How old is the Kingdom of Edom?

When this paper appeared in the British journal Antiquity at the end of 2004, it immediately attracted a lot of attention, particularly outside the professional archaeological world, with a number of newspaper articles prompted by a press release from the principal authors (see, for example the report in McMaster Daily News, dec. 20, 2004) . The reason for the media attention is the claim it seems to be making for the historicity of ‘Biblical Edom’. The never-ending discussion about whether archaeology can prove (or disprove) the historical truth of the Bible has been given a new and powerful push with this paper. But what is all the fuss about?

Khirbet en-Nahas is one of the most important sites for the archaeology of southern Transjordan, the region identified with ancient Edom. It consists of a copper-working site, followed by a possible fortress, in the Wadi Faynan, on the east side of the Wadi Arabah. The site has been re-excavated by the authors of the article, who have collected a number of radiocarbon samples from several layers, and they present the results in this paper, embedded in the chronology that the 14C dates provide. The results, they claim, are ‘spectacular’. The dates range from the 12th to the 9th century BC and prove beyond doubt that roots of the Edomite kingdom lie in the 12th or 11th century.

However, these dates are exactly the problem. Two sets of data are being used. The calibrated radiocarbon dates are presented in a table, but the dates used in the text are BCal dates, which differ from the tabulated calibrated radiocarbon dates (see table below). BCal (Bayesian radiocarbon Calibration tool) dates are calculated by incorporating the calibrated radiocarbon dates into other chronological information about the site, such as stratigraphic sequences of layers, or absolute dating information from other sources (such as texts). Using this information the BCal program can narrow down or modify the radiocarbon ranges. However, the authors do not specify any of the additional sources they have used to reach the results of the BCal dates, some of which (especially the earliest ones, see table) are considerably older than the radiocarbon dates. Since these early dates in particular are vital to the argument of the authors, they must be transparent and publish the other chronological data they have used.

Table comparing the radiocarbon dates of the samples with the BCal dates mentioned in the text

If we look, for example, at the ‘fortress’ in area A: The calibrated 14C range of stratum A4a, which predates the construction of the ‘fortress’, is 1010-920 BC. The BCal date range for Stratum A4a according to the text is 1130-970 BC, with a modal (highest probability) value of 1120 BC. There is thus a discrepancy of at least 110 years between the BCal dates and the calibrated radiocarbon dates, with no explanation provided.

The earliest occupation phase in Area S, a metal working area, has a calibrated 14C range of 1130-1015 BC. According to the text the BCal results date this stratum to 1260-1240 BC, and 1215-1012 BC, with a modal value dated to pre -1190, over half a century earlier than the earliest limit of the radiocarbon range. This is a possible cooking installation and it is followed by a heavy layer of metal waste, dated to 1005-965 (14C) or 1055-915 (BCal). On top of this was a four-room building, dated to 905-830 (14C) or 970-830 (BCal). Finally, two dates from a slag heap in area M have not been submitted to BCal analysis.

The authors need to be more specific about the archaeological or other data they have used to reach their extremely early BCal dates, before they can make any claims based on these dates.

The ‘datable artefacts’, according to the authors, corroborate the Early Iron Age (c. 1200-1000 BC) date of the first occupation. However, a closer look at these artefacts, notably the pottery, shows that all these ‘early dates’ are problematic. The dating of both Negbite pottery and Midianite pottery is problematic, as the authors readily admit: both types of pottery have been found in contexts dating from the very end of the Late Bronze Age, c. 1200 BC, to Iron Age II, dated between the 9th and 7th centuries BC. Nevertheless the authors use the presence of these types of pottery as an argument for a date in the 12th century.

At each step, it seems, the authors are attempting to push the dates as early as possible, on average about a hundred years or so earlier than the calibrated radiocarbon evidence allows for. The irony is that the authors claim that their ‘high-precision radiocarbon dating is liberating us from chronological assumptions based on Biblical research’, but their paper clearly manipulates the dates according to chronological assumptions that are not articulated. This lack of transparency is unacceptable.

