Radiocarbon chronology of Iron Age Jerusalem reveals calibration offsets and architectural developments

Significance

Establishing a detailed absolute chronology in an actively inhabited urban environment is challenging. The key to the solution is to apply stringent field methodologies using microarchaeological methods, leading to dense, radiocarbon-dated stratigraphic sequences. In Iron Age Jerusalem, 103 ¹⁴C measurements on samples from a range of contexts were used to reconstruct Jerusalem’s urban history. By wiggle matching against the calibration curve, a decadal resolution, not usually possible during the problematic 300-y-long Hallstatt plateau, was achieved. Results also revealed excursions in ¹⁴C concentration that were outside the ranges of the calibration curve, verified by a set of 100 calendar-dated tree rings. This field and lab approach could well be applicable to dating other urban contexts.

Abstract

Reconstructing the absolute chronology of Jerusalem during the time it served as the Judahite Kingdom’s capital is challenging due to its dense, still inhabited urban nature and the plateau shape of the radiocarbon calibration curve during part of this period. We present 103 radiocarbon dates from reliable archaeological contexts in five excavation areas of Iron Age Jerusalem, which tie between archaeology and biblical history. We exploit Jerusalem’s rich past, including textual evidence and vast archaeological remains, to overcome difficult problems in radiocarbon dating, including establishing a detailed chronology within the long-calibrated ranges of the Hallstatt Plateau and recognizing short-lived regional offsets in atmospheric ¹⁴C concentrations. The key to resolving these problems is to apply stringent field methodologies using microarchaeological methods, leading to densely radiocarbon-dated stratigraphic sequences. Using these sequences, we identify regional offsets in atmospheric ¹⁴C concentrations c. 720 BC, and in the historically secure stratigraphic horizon of the Babylonian destruction in 586 BC. The latter is verified by 100 single-ring measurements between 624 to 572 BC. This application of intense ¹⁴C dating sheds light on the reconstruction of Jerusalem in the Iron Age. It provides evidence for settlement in the 12th to 10th centuries BC and that westward expansion had already begun by the 9th century BC, with extensive architectural projects undertaken throughout the city in this period. This was followed by significant damage and rejuvenation of the city subsequent to the mid-eight century BC earthquake, after which the city was heavily fortified and continued to flourish until the Babylonian destruction.

Absolute high-precision radiocarbon dating is pivotal in efforts to resolve complex historical and archaeological sequences. Jerusalem in the Iron Age (1200 to 586 BC) is a key site in its widespread archaeological importance but has not previously been a target for such an effort. The city has a history of 150 y of intensive archaeological excavations, stemming from the interest in ancient texts, particularly the biblical writings, which hold a thick and detailed historical record of king lists and significant political events. Surprisingly, though, radiocarbon dating was rarely applied for this period in Jerusalem (1–3), and the chronological frameworks were based solely on pottery typology, stratigraphy, and integration of textual sources. The complexity of the site’s formation processes, especially at the Southeastern Ridge (also known as the “City of David”), due to its sloping topography and the frequent rebuilding of the site from the Early Bronze Age (4) up till the present, presents significant challenges to identifying in situ, well-characterized archaeological contexts for radiocarbon dating.

Besides the archaeological difficulties, large wiggles and plateaus in the ¹⁴C calibration curve (e.g., the “Hallstatt Plateau” during 770~420 BC) mean that radiocarbon samples from the Iron Age period produce multiple dating solutions and very wide calibrated ranges (5).

The calibration curve shape is largely dictated by fluctuations in solar activity, which influence the radiocarbon production rate (6–8). The curve section representing the last ca. 13000 y is based on measurements of radiocarbon locked away in tree rings, which can be independently dated by “dendrochronology”. The majority of the measurements in the calibration curve are based on subsections of 5 to 20 rings lumped together (9), but with the introduction of accelerator mass spectrometers (AMS), which require smaller sample sizes, the dating of single tree rings became both practical and possible (9–12). Single-year data from such studies, including part of the Iron Age period (856 to 626 BC), are incorporated in the most recently published radiocarbon calibration curve, IntCal20 (9, 10, 13). Even further dating precision improvement can be achieved by the process known as “wiggle matching” (14), in which fluctuations in the structure of the curve can be used to fit similar structures formed when radiocarbon measurements are produced from tightly secured, sequential archaeological contexts with unambiguous stratigraphic provenance (15, 16).

The high-resolution data also provide insights into the origin and scale of regional and interlaboratory offsets from the global radiocarbon value (17). While certain regions appear to have clear offsets connected with growth environments or growth seasons (18), other argued regional offsets relevant to the Levant (19) have now been shown to be the result of interlaboratory differences incorporated within the structure of the calibration curve. Nevertheless, a case continues to be made for a region-specific effect in the southern Levant (20). When recognized and precisely measured, a better understanding of identified offsets, whatever the cause, can be used to improve regional chronologies.

