Radiocarbon Dating the Iron Age in the Levant: A Bayesian Model for Six Ceramic Phases and Six Transitions
Finkelstein, Israel, Piasetzky, Eli, Antiquity
The authors of the present paper recently explored the possibility of constructing a framework for Iron Age chronology in the Levant by assigning radiocarbon dates to a set of destruction layers (Finkelstein & Piasetzky 2009). The data that have been collected in Israel (e.g. Carmi & Ussishkin 2004; Mazar et al. 2005; Boaretto 2006; Sharon et al. 2007a) combined with the detailed information on the stratigraphy and ceramic phases of the Iron Age (e.g. Mazar & Panitz-Cohen 2001: 273-6; Herzog & Singer-Avitz 2004, 2006; Finkelstein & Piasetzky 2006a), now encourage a bolder step: the dating of the entire Iron Age sequence of the Levant--each of the ceramic phases and the transitions between them. This venture also has rewards for the archaeology of neighbouring regions, such as the Aegean basin (e.g. Coldstream 2003; Coldstream & Mazar 2003).
In what follows, we use 142 samples (336 determinations) from 38 strata at 18 sites, measured by three methods in five laboratories, in order to establish a Bayesian model for the entire period between the collapse of the Egypto-Canaanite system in the second half of the twelfth century BC and the Assyrian conquests in the late eighth century BC (dates anchored in historical sources). No chronological anchors for this period exist in the archaeological record, e.g. well-dated Egyptian or Assyrian finds in primary stratigraphic contexts (Finkelstein 1996), hence ceramic phases have hitherto been dated in accordance with the interpretation of biblical texts. Yet these texts were only put into writing starting in the late seventh century BC and are designed to convey the ideology and theology of the authors; their reliability for reconstructing events that ostensibly occurred centuries before their compilation, therefore, is questionable (e.g. Na'aman 2002; Romer 2007).
Following the stratigraphy of many key Iron Age sites in the Levant we list the well-defined ceramic phases as sequential horizons. Though a given pottery form could have been used in more than one phase, the ceramic assemblages did not overlap. The exact nature and pace of the transition between the phases is not always clear. In some cases a ceramic tradition could have ended as a result of a military upheaval, but there were also transitions that took many years. Below we assume the simplest possible model of abutting the sequential phases with no overlaps and no gaps. Figure 1 shows four ceramic assemblages from Megiddo. The locations of the sites sourced are shown in Figure 2, and the particular strata used to provide samples at each site are shown in sequence in Figure 3. The character and assemblage of these strata are summarised in Supplement 4 (online).
Selection of data for the model
We worked with a clear set of criteria for accepting or rejecting measurements and with a simple and consistent procedure for interpreting the data. All available determinations from loci which were both safely assigned stratigraphically and well classified from the point of view of ceramic typology were incorporated in the model. Some of the strata that provided samples for radiocarbon dating had come to an end in heavy conflagrations that resulted in a thick collapse that had buried pottery vessels and other finds as well as clusters of charred grain seeds and olive pits. These cases provide the most secure provenance for radiocarbon dating.
Due to the risk of 'old wood effect', only short-lived samples were included (for bias created by 'old wood effect' see Schiffer 1986; Sharon et al. 2007a: 5-6; Finkelstein & Piasetzky in press). Results from all laboratories were included (contra Mazar et al. 2005, who doubted the accuracy of one laboratory without providing a convincing reason--Finkelstein & Piasetzky 2006b). The well-defined ceramic assemblages, mainly from destruction layers, and the fact that we are dealing with a small territory, assure safe spatio-temporal modelling. …