The second claim that the authors make is that of the rise of secondary state formation (implicitly: a kingdom) in Edom in the 10th century, on the basis of the presence of the copper industry and the ‘fortress’. Unfortunately, one ‘fortress’ does not make a kingdom. Or, to put it differently, most kingdoms may have fortresses, but not every fortress belongs to a kingdom (and, of course, the interpretation of the structure as a fortress is no more than a hypothesis). Neither does industrial production require a state structure. Recent research suggests that local corporate groups are very capable of conducting and maintaining large-scale industrial activities, and building up the infrastructure, such as fortified buildings and housing, that comes with it. So far nothing else has been found in southern Transjordan to justify the incorporation of the Khirbet en-Nahas ‘fortress’ in a larger polity. The presence of a 10th-century ‘fortress’ at Khirbet en-Nahas is no indication, let alone proof, of the early rise of the Edomite kingdom. In fact, if, as the authors claim, the copper industry-cum-fortress of Khirbet en-Nahas would be evidence of an Edomite kingdom, we may wonder why it ceased to exist exactly at the time when the other features of that kingdom make their appearance, the 8th and 7th centuries.

We do not underestimate the importance of the excavations at Khirbet en-Nahas. It is one of the most important sites in the region, and can give us much information about the technology and economy of copper production, and the social organization of the region in a period of which little is known. However, the evidence published in the Antiquity paper is confused and misleading, leading to claims that cannot at present be substantiated.

Three authors of the original paper in Antiquity have responded to this review.

Our response to: How many fortresses do you need to write a preliminary report?

The March 2006 issue of Antiquity (Project Gallery) will publish our formal response to the original paper by Levy et al. in Antiquity 402. However, several points raised in the present response by Levy et al. to our critique above merit an immediate reaction, since they may not be addressed fully in our Antiquity piece:

· In their response, Levy et al state that “it would be very unprofessional from our side to discard any of the available non-archaeological sources such as the Bible” (p.3). However, the Hebrew Bible as a source is simply too late (eighth century BC at the earliest, and probably later, according to common scholarly opinion) to provide reliable information on developments in the tenth century. Therefore, it must be excluded as a primary source.

· Further, Levy et al. state that, so far, the dating of Edomite pottery and consequently the main highland sites is based solely on the find of the Qos Gabr seal. The possibility of earlier dates for this pottery must therefore not be excluded (Levy et al. p. 3). However, this is being highly economical with the truth. In assessing the chronology of ‘Edomite’ pottery, the final publication of the excavations at Busayra by Piotr Bienkowski (2002) took into account:
o C14 dates from the Faynan area
o ceramic parallels from Transjordan and Palestine
o well dated imported Attic pottery
o well dated inscriptional material (NOT just the seal impression of Qos Gabr).

This analysis indicated that pottery assemblages from the Faynan area C14-dated to the ninth century BC were quite different from Busayra and other ‘Edomite’ sites, indicating a date later than the ninth century BC for the latter material. ALL the other evidence pointed to a date no earlier than the late eighth century BC, with this pottery tradition continuing to the end of the Persian period at the earliest, and possibly into the Early Hellenistic period.

· Levy et al. claim that there is a constellation of fortresses that can be dated to the tenth century BC: Nahas, Hazeva and Tell Kheleifeh (p. 7). However, a tenth-century date for Kheleifeh is built solely on Nelson Glueck’s interpretation of the material, which was considerably earlier than is accepted now. There is therefore an element of circular reasoning here: an early date for Kheleifeh implying an early date for Edomite pottery in general, an issue we have dealt with above.
The fortress at Hazeva, which is C14-dated to the tenth-ninth centuries, was part of a chain of settlements that flourished in the Beersheba Valley, and which may have been part of a trade route, perhaps even connected to the Arabian trade. It seems more likely, therefore, that Khirbet an-Nahas flourished as a result of its connection with this trade route, which would also explain some of the finds at Khirbat el-Msas. Thus, there is no need to create unsubstantiated connections with sites in the Edomite Highlands.

· Levy et al. accuse us of failing to see the significance of the Nahas results as “the first attempt to apply radiocarbon dating to stratified archaeological deposits from Iron Age architectural constructions in Edom”. We disagree. We are fully aware of the importance of the C14 dates at Nahas, and we accept them as reliable dating information about the site.
The C14 dates from Khirbet an-Nahas are significant enough in themselves. We fail to see, therefore, the need for a tool that, however useful it may be in other contexts, in this context only confuses the issue. The results from the Bayesian Calibration presented by the authors have not only pushed the C14 dates back, but also considerably widened their range, and made the results less reliable.
We have attempted to replicate the results of Levy et al's Bayesian calibration, using their published C14 dates and stratigraphic information. We used the BCal programme that is provided by Sheffield University (http://bcal.sheffield.ac.uk/). Our results differ significantly from those presented by the authors of the paper. The outcome of our tests puts especially the earliest dates 40-50 years later than the Antiquity paper does, and the highest probability values generally fall within the original, calibrated C14 ranges.