On the archaeological side, recent advances in sampling methods and context characterization using microarchaeological tools (21) allow for more confident identification of in situ microstratigraphy and minimize chronological misfits (outliers) between ¹⁴C dates and archaeological contexts (15, 16, 22, 23). Using a combination of these improved approaches in terms of both archaeological sampling and radiocarbon calibration, we overcame the archaeological complexity of Jerusalem, refined the radiocarbon Hallstatt Plateau, and identified two regional radiocarbon offsets during the Iron Age period.

Building the Chronological Ladder

Excavations since the mid-19th century AD in different parts of the Southeastern Ridge (24, 25)—south of the present-day Old City of Jerusalem—have yielded varying stratigraphic schemes (26), best seen as employed by Y. Shiloh (27, 28), who recognized six Iron Age strata (Strata 10 to 15, SI Appendix, Table S1). Recent excavations have further refined this sequence, with intensive building activities spanning the end of the 9th through the early 6th centuries BC, making this period in Jerusalem a key candidate for high-resolution absolute chronology study. In recent years, numerous publications of radiocarbon chronologies for various periods and areas within Jerusalem have been published (4, 22, 23, 29), but no extensive ¹⁴C study of the Iron Age was made.

Our approach to understanding the city’s Iron Age history combines 1) over 100 ¹⁴C dates obtained from recent excavations in different parts of the Southeastern Ridge; 2) Extending a series of annually resolved tree-ring ¹⁴C measurements between 624 to 572 BC from European and American trees (Irish oak and Bristlecone pine) for high-precision comparison with the decadal calibration curve data currently available for this period and our archaeological ¹⁴C determinations; 3) development of decadal chronological resolution for ceramic seriation of pottery assemblages which were retrieved from the tightly radiocarbon dated contexts analyzed in this study; 4) detailed, microarchaeology-aided evaluation of archaeological context integrity.

This approach, alongside the attribution of historically documented events, enabled us to use two chronological pegs to resolve some ambiguity in the position of the ¹⁴C dates along the calibration curve within the Hallstatt Plateau period. The first peg is the series of Babylonian war campaigns at the onset of the 6th century BC. The campaigns, culminating in the city’s destruction in 586 BC, are a secure chronological anchor supported by the strong correlation between the Babylonian chronicles and the biblical description in Jeremiah and 2nd Kings that note that the temple, palace, and houses of Jerusalem were set on fire on the 19th y of Nebuchadnezzar king of Babylon (Jeremiah 52:12-13, 2nd Kings 25:8-9), whose reign began in 605 BC based on Babylonian chronology (30, 31), leading to its 19th year in 586 BC. This, alongside the consistent archaeological evidence from Judah of destruction layers dating to this period (32, 33), coupled with additional information, such as the appearance of Judahite exiled communities in Babylon and the recount of the Babylonian destruction in the works of later historians, such as Flavius Josephus (Antiquity of the Jews–Against Apion), all indicate that the Babylonian destruction of Jerusalem in 586 BC, which determined the end of the Judahite Monarchy and the Iron Age in the southern Levant, is a historically sound event that can be relied upon. A clear destruction layer containing the typical pottery assemblages of this temporal phase has been comprehensively characterized in different parts of the city (28, 34–36).

The second crucial chronological peg involves a mid-8th century BC earthquake, occurring more than 150 y prior to the Babylonian destruction of Jerusalem (corresponding to Shiloh’s Stratum 12b, SI Appendix, Table S1). A clear-cut destruction layer attributed to the earthquake was uncovered on the eastern slopes of the Southeastern Ridge (37), while further reinforcing evidence derived from paleoseismic data (38–40) and the mentioning of this momentous event in the biblical text (Amos 1:1). Widespread evidence of the earthquake has also been uncovered in other excavations within the Southeastern Ridge (28, 41), as well as in other Iron Age sites, allowing archaeologists to draw stratigraphical and chronological correlations across a wide geographic expanse (42–46).

Excavation Areas and Contexts

We retrieved samples for radiocarbon dating from three excavation areas located in different parts of the Southeastern Ridge (Fig. 1):

• 1.Areas C and U: These areas are located upslope from the Gihon Spring—the city’s most important water source in the Iron Age (29, 47). Sampling for this study focused on the Spring Tower, the Fortified Passage (48), the nearby Mid-Slope fortification and surrounding structures (49), particularly several buildings dating to Iron Age II (50). See SI Appendix, S3.1 (area U) and SI Appendix, S3.2 (Area C) for detailed contexts.
• 2.Area E: This area, excavated initially by Y. Shiloh (27, 28) and located on the eastern slopes of the Southeastern Ridge, c. 100 m south of the Gihon Spring, contains a domestic quarter. His extensive excavations in the 1970s established the benchmark stratigraphy and pottery typology for Iron Age Jerusalem (SI Appendix, Tables S1 and S3.4 for detailed contexts).
• 3.Areas 10 and 70-South: This area, also known as Giv‘ati Parking Lot (GPL), is located to the west of the Southeastern Ridge summit, along a slope leading down toward the deep “Central Valley”, separating the “Southeastern Ridge” from the Western Hill, known today as Mt. Zion (51). The excavations exposed a monumental building dating to Iron Age IIB–C (Building 100), comprising a row of three rooms and an open area to the north. The building was destroyed by the city-wide conflagration in 586 BC (SI Appendix, S3.3 for detailed contexts).

