Fluctuating radiocarbon offsets observed in the southern Levant and implications for archaeological chronology debates

Significance

We observe a substantive and fluctuating offset in measured radiocarbon ages between plant material growing in the southern Levant versus the standard Northern Hemisphere radiocarbon calibration dataset derived from trees growing in central and northern Europe and North America. This likely relates to differences in growing seasons with a climate imprint. This finding is significant for, and affects, any radiocarbon application in the southern Levant region and especially for high-resolution archaeological dating—the focus of much recent work and scholarly debate, especially surrounding the timeframe of the earlier Iron Age (earlier Biblical period). Our findings change the basis of this debate; our data point to lower (more recent) ages by variously a few years to several decades.

Abstract

Considerable work has gone into developing high-precision radiocarbon (¹⁴C) chronologies for the southern Levant region during the Late Bronze to Iron Age/early Biblical periods (∼1200–600 BC), but there has been little consideration whether the current standard Northern Hemisphere ¹⁴C calibration curve (IntCal13) is appropriate for this region. We measured ¹⁴C ages of calendar-dated tree rings from AD 1610 to 1940 from southern Jordan to investigate contemporary ¹⁴C levels and to compare these with IntCal13. Our data reveal an average offset of ∼19 ¹⁴C years, but, more interestingly, this offset seems to vary in importance through time. While relatively small, such an offset has substantial relevance to high-resolution ¹⁴C chronologies for the southern Levant, both archaeological and paleoenvironmental. For example, reconsidering two published studies, we find differences, on average, of 60% between the 95.4% probability ranges determined from IntCal13 versus those approximately allowing for the observed offset pattern. Such differences affect, and even potentially undermine, several current archaeological and historical positions and controversies.

A long-standing assumption on theoretical and empirical grounds holds that, because of rapid mixing (on the order of less than a month), the premodern atmospheric radiocarbon (¹⁴C) levels for the midlatitudes are effectively uniform on an annual basis for each hemisphere, thus permitting use of standard northern and southern hemisphere ¹⁴C calibration curves for the Holocene (1–4). However, a number of investigations indicate possible spatial variations in contemporary ¹⁴C levels (5–9). There are indications that such regional ¹⁴C offsets, in some cases, show temporal variability associated with fluctuations in climate processes and changes in solar activity and ocean circulation (6, 9). Observed regional offsets are typically linked either with changing impacts of ¹⁴C reservoirs on an area over time, as in the case of East Asia (8, 9), or with differences in the timing of growing seasons for plants, and hence ¹⁴CO₂ uptake within a hemisphere, leading to the representation of differing parts of the intraannual ¹⁴C cycle, in particular when maximized under certain conditions (6, 7, 10–13).

Such offsets have paleoclimate relevance, but they are also of direct archaeological and historical importance. As ¹⁴C dating and derived timescales become more precise, potential regional offsets in contemporary ¹⁴C levels, and especially the issues of their scale and temporal stability, become relevant to research exploiting the limits of ¹⁴C dating resolution. The southern Levant has seen intense research efforts aimed at high-resolution ¹⁴C chronologies for the Late Bronze and Iron Ages (∼1200–600 BC), some trying to link with early history and Biblical chronology (14–21). Despite large datasets and sophisticated analytical programs, the issue of regional ¹⁴C offsets versus the default Northern Hemisphere (NH) standard record of IntCal13 (1) has not been seriously considered for the southern Levant.

Here, we investigate the question of the existence of possible offsets in atmospheric ¹⁴C values in the southern Levant over the period AD 1610–1912. We analyze ¹⁴C levels in known-age native Juniperus phoenicea tree rings across this period, employing a tree-ring chronology constructed from samples from historic structures at Taybet Zaman (TZM) in southern Jordan (∼30°15′17″N; 35°27′35″E) (Fig. 1A). This tree-ring chronology was cross-dated and securely placed in calendar time, employing standard dendrochronological methods (22, 23), against an existing J. phoenicea reference chronology from southern Jordan, dating AD 1469–1995 (24, 25) (Materials and Methods, Fig. 1B, and SI Appendix, section 1, Figs. S1 and S2, and Table S1).

