must he taken into account.


In terms of possible regional correlations between architecture and governmental planning, it should he noted that both Stratum V and IV of Tel Beer Sheva had a solid wall, like the Middle Fortress at Tell el-Qudeirat. Casemate walls built on top of the remains of the previous solid walls occur at Tel Beer Sheva in Stratum III (Herzog 1993) and at Tell el-Qudeirat with the Upper Fortress (Cohen 1983, 1993a). Herzog (1993) suggested that Stratum V of Tel Beer Sheva-characterised by a solid wall—might have been destroyed by Pharaoh Shishak. The only "C date from Tell el-Qudeirat that might fit the Shishak campaign is the destruction layer associated with the Middle Fortress, which also had a solid wall.

The elliptical Lower Fortress was smaller and had a different shape than the rectangular Middle and Upper Fortresses at Tell el-Qudeirat, which are decisively younger in age. Most Iron Age settlements in the Negev-Sinai region are characterised by elliptical or irregular shaped fortresses, including Horvat Haluqim, Nahal Ha'Elah and the Lower Fortress at Tell el-Qudeirat. The most probable calibrated '4C date of 1103-1050 BCE for the destruction of the Lower Fortress is about 150 years older than the suggested date for its destruction by Cohen (1980, 1983, 1993a). The above '4C date would place the Lower Fortress firmly in the Iron I period, as favoured by Rothen-berg (1972, 1988), Aharoni (1978), Herzog (1983), Finkelstein (1984, 1988) and considered possible by Meshel (1979).

We note that the old-wood effect may lower the date to some extent at Tell el-Qudeirat. However, the powdery charcoal mixed with soil from the destruction layer of the Lower Fortress is generally not characteristic for old wood. Large trees of an old age tend to give chunks of recognizable woody charcoal, such as found often at Tel Dan. But even the radiocarbon results from such woody charcoal at Tel Dan are only rarely older than 50 or 60 years in comparison to short-lived seeds (Bruins et al. [Chapter 19, this volume]). Therefore, particularly in arid regions, usually devoid of trees, the inherent age of fine charcoal is in most cases probably not more than 10-30 years, or even much less. Annual vegetation growing after the winter rains withers in the spring. Burning of such vegetation would give short-lived powdery charcoal similar in age to seeds. Desert shrubs are older than annual plants and charcoal derived from such shrubs may have an age of ca. 2 to 20 years, occasionally even older, but on average below 10 years. Though the exception may always be present, a small to medium old-wood effect is probably the rule.

The destruction date for the elliptical Nahal Ha'Elah fortress in the Central Negev seems on the face of it too old for the Shishak campaign, as the most probable age range is 1006-972 (46.2%) BCE in the 10 range.

The 75 cm thickness of the anthropogenic agricultural soil layer at Horvat Haluqim, situated in a field terraced to catch runoff water, could not have been formed in just 40 years within the 10th century BCE. The Iron Age radiocarbon date on a small piece of bone from a sheep or goat, found in this soil layer, indicates that the terraced field was already established before the period of Solomon. The most probable age range is 1052-971 (45.7%) BCE, favouring the second half of the 11th century or the first decades of the 10th century BCE.

The younger Iron Age date on a small fleck of charcoal gives a most probable age range within the period 828-759 (41.2%) BCE and a second most likely age within the period 634-591 (14.1%) BCE. The result confirms that the thick anthropogenic soil layer was not formed in just one generation during the 10th century BCE, but could have begun in the 11th century (animal bone) or even well before. Agriculture and manuring continued, or were resumed after possible gaps, in the 9th, 8th or 7th centuries BCE (charcoal fleck). Micromorphological research of many thin sections from this soil layer showed the common presence of very small charcoal and bone fragments, even below 0.01 mm in size, pointing to home refuse as the source of the fertilizer.