Fig. 1.

Location of the excavation areas in the Southeastern Ridge, south of the Old City of Jerusalem. See also SI Appendix, S1.

Results and Discussion

A total of 103 archaeological samples were dated, consisting of single charred items, mostly seeds (2/3 of the samples). No mortar, plaster, or organics in sediments were dated but were assigned dates through organic remains preserved within the construction materials, such as fresh or charred straw. We also dated an additional 100 single tree rings from Irish oak and Bristlecone pine, improving the ¹⁴C calibration-curve resolution in the segment of the Hallstatt Plateau critical to this part of Jerusalem’s chronology. In reporting the radiocarbon dating results throughout this manuscript, we use the 68.3% probability. The extended SI Appendix contains detailed descriptions of the dated samples, contexts, and methods used.

High-Precision Dating in the Hallstatt Plateau (Area U).

Room 17063 in area U (Fig. 1) provided an opportunity to build a chronological skeleton on which the radiocarbon levels in Jerusalem could be assessed. Dating here focused on a vertical series (1.2 m depth) of eleven superimposed floor surfaces and the accumulations overlying them within one room in Area U (Floors 11–1, Room 17063; Figs. 1 and 2; for detailed description and microarchaeological context characterization, see SI Appendix, S3.1; the modeled results are also presented in SI Appendix, Fig. S12). This effort yielded ¹⁴C concentrations closely tracking the fluctuations (wiggles) over much of the calibration curve within the Hallstatt Plateau. This outcome demonstrates the potential to obtain a high-precision radiocarbon chronology within a relatively flat section of the calibration curve.

Fig. 2.

Area U, Room 17063: Most probable age distributions based on the modeled stratigraphic sequence (SI Appendix, S3.1), shown on the calibration curve. The different colors mark different floor numbers. The measurements along the sequence follow the shape of the calibration curve, enabling high resolution within the plateau region. The narrow timespans calculated for the floors within the Hallstatt plateau are shown in the bar at the Bottom of the figure. Sample locations are in the Top-Right picture. Striped circles mark samples from the horizontal excavation and not directly from the shown section. The earthquake event was sampled in the adjacent room (Room 17130). Its calibrated radiocarbon age, combined of six separate measurements, is shown in black.

The deepest in situ surface in Room 17063 (Floor 11), a firepit on the orange clay used to level the underlying bedrock—was independently dated with four separate samples to reach high precision (providing a combined date of 2545 ± 11 ¹⁴C BP. R_Combine details are in SI Appendix, Table S5) (52). The calibrated date corresponds to c. 760 BC, a date slightly earlier than the beginning of the Hallstatt Plateau, constraining the beginning of the overlying archaeological sequence before the plateau timespan. Dates obtained from an accumulation of collapsed building materials belonging to Floor 11 cover several decades, beginning from the 9th century BC till around 725 BC (Fig. 2, blue samples). Floors 10–6 (c. 30 cm thick; Fig. 2, purple and red samples) represent the next phase of occupation in Room 17063, with dates of 730 to 705 BC. These samples are relatively enriched in ¹⁴C compared to the expected values of the calibration curve, forming an offset from the curve (see below). The three subsequent occupation phases (Floors 5–3, Fig. 2, orange and yellow samples), with decreasing radiocarbon amounts, form a sequence within the years 710 to 670 BC. Underneath the topmost floor, made of white earth (Floor 1), was a small ash-pit (assigned as Floor 2), dating to 690 to 670 BC, while the topmost accumulation—a broken installation containing many grape pips—yielded a date corresponding to a peak in a wiggle in the calibration curve, at either c. 670, 630 or 600 BC (SI Appendix, Fig. S7). Another date from the same floor (Floor 1) falls at the bottom of the same wiggle, around 660 BC, anchoring the time when the floor was constructed to around the mid-7th century BC.

Attempts were made in the past to distinguish the early and late 7th century BC based on typology (53). However, relative schemes for the chronological seriation of Iron Age pottery types generally have a century-scale resolution. Here, we demonstrate the potential for creating a high-resolution decade-scale framework for certain types within the late Iron Age pottery series (Fig. 3). The pottery types retrieved from the directly dated accumulation above the different floors of Room 17063 allow us to devise absolute ages for the time ranges during which these vessel types were used in the city and to establish a solid age comparison for other important Iron Age sites. As Room 17063 is rather small, a larger-scale, pottery-rich excavation can serve as a new scaffold for the southern Levant Iron Age pottery when excavated and sampled as demonstrated here for high-resolution radiocarbon dating.

Fig. 3.

High-resolution dating of specific pottery types within the Hallstatt plateau. “Single-layer vessels” refer to types that in this room were found only in one of the layers. “Multilayer vessels” refer to types that were found in several layers within the room. Absolute chronology on the Left is based on the new radiocarbon dates obtained in this work.