Fig. 1.

(A) Map showing location of study area and sampled sites, TZM and BADG. (B) The time periods covered by the TZM J. phoenicea known-age tree-ring chronology and its cross-dating versus the existing chronology (24, 25).

Southern Jordan is part of the midlatitude NH. During the period when ¹⁴C released from atmospheric nuclear explosions (bomb ¹⁴C) exaggerated contemporary intrahemisphere ¹⁴C levels, this area falls in the middle of NH zone 2, well away from the dynamic interface regions around the Intertropical Convergence Zone (26). It would therefore be anticipated that the standard midlatitude NH ¹⁴C calibration dataset, IntCal13 (1), should be applicable for plant material from southern Jordan, as only during the peak of bomb ¹⁴C production, ∼AD 1955–1970, does the NH zone 2 region vary from NH zone 1, the source of most of the wood employed to build IntCal13 (central and northern Europe and North America) for the last few thousand years (1, 26). To investigate and test this assumption, we compare ¹⁴C ages obtained at the Arizona (AA) and Oxford (OxA) AMS ¹⁴C laboratories on known-age 5-y sections of tree rings dissected from the TZM timbers (SI Appendix, section 2 and Table S2) with the corresponding values of IntCal13 (1), and also with previous OxA ¹⁴C data on known-age plant material from 18th to 19th century AD Egypt which have been argued to demonstrate a 19 ± 5 ¹⁴C years offset for plants growing in Egypt in premodern times (7) (Fig. 2).

Fig. 2.

(A) The ¹⁴C ages from the known-age JJ samples, and previously published dates on known-age annual plant matter from Egypt (7), plotted against the NH radiocarbon calibration curve (IntCal13) (1) (1σ errors shown). (B) Differences between the JJ ¹⁴C ages and IntCal13 and comparison of these trends versus 5-y moving averages of reconstructed precipitation for southern Jordan (24) and regional temperature from Red Sea corals (28). (C) Overall offsets in ¹⁴C ages between the JJ samples and IntCal13 by laboratory, showing (light gray) the neutral prior (0 ± 20) versus the calculated posterior densities from each laboratory’s data versus IntCal13 (red and blue regions) (27).

Results

The Jordanian juniper (JJ) samples, on average, yield older ¹⁴C ages compared with the corresponding IntCal13 values (Fig. 2). The average offset for the collected JJ data, calculated as a Delta_R query in OxCal 4.3 (27) versus IntCal13 with curve resolution set at 1 y with a neutral prior of 0 ± 20, yields a posterior of 20.6 ± 4.5 ¹⁴C years for the AA data and 19.1 ± 2.8 ¹⁴C years for the OxA data (Fig. 2C), or 18.6 ± 2.5 ¹⁴C years if the OxA and AA datasets are combined (SI Appendix, Figs. S3–S5 and Tables S4 and S5). All these values are very comparable with the average 19 ± 5 ¹⁴C years offset determined previously for plants growing in Egypt (7, 12). We note that the offsets we observe are represented in data from more than one TZM tree, indicating a general pattern and ruling out any single sample/tree issue (SI Appendix, Fig. S6). In addition, by way of independent replication and confirmation, another ordered series of JJ samples from a different location (4 km NNW of Al-Bayda, site code BADG, ∼30°25′18″N, 35°26′58″E) (Fig. 1A) also exhibit similar and contemporary ¹⁴C offsets versus IntCal13 (SI Appendix, sections 1 and 2 and Figs. S7–S17).