We fully support a programme of C14 dates on Iron Age sites in southern Jordan and the Negev, but with the proviso that we establish first a framework of calibrated C14 dates not influenced by the subjective biases inherent in using the Bayesian tool.

In their response to our methodological critique on the Khirbet en-Nahas dating results, Levy et al. advise us to ‘read the book’ (their publication of the Khirbet en-Nahas C14 dates and BCal modelling in Levy and Higham 2005), and their extensive response to our critique on this website (link). This, they say, would have solved many of the misunderstandings. None of these sources were available when we wrote our critique, which addressed simply the published paper in Antiquity 2004.
We have now read the book, and while we agree that it solves some of the problems, much of the confusion remains.

Our main critique, that the BCal modelling of the calibrated C14 dates pushed these dates back significantly, is addressed in Higham et al.’s paper. In short: the BCal modelling does NOT push the dates back. If we look at Area A: “The Oxford results show that the modelling has had a limited influence. The posterior distributions show little difference when compared with the original radiocarbon likelihoods themselves”. And for area S: “The Bayesian analysis yielded little additional chronometric data compared with that derived from the radiocarbon likelihoods.” (Higham et al. 2005,167).
Comparison of the tables in this paper suggests that in order to reach the spectacular early dates published in Antiquity 2004, the authors have used the maximum (95.4%) probability range of the BCal results. However, how they reached the modal values (the “value with the highest probability”, according to the Antiquity paper) remains enigmatic.

In the new publications, the data published in Antiquity 2004 have been combined with the results of new radiocarbon samples, but the confusion continues. For example, Levy et al. 2005 provide a sequence for area A based on these new results, stating that stratum A4a is dated by two samples (GrA 25318 (calBC 1210-1045) and GrA 25354 (calBC 1185-1180, 1125-945) (Levy et al. 2005,138). These samples were taken from stratum A3, but ascribed to A4a, seemingly because they were too early. The paper by Higham et al. discards these two results, ironically on the basis of the Bayesian modelling, as unreliable (Higham et al. 2005,170).
The four-chamber gate of A3 is dated by Levy et al. to the early 10th century on the basis of the original Oxford lab sample (OxA 12366 [calBC 1000-985]). Two other, new samples, GrA 25321 and GrA 25322, which would date the building to the 9th century (Levy et al. 2005,138; Higham et al. 2005,170) are pushed into the next stratum. Higham et al. date the transition (boundary probability) between A4a and A3 to 900 BC, and consequently, the building of the gate to after 900 BC. So, again, there is a discrepancy between the results of Higham et al.’s analysis of the radiocarbon data, and Levy et al.’s interpretation of them, of about 100 years.

We appreciate the fact that the authors of the Antiquity 2004 paper had little control over the media-hype that followed the publication of that paper (a quick websearch reveals that practically all sites that refer to this paper feature one specific co-author, who has since withdrawn from the discussion).
However, Levy et al.’s conclusions in their 2005 paper still push the radiocarbon results back in order to fit the question whether David or Solomon built the Nahas four-chamber gate, and are liberally peppered with quotes from the bible. This hardly supports their claim that the new results finally release us from bible-related interpretations of archaeology in the region.

Higham, Th., J. van der Plicht, Ch. Bronk Ramsey, H.J. Bruins, M. Robinson and Th.E. Levy. 2005. Radiocarbon dating of the Khirbet en-Nahas site (Jordan) and Bayesian Modeling of the results. Pp. 164-178 in Levy Th. E. and Th. Higham, eds. 2005. The bible and radiocarbon dating. Archaeology, text and science. London: Equinox.

Levy Th. E. and Th. Higham, eds. 2005. The bible and radiocarbon dating. Archaeology, text and science. London: Equinox.

Levy, Th. E., M. Najjar, J. van der Plicht, N. Smith, H.J. Bruins, and Th. Higham. 2005. Lowland Edom and the High and Low Chronologies. Pp. 129-163 in Levy Th. E. and Th. Higham, eds. 2005. The bible and radiocarbon dating. Archaeology, text and science. London: Equinox.