We observe that pottery types of important diagnostic value, broadly assigned to the Iron Age terminal phases (54), such as rosette stamped handles, stepped-rim and folded-rim holemouth jars, stump-base lamps, flat platters with an out-turned rim, and bag-shaped storage jars, were only found above the uppermost floor of Room 17063. As these types occur in our sequence from c. 650 BC onward, they serve as evidence that the part of the sequence below the uppermost floor ended in the early 7th century BC, supporting the conclusion based on wiggle matching of the radiocarbon dates.

Furthermore, pottery types typical of the early part of the late Iron Age, the 8th-century BC (54), such as storage jars with a thick inverted rim and spouted storage jars, appear exclusively in connection with Floors 9–11, dated to 780 to 715 BC. Folded rim bowls occur within Floors 4–8 (730 to 690 BC), and carinated cooking pots with a ridged rim, occurring within Floor 4 (710 to 690 BC) are often found in the same ceramic assemblages. This is typical of many Iron Age forms, which are thought to appear for long periods.

This is the first step in improving the ceramic dating. As the data accumulate and the assemblages of narrow slices of time are published, the absolute ceramic data will be broadened and improved, particularly if large assemblages are exposed.

Reconstructing Jerusalem’s Urban History (Fig. 4).

Fig. 4.

Modeled or unmodeled radiocarbon ages of 68.3% probability distribution, organized by archaeological areas and contexts. Within the contexts, samples are ordered according to stratigraphy, oldest at the Top. Where no stratigraphy could be established (see context with *), samples are ordered according to their radiocarbon ages. Thicker bars represent combined measurements related to a single context. The thickness is relative to the number of combined results shown Right to the bar. R_Combine details are in SI Appendix, Tables S3 and S5. Light gray parts of the probability distribution are the regions that are very unlikely, based on the archaeological considerations.

The 12th to 10th centuries BC: founding of Iron Age Jerusalem (various areas).

Altogether, almost 20% of the samples (18 dates) fall within the timeframe of the early Iron Age (12th to 10th centuries BC, Fig. 4). This is highly significant, since only in three cases (RTD10780, 9598, 9585) do the dates derive from contexts with clearly associated early Iron Age pottery, while the remaining dates come from charred remains from building materials. The abundance of early Iron Age dates, measured from all the areas in our study, clearly indicates widespread occupation of yet undetermined character, often underestimated due to the limited architectural contexts attributed to this period.

Our excavation in Area E (section T5), within a previously unearthed room of an Iron Age structure (SI Appendix, S3.4) (28), exposed a sequence of layers, attributed by Shiloh to Stratum 12 (the 8th century BC). Four dates assigned to the 16th, 12th, 10th, and 9th centuries BC were retrieved from a half-a-meter-thick accumulation of sediment and stones from that Stratum, apparently representing collapsed wall material (SI Appendix, Fig. S44, red circles). The accumulation was superimposed atop a series of crushed limestone floors, with a c. 800 BC terminus post quem determined for their construction, based on samples from a thin layer of ash (RTD8781) and soft phytolith-rich sediment with seed clusters (RTD8782, 8527), all underlying the lowermost earthen floor (SI Appendix, Fig. S44, green circles). These dates rule out the possibility that the collapse of the encompassing structure occurred at any other time than the late Iron Age, although its walls may have been constructed earlier in the Iron Age. Occupation layers from Shiloh’s Strata 15 to 14 (12th–10th centuries BC) within the same sequence, sandwiched between Stratum 12 and Middle Bronze Age strata 17 to 18 (23), yielded a 13th to 12th-century BC date from an in situ ash layer (RTD9585) and a 10th-century BC date on a piece of bone (SI Appendix, Fig. S44, yellow circles RTD9598), in agreement with the established relative chronology for these strata.

Tenth-century BC dates were determined based on two separately measured charred seeds (RTD10645, 11352) from mortar material in Area 10 (Building 100; Fig. 5), while a range of dates in the 12th to 9th centuries BC were measured from an accumulation below and above a segment of a floor within Building 100, which was understood to predate the building’s time of construction (Fig. 5). Another 10th-century BC date was retrieved from a fill sealed by a wall with corresponding pottery in Area 70-south. In this area, the radiocarbon samples measured were taken from the same locations as previously measured OSL (optically stimulated luminescence) dates (55). Interestingly, in two out of the three compared dates, the OSL dates are older, to the extent that the ranges of the two methods do not overlap (SI Appendix, Figs. S39 and S41).

Fig. 5.

Chronology of Area 10, Building 100. (A–E) Location and images of the dated samples. (A) Top plan of the building with the rooms’ designated letters. (B) Room C: Two plaster layers covering the wall and mortar between the stones. (C) Room A: Sycamore log separated in the field into segments along the radius for dating. The arrow marks the radius from the pith to the outermost ring. (D) Room B: An early floor and the passage where the bat skeleton was found. (E) Room C: A burnt destruction layer attributed to Nebuchadnezzar from 586 BC. (F) Stratigraphy-based model. Light gray parts of the plots mark the unmodeled calibrated ranges, while the dark parts mark the modeled calibrated ranges. Modeled results of the five consecutive samples from “Sycamore 1” log used for the second story construction are placed on the calibration curve. Two possible solutions exist for the outermost ring’s date. The green line marks the more probable felling date. Note that the calibration curve’s y-axis is flipped compared to Figs. 2 and 6. Three samples marked with? are identified as outliers and removed from the modeling.