However, importantly, it is also evident that this ¹⁴C offset appears to fluctuate over time (Fig. 2 A and B and SI Appendix, Figs. S3, S5, S6, and S8–S10). Thus, while real and relevant for ¹⁴C dating and analysis in the southern Levant, this situation also suggests that it is likely inappropriate to consider any average offset value, or potential correction, as generally relevant or applicable for any specific period without having data available for that specific time interval. While no correlation of the changing offset is evident with reconstructed precipitation for southern Jordan (24), there do appear associations between an increased offset and warmer temperatures as reconstructed from Red Sea corals in the period after ∼AD 1835 (28) (Fig. 2B), or generally for the extratropical NH (29), and with reversals in IntCal13 which correspond to increased solar irradiance (1, 3, 6, 29) (Fig. 2A and SI Appendix, Figs. S3, S6, and S8–S10). The offset period ∼1685–1762 (Fig. 2 and SI Appendix, Fig. S3), for example, starts around the change from the cool and (in the Mediterranean) dry conditions in the 17th century leading to the Maunder Minimum (peak 1645–1700), and corresponds especially to the warmer, post-Maunder Minimum conditions, particularly the long, stable, wetter period ∼1700–1750 noted in much of the Mediterranean (30). Despite expected variability in the ¹⁴C measurements of the AA versus OxA laboratories on the same material, the fluctuating offset is clear, independently, in the data from both laboratories. The boxes illustrated in Fig. 2B offer a subjective breakup of the data, indicating four possible time divisions, with two boxes offering little significant average offset and two boxes indicating a significant offset. With one partial exception, at AD 1855, the OxA ¹⁴C data on known-age annual plant material from Egypt produced similar age estimates (and, even for the 1855 exception, the JJ data from both AA and OxA show a consonant shift to more recent ¹⁴C ages centered at this year). The JJ ¹⁴C ages for some periods compare very well with values from the Southern Hemisphere (SH) ¹⁴C calibration dataset, SHCal13 (2) (Fig. 3A). On average, the JJ ¹⁴C data lie midway between the NH and SH ¹⁴C calibration curves, with offsets around half the average interhemispheric offset of 43 ± 23 ¹⁴C years (2) (Figs. 2 and 3A). Laboratory quality controls at both the AA and OxA laboratories (SI Appendix, section 2) and the dating of other known age samples indicate good agreement and only negligible differences for midlatitude NH samples with IntCal13 for both AA (31, 32) and OxA (Fig. 3 B and C and SI Appendix, Figs. S18 and S19) (33). Thus, the consistent and fluctuating ¹⁴C offsets observed for these JJ samples by both the AA and OxA laboratories versus IntCal13 appear real.

Fig. 3.

(A) Comparison of the JJ ¹⁴C data by laboratory versus both IntCal13 (1) and SHCal13 (2) (1σ errors shown). Known-age tree-ring ¹⁴C measurements from Bcharrê, (northern Lebanon), Stavros tis Psokas (western Cyprus), and Çatacık (western Turkey), are also shown (44). (B). OxA data on other known age NH tree rings (61, 62) versus IntCal13 (1). (C) Overall ¹⁴C years offset OxA NH data in B versus IntCal13; light gray region shows the neutral prior (0 ± 20) versus the calculated posterior region (blue) (27).

Discussion

The explanation proposed for the observed Egyptian offset (7, 12) was the different, almost opposite, growing season (winter to spring) for plants in Egypt in antiquity (before the Aswan Dam constructions in the 20th century AD) versus the spring and especially summer growing season for the central and northern European and northern North American trees comprising the Holocene IntCal13 dataset (1, 34). Since these near-opposite growing seasons correspond to periods of peak variability in natural (premodern) intraannual ¹⁴C level fluctuations, growing-season variability could readily account for the ∼2.5‰ ¹⁴C offset observed for Egypt (6, 7, 13, 34–38). Juniper trees in southern Jordan grow from autumn to early summer (24, 25), also largely in antiphase with trees in central and northern Europe and northern North America. In contrast, ¹⁴C ages on known-age wood from conifer trees growing at higher elevations in the northern Levant (Bcharrê, northern Lebanon), western Cyprus (Stavros tis Psokas), and western Turkey, which have growing seasons spanning, variously, from spring to summer (depending on temperature and moisture availability) (39–43), typically show no measureable offset (6, 44), even when the Jordanian trees are exhibiting larger offsets from IntCal13 (e.g., AD 1685–1760, 1835–1910) (Fig. 3A).