See also:
I. Finkelstein: Khirbet en-Nahas, Edom and Biblical History. Tel Aviv 32 (2005):119-125

Larry G. Herr
(Department of Religious Studies, Canadian University College
Lacombe, Canada)

There are a few other observations why the view that Edom began in the Araba during the Iron I period and was a “state” during the 10th century is premature. These address archaeological and “ethnic” concerns.

1. Just because the site is located within the territory that will become Edom at the end of Iron II doesn't make it Edom in Iron 1. That seems to be Levy et al's assumption and it is a problematic jump in my view. Where is the archaeological continuity between the two periods of settlement?

2. We cannot include Khelayfa (Kheleifeh) as a 10th century site. Nor can it be used for early Iron II. Please see the final publication of Glueck’s excavations prepared by Gary D. Pratico as a reappraisal (Nelson Glueck’s 1938-1940 Excavations at Tell el-Kheleifeh: A Reappraisal. Atlanta: American Schools of Oriental Research, 1993). This was a late Iron II site. It is so with regard to both pottery typologies and the paleography of the inscriptions.

3. The local pottery must be published in order to check the typological dating and typological connections with other regions. We have almost nothing from the Iron I period or Iron IIA from southern Jordan. We must use the pottery from elsewhere to make judgments. Further, are the ceramic connections to the east or west? How about NAA and/or petrographic studies to find out where the local pottery was manufactured? To state that this area, rather than the highlands, is the core and original territory of Edom then mandates a continuum of uninterrupted settlements, and associated local pottery and other material culture, throughout the Iron Age. We have no means to check the claim without this evidence.

4. What's to hinder, for instance, the copper miners coming from the west? Could they be Egyptians who were also mining copper in Sinai and probably Timna? Or were they locals (nomads? or itinerants?) employed by westerners, such as Egyptians? The picture would be much like the Abujaber family (and others), who employed the local bedouin to farm their large agricultural estate around Yaduda south of Amman in the 19th c (see Raouf Abujaber, Pioneers over Jordan. London: I. B. Tauris, 1989). Could the site be a fortified industrial site of Iron I-II in the Araba like Yaduda was a fortified agricultural site? It was the copper that brought miners and support personnel to the location. There are many parallels to similar "boom" settlements, such as, for instance, Dawson City in the Yukon--for one year, and one year only, the largest city west of Chicago and north of San Francisco in 1898, simply because of gold. The ethnic makeup of Dawson City was varied. Formally it was part of Canada, and Canadians controlled the border and policed the city, but most of the miners were Americans. Granted, this is only a crude analogy, but until you eliminate such possibilities, other suggestions are nothing more than speculation.

5. Or could the inhabitants have come from the south? Why not use the Midianite pottery and suggest they are Midianites mining the copper for their caravans between the Levant and Egypt? Just another possibility.

6. We also need to establish why mining copper at a fortified site implies a "state" in the region, especially when there is not yet any clear evidence for the settlement of other parts of the region. Was 19th c Yaduda a state or part of a state located nearby? If we say it was part of the Ottoman system and imply a similar situation at this site, then we are suggesting the state could have been centered very far away--perhaps in Egypt or Gaza. In that case, it wouldn’t be an “Edomite” state, but it would be a site at the end of a long tentacle originating in another “state.” Perhaps it was like a colony.

Locus	Stratum	Context	Cal BC date	Bayesian data
356	S4	Cooking installation, basal layer	1130-1015 BC	1260-1240 BC 1215-1020 BC (highest probability value pre-1190 BC)
341	S3	Earliest industrial slag layer; under building foundations	1005-965 BC	1055-915 BC
336	S2b	Main occupation phase of building	905-830 BC	970-830 BC (modal value 895 BC)
331	S2a	Re-use of Room 2	895-875 BC	900-760 BC (modal value 815 BC)
95	A4a	Ashy layer below surface over bedrock	1010-920 BC	1130-970 BC (modal value 1120 BC)
94	A3	Surface connected to original gate structure	1000-985 BC	1005-870 BC
92	A2b	Massive smelting inside gate chamber	900-875 BC	920-815 BC (modal value 885 BC)
61	A2a	Installation with human remains, outside gate	900-805 BC	990-790 BC (modal value 835 BC)
539		Tap slag layer	910-886 BC
511		Tap slag layer	829-801 BC