A wide range of dates, spanning the 12th to 8th centuries BC, was determined for two plaster layers of a small room in Area C (Room SP 2482), abutting the Fortified Passage connecting the city and the Spring Tower (SI Appendix, Figs. S21, S23, and S25), while dates lying beneath the Spring Tower, part of the defense of the Gihon Spring, span between 10th to 9th centuries BC (29) (SI Appendix, Figs. S22 and S25). A 12th-century BC deposit of thick ash with abundant seeds was dated on the bedrock next to wall W20021 in a nearby Area U (SI Appendix, Figs. S18 and S20). A twig found within Room 17049 in Area U (RTD11182), perhaps representing roofing material, was dated to the 10th century BC (Figs. 1 and 4 and SI Appendix, Figs. S13 and S17).

The 9th century BC: urban expansion (Areas U, 10 and 70-south, E).

An appreciable number of dates from different excavation areas and a wide variety of contexts were dated to the 9th century BC, strongly suggesting that extensive building activity along the Southeastern Ridge was taking place at this time (Fig. 4). Thus, several seeds from an ash layer found below a thin wall in Area U (Room 19040), indicate a 9th century BC date for the construction of this room and adjacent structures, as well as the hewing of a series of rock-cut rooms to which the architectural remains were connected based on stratigraphic observations (SI Appendix, Figs. S18 and S20). Also dating to this century in Area U was a collapsed refuse of building materials, uncovered in Room 17063, built directly on top of the bedrock (RTD 9180, Fig. 2 and SI Appendix, Figs. S4, S9, and S12). In Area C, the construction (or rebuilding) of the Spring Tower surrounding Gihon Spring, the main water source of ancient Jerusalem, was dated to the same period (29) (SI Appendix, Figs. S21, S22, and S25), as well as an occupation surface in Area E directly underlying a series of white-crushed limestone floors (c. 800 BC), which were therefore laid down in the 9th century BC (SI Appendix, Fig. S44, green circles).

Building 100.

Importantly, the construction of the monumental Building 100 in Area 10 (Building Phase 10/IX) (56) was dated to the 9th century BC, together with additional construction activities in nearby Area 70-south (Fig. 5. See SI Appendix, S3.3 for detailed description and OxCal models). Three samples (RTD10747.1-2, 10751) from a building phase immediately preceding that of Building 100 (Building Phase 10/X) were dated, two to the 10th and one to the 9th century BC; the samples were retrieved from a context underlying a patchy surface of crushed limestone (Floor 1591; SI Appendix, Figs. S33–S35) that rested only 10 cm above bedrock and was cut by two of the walls of Building 100 (W1338 and W1318 of Room B). Three samples overlying the crushed limestone surface were dated (RTD10748-50), one to the 12th century BC and two to the 9th century BC. Calculating the time of use of the two surfaces with an outlier model in OxCal (57) constrains the construction date of Building 100 to 900 to 850 BC (Fig. 5 D and F and SI Appendix, Fig. S42).

Apparently, sediments including older material were used for construction as the two dates measured from mortar between the stones of one of its walls (W1352 of Room C) gave 10th-century BC dates (RTD10645, 11352). Two layers of plaster securely covered this mortar. Three dates (charred straw fragments and seeds) from the inner plaster coating fell within the Hallstatt Plateau (RTD11346, 11350-1; Fig. 5 B and F and SI Appendix, Figs. S31 and S32). Another Phase 10/IX was dated 790 to 760 BC and was attributed to the first major renovation of Building 100. The dated material was skeletal remains of a bat from a fill context within a passage connecting two of the building’s rooms (Rooms B and C; W1352, Fig. 5 D and F and SI Appendix, Fig. S36). The passage was blocked by fieldstone walls on both sides and sealed from above by a 1 m thick fill beneath the renovated floor.

The following dates reinforce the suggestion that the city expanded westward in the 9th century BC, and possibly earlier (48, 58): the secure date of the bat context; the assemblage of earlier dates from within the walls; and the accumulation above the terraced bedrock, which was intentionally quarried as part of large-scale terracing of the slope undertaken before Building 100. The building’s first renovation yielded time ranges that overlapped those of the dates for the major mid-8th century BC earthquake (see below, 766 to 750 BC), raising the possibility that the earthquake was the cause of the renovation. A second renovation took place at 680 to 670 BC when a second story using sycamore beams was erected (Fig. 5 and SI Appendix, Figs. S37 and S38). The building was replastered at least twice as evident from a wall of Room A. The whole building was destroyed by fire on the historically secured date of 586 BC (see below) (32).

The mid-8th century BC: Earthquake (Area U, E).

Construction damage, including abundant collapsed stones and building materials attributed to the mid-8th century BC earthquake that devastated the city, was uncovered on top of leveled bedrock in another room of Area U (Room 17130), abutting Room 17063 to its south. A row of complete 8th-century BC in situ vessels, originally placed against the room’s northern wall, was found smashed beneath collapsed stones, and behind them a piglet skeleton in standing position (50), indicating it had been caught in the room’s sudden destruction.