If we consider the plant taxa typically recovered from archaeological contexts in the southern Levant and subject to ¹⁴C dating, these have traditional growing seasons (subject to some intraregional geographic variations). Whereas a first group of crops comprising wheat, barley, oats, peas, lentils, and vetch grow winter to spring (with harvest April to May), a second group comprising chickpeas, sesame, flax, millet and some grapes, figs, and pomegranates grow later (harvest June to August), and a third group comprising other grapes, figs, pomegranates, and olives grow after that (harvest September through November; e.g., olives flower April to May, fruit grows in the summer, and harvest is around November) (45). Thus, a growing-season−related ¹⁴C offset versus central and northern Europe should apply to the first group, but not the second group, and then apply again, partially to more fully, to the third group. The growth periods for native tree species contributing charcoal at archaeological sites in the southern Levant likewise vary, but the typical pattern sees a period of dormancy over the hot, dry, summer months (42), so much of the growing season will be out of phase with central and northern European oak trees. If we examine the 121 ¹⁴C dates listed in one major study on Iron Age Israel as representative (17), 44% are on seeds/grains/semolina, 33% are on olive pits, 22% are on charcoal, and just 1% are on grapes. Thus, the majority of these samples likely fall outside the main spring to summer growing season represented by central and northern European oaks (46)–source of the earlier first millennium BC calibration data (1)–and so would be affected by a growing-season ¹⁴C offset.

However, a systematic growing-season−related ¹⁴C offset seems unlikely to be the sole explanation for the JJ ¹⁴C data, since the offset appears to vary over time from significant to insignificant values. The offset trend is most clear during regional warming periods, e.g., after AD 1835 (28) (Fig. 2B), or in the period following the Maunder Minimum (∼AD 1700–1750) (in Results). Such changes in climate likely modified the local growing season. For example, warmer conditions probably brought both the start and end of the growing season forward in the southern Levant, exaggerating the growing-season offset versus central and northern Europe. An increased scale of observed offset from the early 20th century might be associated with the increased Suess effect from fossil fuel use evident over Europe from about this time (47) (and thence transport to the East Mediterranean), especially since the Suess effect, which produces older observed ¹⁴C ages, peaks on an intraannual basis in the winter months for the NH (34, 35, 48). It would thus be reflected in the JJ wood growing across the winter months, and act to exaggerate differences in ¹⁴C values within the NH when compared with wood reflecting spring and especially summer ¹⁴C values, as IntCal13 comprises (1), when the Suess effect is at its annual minimum. However, the larger offsets observed in our data in the mid to later 19th century (Fig. 2B and SI Appendix, Fig. S3) occur before any plausible Suess effect and therefore likely reflect natural processes.