Comparing the Iron Age 14C dates from Sinai and Negev with those from Khirbet en-Nahas in the eastern Arabah Valley in Jordan (Levy et al. 2004), it is quite remarkable that grosso modo [in a rough way] a similar BP time range is found for the older part of the Iron Age. The oldest dates are 2930 ± 30 BP (GrN-12330) in relation to the Lower Fortress at Tell el-Qudeirat and 2906 ± 39 BP (HD-14057) concerning the Slag Mount East (Hauptmann 2000). Moreover, also the period 2880-2825 BP appears in both areas. The oldest Iron Age date so far from the agricultural terrace at Horvat Haluqim (2860 ± 40 BP, GrA-14398) and a destruction date for the elliptical fortress at


Nahal Ha'Elah (2840 ± 15 BP, GrN-15552) are rather similar to three dates (2880 ± 28, HD-14302; 2876 ± 38, HD-14308; 2864 ± 46, HD-14113) from the Slag Mound West (Hauptmann 2000) and to Stratum A4a in Gate 2002 of the Khirbat en-Nahas Area A (2825 ± 32, OxA-12365). These chronological data are most significant, showing the regional scale of Iron Age settlement and activities in north-eastern Sinai, the Negev Highlands and the eastern Arabah Valley (Levy et al. 2004) for the early part of the Iron Age. These four sites are relatively close to each other, as can be seen in Figure 21.1.

In conclusion, the radiocarbon dates from the three successive fortresses at Tell el-Qudeirat are internally consistent in stratigraphic terms. The results indicate that the Upper Fortress was proba-bly destroyed by the Babylonian campaigns, as suggested by Cohen, though a 601/600 BCE histori-cal destruction date would fit better than the alternative 586 BCE option. The Middle Fortress appears older than suggested by Cohen. It is the only radiocarbon date that can possibly be linked, in chronological terms, with the Shishak campaign. The thick solid wall of this fortress appears similar in architectural construction to that of Stratum V of Tel Beersheba, the destruction of which is also associated with the Shishak campaign (Herzog 1993). The Lower Fortress at Tell el-Qudeirat and the Nahal Ha'Elah Fortress, both elliptical in shape, have destruction layer dates that appear older than the Solomonic period. The possible old-wood effect must be taken into consideration, but fine charcoal tends to be rather short-lived. If the old-wood effect is minimal, even the 12th century BCE is a reasonable option for the Lower Fortress at Tell el-Qudeirat in terms of its radiocarbon date.

Considering all the different theories proposed for the elliptical Iron Age fortresses and related settlements, briefly presented in the introduction, it seems that the suggested chronologies and historical associations by Cohen and Haiman are the most unlikely, while the 11th and early 10th centuries BCE appear most probable. However, even older dates for the beginning of these settle-ments cannot be ruled out, as the radiocarbon dates were derived from destruction layers. Indeed, the oldest date obtained so far, from the agricultural soil layer at the site of Horvat Haluqim, backs the above picture. Here the old-wood effect cannot be used as an excuse, because the date is based on a sheep or goat bone from within the anthropogenic agricultural soil layer. Nevertheless, more dates are necessary to substantiate and refine this preliminary radiocarbon dating assessment.


We thank Dr Rudolph Cohen and the Israel Antiquities Authority for their cooperation in the sampling and provision of the organic material and the permit for the geoarchaeological excava-tions at Horvat Haluqim. Despite disagreements concerning chronology, we acknowledge the great value of the large amount of fieldwork and excavations conducted in the region by Dr Cohen and his colleagues, including Dr Haiman. We thank the technical staff of the Centre for Isotope Research (University of Groningen) for performing the radiocarbon measurements.


Aharoni, Y. (1967) Forerunners of the Limes: Iron Age Fortresses in the Negev. IEJ 17: 1-17. —(1978) The Archaeology of Eretz Israel (Jerusalem: Shakmonah).

Barkay, G. (1992) The Iron Age In The Archaeology of Ancient Israel, edited by A. Ben-Tor (New Haven: Yale University Press): 302-73.

Bruins, H.J. (1986) Desert Environment and Agriculture in the Central Negev and Kadesh-Barnea during

Historical Times (PhD Dissertation, University of Wageningen; Nijkerk: Stichting Midhar Foundation). —(2003) Man and Landscape in the Negev Highlands: Runoff Systems. Horizons in Geography 57-58:

146-58 (Hebrew).


By Steve Rudd: Contact the author for comments, input or corrections.


Click to View