Six ¹⁴C dates were obtained from various charred organic materials within a deposit of thick ashy sediment found in the middle of the room, on top of bedrock, and below the accumulation of collapsed building materials. The combined calibrated date of the six determinations intersects the Hallstatt Plateau at four junctures in the span of 766 to 580 BC (SI Appendix, Fig. S15 and Table S5). Of these, we consider the earliest probability distribution of 766 to 750 BC as the most likely range. This assertion is further reinforced by a close stratigraphic and chronological correspondence between the earliest occupation surface of Room 17130 and that of adjacent Room 17063, with its tight sequence of dates (Fig. 2 blue plots, c. 830 to 720 BC). Furthermore, the presence of 8th-century BC pottery in association with Room 17130 (37), with similar qualities to the material uncovered in Shiloh’s Stratum 12b (28, 50), offers further reinforcing evidence of the proposed chronology. The combined uncalibrated date of 2507 ± 8 uncal BP can be used to identify the same seismic event in other Near Eastern sites and to inform the calibrated result within the Hallstatt Plateau.

The late 8th century BC: Rebuilding and fortification (Areas E and U).

Widespread building activity of the 8th century BC, some involving the renovation of previously erected structures, is attested in Areas U and E, where we have much evidence of collapse due to the massive earthquake of that century. A partial collapse of the Spring Tower and the Fortified Passage was also documented (48). The erection of a new fortification, along the eastern slope of the Southeastern Ridge, is shown to overlie building remains which seem to have been damaged in the earthquake. Our dating of a construction used to block the entrance to Room 17063 at 730 to 710 BC (Floor 10, RTD9128, Fig. 2, purple sample) suggests that this activity postdates the earthquake and coincided with the erection of the new fortification line. The floor was laid down within this room above collapsed debris after the earthquake abuts the blockaded entrance and hence, postdates it. The Mid-Slope Fortification formed a support wall for buildings erected west and upslope of the wall, such as the building to which Room 17063 belonged. The room was boarded up because the new Mid-Slope city wall was built against it, effectively nullifying the said entrance.

Therefore, while the major building project of Jerusalem’s 8th-century BC fortifications was previously assigned to King Hezekiah in the late 8th century BC (e.g., ref. 59) based on our chronology, these activities can now be associated with the latter years of King Uzziah, whose reign spanned the mid-8th century BC, suggesting the city was fortified during the Syro-Ephraimite war (2nd Kings 15 to 16) (49).

The 7th–early 6th centuries BC: Renewed expansion and stability (Areas 10 and U).

Our dating of a charred Ficus sycomorus wood beam (“Sycamore 1”) from Room A of Building 100 allowed wiggle-matching of a sequence of five dates taken at regular intervals along the beam’s growth radius (Fig. 5 C and F and SI Appendix, Fig. S38). The most probable felling time for this beam was determined at 680 to 670 BC or 610 to 590 BC. The outermost rings of two other charred F. sycomorus beams (RTD10652, 10752) and one Ceratonia siliqua beam (RTD10653) date either a decade before or after the wiggle-matched radiocarbon peak of Sycamore 1. The beams are understood to have been the supports of a massive floor of polished plaster, built in a second renovation stage as the second story of Building 100 (36). Drawing on historical considerations, we assign the construction of the second story to c. 680 to 670 BC, during the long reign of King Manasseh (60) rather than 610 to 590 BC, 10 y before the historical date for the destruction of Jerusalem in 586 BC. The building was finally toppled by the massive conflagration of 586 BC, as dated using pottery found on the building’s destruction deposit floor (31–33, 61).

The lengthy period of use of Building 100, determined from our radiocarbon dates, may be emblematic of the city’s long-term economic flourishing and relative political stability (62, 63) up until the Babylonian invasion at the beginning of the 6th century BC. The city’s continuous occupation indicates a time of demographic growth while recurring conflicts with regional empires negatively affected settlements elsewhere in the region.

Remarkably, while our radiocarbon determinations demonstrate a ~300-y use of Building 100 (Fig. 5), the pottery found in association with the building belongs almost entirely to the end of the 7th to the early 6th centuries BC, in the terminal Iron Age (for a similar assemblage, see ref. 35). Similarly, in Area U, long-term use is documented by the radiocarbon dates for several bedrock floors and the uppermost white-sediment floor of Room 17063 (see above). Also, in Area E, a white-sediment floor of Shiloh’s Stratum 12 (generally assigned to the 8th century BC) yielded four older dates. These observations suggest that these well-prepared, long-used floors were continually kept clean while in use so that most of the material accumulated on top of them originates from the final phase of occupation and the collapse of wall material from the structures. Thus, the radiocarbon determinations from various building materials and floors provide a more realistic approximation of the chronology of the analyzed structures and the history of occupation of buildings and entire settlements, than does the pottery and other remains of material culture.

Offsets from the Calibration Curve.