Our finding of a fluctuating ¹⁴C offset for the southern Levant versus IntCal13 (1) (and so against the IntCal13 ¹⁴C record from central and northern European and North American wood) potentially complicates previous studies where average offsets identified from particular periods or sets of samples were then considered as generally relevant through time, in particular in the case of Egypt (e.g., refs. 7 and 12). Instead, in cases like the southern Levant, where there appears to be a potential substantive growing-season (or other) difference which may provide a basis for intraannual offsets in ¹⁴C values as recorded in plant matter, our dataset indicates the need for a regional calibration time series if appropriate corrections are to be made for any particular time interval. Where such calibration time series are not yet available (namely, before AD 1610 for the southern Levant case at present), our dataset better indicates the circumstances under which a likely potential range of error may apply for earlier periods—assuming that similar conditions and process apply in earlier periods and accepting some possible variations—rather than offering any specific average correction factor. If we consider the combined OxA and AA dataset in Fig. 2A (as in SI Appendix, Fig. S3), then, overall, the offset is around 19 ± 3 ¹⁴C years (SI Appendix, Fig. S4A) applying an OxCal Delta_R calculation (27) with a neutral prior of 0 ± 20, or 16 ± 5 ¹⁴C years comparing observed values versus (linear interpolated) IntCal13 (1) values and errors (so around 2 to 2.5‰). These values and the ones for the separate OxA and AA datasets (Fig. 2C) are all strikingly similar to the 19 ± 5 ¹⁴C years offset observed previously from plant material from Egypt (7, 12). Where an offset applies, this suggests the approximate scale of a likely minimum southern Levant offset. However, the offset between the JJ and IntCal13 across two (subjectively selected) intervals with larger apparent offsets, between AD 1685–1762 and between AD 1818–1912, is a little larger at about 24 ± 5 ¹⁴C years (SI Appendix, Fig. S5), or around 3‰. Such possible offsets (or intraannual variations) of, overall, around 2 to 3‰ are plausible in terms of the known cycle and scale of natural premodern (preindustrial) intraannual ¹⁴C variations (7, 13, 34–38). Our findings highlight that it is important now to extend the time period of comparison if we are to determine whether such a scale of offset for the southern Levant occurs regularly at times when there are reversals and plateaus in the ¹⁴C calibration curve and/or regional or wider general warming episodes. On the basis of the currently available comparison, AD 1610–1910, we might anticipate the possibility of offsets relevant to ¹⁴C dating in the southern Levant ranging from about 19 ± 5 ¹⁴C years to 24 ± 5 ¹⁴C years at such times.

The findings reported here have immediate implications for high-resolution archaeological dating in the southern Levant. If the period AD 1610–1910 is representative in terms of a fluctuating offset versus the NH IntCal13 dataset, then, for substantial periods, and especially those where there are reverses or plateaus in the ¹⁴C calibration curve and/or a local or wider warming climate regime, there is likely a small, fluctuating, but substantive ¹⁴C offset in operation in the southern Levant which is of relevance to ¹⁴C dating. The impact on archaeological and other ¹⁴C dating will vary over time because this offset appears to fluctuate and because of the shape of the ¹⁴C calibration curve (1). To explore the potential scale of this issue, based on our JJ dataset (Fig. 2A and SI Appendix, Figs. S3–S5), we consider a possible modification of the IntCal13 ¹⁴C calibration curve (1) for the period ∼1200–700 BC, covering the debated Iron Age chronology period in the southern Levant. We apply the average 24 ± 5 ¹⁴C years adjustment observed across the periods exhibiting a substantive offset in our dataset (SI Appendix, Fig. S5) to those parts of the IntCal13 curve which exhibit reversals or plateaus in ¹⁴C values (Fig. 4A). Needless to say, this is an approximate and subjective adjustment; the exercise is aimed to be indicative and not robust.

Fig. 4.

(A) IntCal13 ¹⁴C calibration curve ∼1200–700 BC approximately adjusted by 24 ± 5 ¹⁴C years in the periods where curve taphonomy suggests a substantive JJ offset might apply based on Fig. 2 and SI Appendix, Fig. S3 (27). (B) (Bottom, vertical bars) Comparison of the calibrated calendar age ranges at 68.2% and 95.4% probability with, and without, both the average 19 ± 5 ¹⁴C years offset or the approximate JJ adjusted IntCal13 ¹⁴C calibration curve for the date sets from Tel Rehov (14) modeled as a sequence in OxCal (27), and the boundaries labeled as indicated from a rerun of the Bayesian chronological model from Khirbat en-Nahas (18). (Top, stars) Comparison of the noncommon (nonoverlapping), versus common (overlapping) ranges calculated when comparing the JJ adjusted ranges versus those from IntCal13 expressed as a percentage value: 0% (gray line) equates to exactly the same ranges; the larger percentage numbers indicate progressively less overlap. The average differences across the 17 comparisons are (i) for the 68.2% most likely ranges = 162% and (ii) for the 95.4% most likely ranges = 60%.

The potential impact of these offsets on real archaeological cases is then illustrated by looking at the changes in calendar date ranges achieved with, versus without, these offsets (Fig. 4A) in two high-profile examples: the initial dating of Tel Rehov in northern Israel, central to the early Iron Age and history debate in Israel (14), and the dating of Khirbat en-Nahas in southern Jordan, central to redating the rise of Iron Age Edom (18). The dates are on cereals, olive pits, seeds, and (in one case) charcoal for Tel Rehov, and charcoal and seeds (Phoenix dactylifera) for Khirbat en-Nahas, and all should likely reflect any contemporary southern Levant ¹⁴C offset. We model the published Tel Rehov ¹⁴C dates (minus the calendar date estimates) as a sequence in OxCal (27) with, and without, the above offsets, and, for Khirbat en-Nahas, we rerun the published Bayesian dating model with, and without, the above offsets (Fig. 4B and SI Appendix, section 2 and Tables S6 and S7). To compare the relevance of determining a specific record of such offsets through time, versus merely applying a general average (Delta_R) correction, we also consider the same data but applying a general 19 ± 5 ¹⁴C years Delta_R correction (as, e.g., refs. 7 and 12). The calibrated calendar age ranges for the elements of the Tel Rehov and Khirbat en-Nahas site sequences are shown from the nonmodified IntCal13 dataset, with the general 19 ± 5 ¹⁴C years correction, and with the specific contextualized approximate/estimated southern Levant modified calibration curve (Fig. 4A) in Fig. 4B.

It is notable that there is variation, and, in a number of cases, considerable variation, in the most likely 68.2% ranges, comparing the IntCal13 ranges with either the general 19 ± 5 ¹⁴C years adjustment or the data from the estimated JJ adjusted calibration curve. The 17 cases shown highlight that the application of a general 19 ± 5 ¹⁴C years offset tends to create larger differences in most cases (in 76% versus 24% of cases) versus the JJ adjusted curve, which tries better to model a plausible fluctuating scenario in keeping with observed data AD 1610–1910. This situation highlights the likely problems created if a simple “average” correction is applied to a geographic area when, in fact, the offset in question appears to fluctuate through time. Regardless, however, we may note that every shift is to “lower” or more recent calendar age ranges (whichever adjustment is considered), which is significant when considering recent debates over absolute dates for the Iron Age archaeological periods in the southern Levant. If we compare the 95.4% probability ranges, there are, in several cases, greater overlaps, but, even so, in a number of cases, there are substantial differences, and again the shifts are to lower or more recent calendar ages compared with the nonmodified IntCal13 ranges. If we consider the Fig. 4B comparisons between the IntCal13 ranges and the JJ adjusted ranges (comparing the calendar range of overlap versus the calendar years of nonoverlap between the IntCal13 and JJ adjusted curve ranges: Fig. 4B, Top, stars), then 14 of 17 (82%) of the most likely 68.2% probability ranges vary by ≥50%, and 12 of 17 (71%) of the most likely 95.4% ranges vary by ≥20%. While not always large, the scale of variations evident in many cases is sufficient to be substantive in considerations of Iron Age chronology, especially as current debates over Iron Age chronology in the southern Levant focus on intervals of only a few decades to ∼50 y to 100 y (13–20, 49, 50).

Available paleoclimate data for the southern Levant for the earlier Iron Age are inconclusive, but, after indications of cooler and arid conditions in the period around the close of the Late Bronze Age through initial Iron Age ∼3300–3000 BP (51–55), there are some (not always consistent) suggestions of wetter and/or warming conditions and increased solar irradiance ∼3000–2800 BP in the East Mediterranean region (refs. 52, 54, and 56–58; note that we adjust the ref. 57 timescale following their maximum age correction to match the age of the Santorini eruption as in ref. 53). This might suggest some exaggeration of regional growing-season ¹⁴C offsets in this period, especially around the plateau/reversal in the ¹⁴C record ∼2850–2800 BP/900–850 BC (1), and hence that a larger offset, at least comparable to those identified in the recent periods ∼AD 1685–1762 and AD 1818–1912, is relevant.