In this study, we identified several offsets, in which dated samples from contexts with known expected ages securely dated by independent historic and stratigraphic records, provided ¹⁴C measurements either above or below what might be predicted for these dates based on the existing calibration curve, IntCal20 (9) (Fig. 6).

Fig. 6.

The identified offsets from the northern hemispheric calibration curve IntCal20. The blue dots are the set of measurements that have been used to construct the calibration curve. Samples from the 710 BC 14C enrichment offset from area U, room 17063 (colored ellipses correspond to the modeled ranges in Fig. 2). The black squares mark two samples from Building 100 (Area 10) that most likely correspond to c. 710 BC, but could also represent the c. 660 BC peak (connected by dotted lines). The open circles mark the samples from the 586 BC destruction levels at Area 10 (black circles) and Area U (turquoise circles). 14C measurements of individual rings from Irish oak are marked by brown squares and of Bristlecone pine by green triangles (this work). The Inset is a zoom-in of the 580 to 600 BC region, where the black circles are short-lived samples, and the red circles are twigs and a bone. All y-error bars are 1σ. The x-error bars represent the number of tree rings dated together.

¹⁴C enrichment c. 730 to 710 BC.

The radiocarbon measurements from Floors 10–7 in Room 17063 of Area U (Fig. 6, elongated ellipses), and those of two samples from Building 100 (Area 10)—one from a beam (Sycamore 1, RTD10683) and another from a wall plaster (RTD9346, Figs. 5F and 6, black squares; SI Appendix, Fig. S29), lie beneath the calibration curve. These dates are all from contexts clearly predating the destruction of Jerusalem in the historically secured date of 586 BC and yet, the calibrated measurements obtained postdate the destruction. The five dates retrieved from Sycamore 1, along its radius, demonstrate a rapid decrease in the measured radiocarbon amount, with the earliest determination, from the inner part of the beam, indicating an enriched radiocarbon content, measuring 2396 ± 31 uncal BP. These results suggest a higher influx of radiocarbon around 730 to 710 BC than predicted by the calibration curve. A similar effect was identified at around 2830 BC at the tell sites of Megiddo and Bet Yerah (SI Appendix, S4 and Fig. S48).

It is noteworthy that these results are consistent with some of the IntCal20 raw data for this time period (specifically those reported by Fahrni et al., 10), which also lie below the curve during the time interval in question. The fact that this offset and the one at 2830 BC occur below sharp minimum peaks of the calibration curve indicates that caution should be exercised when calibrating measurements that fall within such zones.

¹⁴C depletion c. 586 BC.

Here we focused on remains previously identified as evidence of the historically dated Babylonian destruction of Jerusalem, comprising burned accumulations at Area 10 (11 samples, Fig. 6, black circles; for context description, see SI Appendix) and finds in Area U, Room 17049 (four dates, Fig. 6, turquoise circles. SI Appendix, Figs. S13 and S17). Ten samples are short-lived (various seeds and linen), four are twigs, and one is a bone (Fig. 6, Inset and SI Appendix, Table S3). We expected most of the short-lived seeds from these contexts to be between 590 to 586 BC, slightly prior to the destruction event since the siege began two and a half years before the destruction occurred. (2nd Kings, Ch. 25) (31, 33). Nevertheless, out of the 15 measured samples, only one coincides with the calibration curve (RTD10755), while the remaining 14 samples lie above it. The 15 samples’ combined radiocarbon age is 2530 ± 6 uncal BP, whereas the one of only the short-lived samples is 2520 ± 7 uncal BP (see SI Appendix, Table S3 for R_Combine details). The latter is 40 to 24 ¹⁴C years above the center of the calibration curve for the years 586 to 590 BC.

Several reasons eliminate a laboratory bias explanation for the two types of offset observed. First, we found two opposite offsets, each with very consistent measurements. Second, the samples were measured in our laboratory over a period of several years, giving consistent results; and third, the 2830 BC offset has the same character as the 730 to 710 BC one and includes samples measured at another laboratory (SI Appendix, Fig. S48).

As the years 624 to 586 BC in the calibration curve are still represented by older measurements where each combines 5 to 20 rings, we performed 100 measurements of individual growth rings (alpha-cellulose extracted from each full ring) of both Irish oak and Bristlecone pine (Fig. 6, brown squares and green triangles, respectively; SI Appendix, Table S4) to test whether the observed offset was due to the calibration curve’s low resolution. Our annually resolved series indicates that some regions of the calibration curve at 624 to 590 BC may need to be shifted higher in the graph (a position with less ¹⁴C), but this up-shift is not evident in the relevant 590 to 586 BC curve segment (Fig. 6, Inset). Therefore, the offset we identified relates to some real site or regionally specific effect.

Dee et al. (18) found an offset with a ¹⁴C depletion of 19 ± 5 y in Egypt due to a shift in the regional growing season (winter/spring) compared to that of the trees of higher latitudes used for assembling the calibration curve (spring/summer). It was later reduced to 12.5 y when comparing AMS measurements with segments in the calibration curve that were also measured by AMS (64, 65). In the present case, among the 10 dated short-lived samples, four are olive pits and one a grape pip (see SI Appendix, Table S3 for the samples’ growth seasons), which grow in the spring/summer and, hence, are not expected to record an offset such as the one recorded in Egypt. Reducing the proposed 19 y Egypt offset from the remaining five samples (pea, wheat, barley, and linen; fall/winter/spring growing season) and combining them with the olives and the grape pip results in a date of 2510 ± 7 uncal BP, still above the calibration curve by 14 to 21 y.