Although, overall, the ¹⁴C offset identified here produces what may seem to be relatively small dating changes, these are revealed to be of a scale that is important for high-resolution chronological work. They are especially important for the contested and detailed chronology debates in archaeological scholarship on the southern Levant region, particularly for those focused on differences of only a few decades to ∼50 y to 100 y in recent “high” (or conventional) versus “low” chronology debates (13–20, 49, 50). Thus, we recommend that users must proceed with caution when dating plant material from the southern Levant with a winter to spring growing season. It also seems likely that the offset we observe fluctuates, and thus is not best compensated for via a static, systematic, adjustment. This potentially complicates the previously proposed Egyptian offset (7, 12). The offset we observe is also relevant to other high-resolution work in the southern Levant based on detailed ¹⁴C chronology, such as paleoenvironmental investigations (52, 54), or for the correct association of radiocarbon-dated contexts and time series with geomagnetic intensity series which show important changes in the earlier Iron Age period in the southern Levant region (59, 60). The growing season (and climate) related ¹⁴C offset we identify changes and undermines the basis and assumptions in existing ¹⁴C work in the southern Levant, and especially in those periods where a larger offset likely applies. In these cases, the effect of the offset can be substantial, and of the scale of the existing range of scholarly debate. This ¹⁴C offset therefore requires attention, and, in particular, further work is necessary to better define its history since it appears to be time-varying (likely with climate associations as these affect growing seasons), especially when attempting to integrate ¹⁴C chronology closely with history. Ideally, a southern Levant radiocarbon calibration curve is required, or at least a longer comparison curve. The ¹⁴C offset observed in this study highlights a topic of general relevance to the radiocarbon field in cases where, within the same hemisphere, there are substantial differences in growing seasons (and hence conditions) for plants compared with the standard growing season represented by the midlatitude IntCal13 (1) ¹⁴C calibration dataset. In the present case, for example, the offset observed points toward more recent (lower) age ranges being more likely for some intervals in the earlier Iron Age in the Southern Levant, but for reasons not currently discussed in the high versus low scholarly debate.

Materials and Methods

We sampled native juniper (J. phoenicea) timbers in historical structures at TZM in southern Jordan (∼30°15′17″N; 35°27′35″E) (Fig. 1A and SI Appendix, sections 1 and 4). Employing standard dendrochronological methods (22, 23), the TZM historical timbers were cross-dated and placed in absolute calendar time, AD 1610–1940, against an existing J. phoenicea reference chronology from southern Jordan, dating AD 1469–1995 (24, 25) (Fig. 1B and SI Appendix, section 1, Figs. S1 and S2, and Table S1). Known-age 5-y sections of the TZM tree rings were dissected with a steel blade under a binocular microscope from the TZM timbers for ¹⁴C dating at the AA and OxA accelerator mass spectrometry (AMS) ¹⁴C laboratories (SI Appendix, section 2 and Table S2). A sequence of ordered, but not known-age, tree-ring samples from the BADG site (∼30°25′18″N, 35°26′58″E) (Fig. 1A) were also dated at the AA and OxA AMS ¹⁴C laboratories (SI Appendix, section 2 and Table S3). The resultant ¹⁴C ages were then compared against IntCal13 (1) (e.g., SI Appendix, Table S4), and also OxA data on known age plant material from 18th to 19th century AD Egypt which have been argued to demonstrate a 19 ± 5 ¹⁴C years offset for plants growing in Egypt in premodern times (7) (Fig. 2 and SI Appendix, Figs. S3, S5, S6, and S8–S13). Where stated (in the text, or where a figure plot indicates “r:1” at the top), the five-calendar-year resolution IntCal13 record was modeled to one-calendar year resolution by linear interpolation. Analysis of the ¹⁴C data employed the OxCal software (27) version 4.3. The OxCal runfiles with the data and coding employed for Fig. 4B are listed in SI Appendix, Tables S6 and S7, and the OxCal runfile for the analyses in SI Appendix, Figs. S8 and S11 are in SI Appendix, Table S5. As an example (employing the relevant data in SI Appendix, Table S2), the OxCal runfile for the analysis shown of the OxA data in Fig. 2C is listed at SI Appendix, Table S8.