Using the 586 BC historical event, we detected a part of the global calibration curve where improved resolution will contribute to our ability to identify and account for regional offsets. Our archaeological data better fit the annually resolved measurement series we assembled for this study than the current IntCal20 calibration curve for the timespan in question. Still, we note that a future more-resolved calibration curve will not be shifted to the extent that can cancel out the observed offset in 590 to 586 BC. Although the 14 to 21 y-difference between our combined date from 10 samples and the calibration curve is within the measurement error, with nine out of the 10 short-lived samples falling clearly above the curve, we lean toward the possibility that they represent a regional or site-specific radiocarbon offset, not represented by the Irish oak and the Bristlecone pine, even though the latter tree grew at a latitude similar to that of Jerusalem. Furthermore, this study highlights that where contexts are dated historically with certainty, as for the examples in this study, these dates can then be used effectively to better tease apart possible global/regional/seasonal/subannual effects on the radiocarbon budget.

Conclusions

It is striking that despite the centrality of Iron Age Jerusalem to research on the history of major biblical narratives relating to the ancient Near East and lingering questions about the origins of state societies, systematic radiocarbon dating has not previously been employed to comprehensively address highly controversial and intriguing issues in the city’s early history. This study presents the first major contribution to this effort by assembling an absolute and high-resolution chronology, covering a substantial part of the Iron Age and the time of some of the most consequential events in the city’s emergence as the capital of a regional kingdom and early state society.

Our study charts a way forward to resolving long-existing challenges in high-precision dating during periods coinciding with the Hallstatt Plateau in the global radiocarbon calibration curve. This involves the careful coupling of well-controlled stratigraphic and contextual data combined with dense sampling, information from ancient textual sources, ceramic chrono-typology, and absolute dating, together with the production of annually resolved ¹⁴C data from calendar-dated tree rings. With this approach, instead of calibrated ranges spanning 300 y, we succeeded in reaching decadal precision and identified two regional offsets from the global calibration curve during the Iron Age, c. 730 to 710 BC and 586 BC, with potentially far-reaching implications for future ¹⁴C applications, and the possibility to chronologically better constrain local series of pottery assemblages, widely employed in the relative dating of Jerusalem and elsewhere.

Here, we assessed the validity and accuracy of our dating framework by employing two widely known historical events that broadly impacted Iron Age Jerusalem as chronological pegs: the mid-8th century BC earthquake and the 586 BC Babylonian destruction. Thus, we are able to describe Jerusalem’s building history over c. 500 y of the Iron Age with sub-decadal detail and fidelity:

• 1.We provide concrete evidence for a widespread human presence in the city during the 12th to 10th centuries BC, as short-lived material from all the areas studied was dated to these centuries. However, these contexts were often lacking diagnostic ceramics. This signifies that radiocarbon dating on a broad scale should be used to provide a more reliable site history, particularly in a densely rebuilt city, where foundations often reach the bedrock.
• 2.The city’s westward spread out of its ancient core is assigned here to at least as early as the 9th century BC, as attested by the radiocarbon dating of the construction of monumental Building 100 in excavation Area 10 (GPL) and the large-scale terracing activities of the bedrock done prior to its construction.
• 3.Major changes in the city’s urban plan were shown to follow the devastating earthquake of c. 760 BC, including the building of a Mid-Slope Fortification and a retaining wall along the eastern slope of the Southeastern Ridge, which are attributed by us to the time of King Uzziah rather than the more conventional assignment to that of King Hezekiah.
• 4.Ongoing development of the city is documented between the 8th and the 6th centuries BC, until the destruction in 586 BC.

Materials and Methods

See SI Appendix, S2 for details. All the samples for radiocarbon dating were collected in the field for chronology-building purposes, combined with microarchaeological analyses of sediments as part of context in situ verification (15, 16, 22, 23, 66, 67). Fourier-transform infrared spectroscopy was used to identify the presence of anthropogenic material, such as burnt clay (68), phosphate (69, 70), and disordered calcite (71). Charred remains were treated according to the acid–base–acid (ABA) method (72). A single charred fragment was used for dating, unless for tiny samples such as straw, where fragments were combined into one sample. Collagen was extracted and purified from bone samples with the ABA procedure followed by gelatinization and ultrafiltration methods (73, 74). Graphitization was done using an EA-AGE3 system (Ionplus). The ¹⁴C content was measured at the DANGOOR Research Accelerator Mass Spectrometry Laboratory at the Weizmann Institute of Science (75). The radiocarbon ages were calibrated and modeled using OxCal version 4.4.2 (52) according to IntCal20 atmospheric curve (9). Unless otherwise stated, all calibrated dates are of the 68.3% probability range.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.