Extended dilation of the radiocarbon time scale between 40,000 and 48,000 y BP and the overlap between Neanderthals and Homo sapiens

Abstract

The new radiocarbon calibration curve (IntCal20) allows us to calculate the gradient of the relationship between ¹⁴C age and calendar age over the past 55 millennia before the present (55 ka BP). The new gradient curve exhibits a prolonged and prominent maximum between 48 and 40 ka BP during which the radiocarbon clock runs almost twice as fast as it should. This radiocarbon time dilation is due to the increase in the atmospheric ¹⁴C/¹²C ratio caused by the ¹⁴C production rise linked to the transition into the Laschamp geomagnetic excursion centered around 41 ka BP. The major maximum in the gradient from 48 to 40 ka BP is a new feature of the IntCal20 calibration curve, with far-reaching impacts for scientific communities, such as prehistory and paleoclimatology, relying on accurate ages in this time range. To illustrate, we consider the duration of the overlap between Neanderthals and Homo sapiens in Eurasia.

The radiocarbon method is the most widely used dating method over the past 55 ka BP. It relies on the beta decay of the ¹⁴C isotope produced in the upper atmosphere by interaction with cosmic-ray particles. Samples of organic material or carbonates dated by ¹⁴C either incorporated their carbon directly from the atmosphere as in plant photosynthesis, indirectly through the food chain, or by various chemical reactions. The ¹⁴C content of a fossil sample is compared to the atmospheric ¹⁴C content, which constitutes the starting reference for its disappearance by radioactive decay with a half-life of 5,700 ± 30 y.

In its raw form, the ¹⁴C method is not accurate as atmospheric ¹⁴C content has not been constant over time, instead having varied due to changes in its production rate and global carbon cycle rearrangements. To calculate a true calendar age from a ¹⁴C measurement, one needs to know the initial atmospheric ¹⁴C/¹²C ratio at the time the sample carbon last exchanged with the atmosphere. We therefore calibrate the radiocarbon method by comparing ¹⁴C measurements against samples for which accurate (“true”) ages have been measured with independent dating techniques such as counting tree rings in subfossil tree logs, counting annually laminated sediments, or dating corals and stalagmites using uranium–thorium (U-Th). Over the past three decades, the resulting radiocarbon calibration curves have been provided by an international working group (IntCal). The new IntCal20 curve covering the past 55 ka BP has just been published (1), updating the previous IntCal13 version (2).

Over the past 55 ka, the ¹⁴C calibration curve shows that ¹⁴C ages are usually younger than true ages (1, 2), that is, the ¹⁴C clock generally ticks at a slower pace than it should. This is due to the overall decrease of the atmospheric ¹⁴C/¹²C ratio over the past 40 ka (Fig. 1D), which partly compensates for the loss by radioactive decay in dated samples. Additionally, the relationship between ¹⁴C ages and true calendar ages is far from linear. Compression of the ¹⁴C time scale is particularly obvious during specific periods called ¹⁴C age plateaus, when the decreasing atmospheric ¹⁴C/¹²C ratio fully compensates for radioactive decay. This implies that archeological sites and artifacts from these periods cannot be dated precisely with radiocarbon. The ¹⁴C age plateaus on the order of a few centuries are linked to modulation of cosmogenic production by variable solar activity. Longer age plateaus may correspond to changes in the carbon cycle and deep ocean circulation, for example during the plateau that occurred at the end of the Younger Dryas climatic event (3–5).

Fig. 1.

(A) First derivative of the IntCal20 curve (1) computed by calculating the ¹⁴C age vs. calendar age gradient over sliding windows of between 5,000- and 2,000-y durations. The purple area shows the 95% confidence interval for the 3,000-y window (to maintain legibility this is the only interval plotted). The green curve shows the first derivative of the IntCal13 curve (2) using the 3,000-y window (to be compared with the purple curve for IntCal20). When the gradient is above (below) unity, the ¹⁴C clock ticks faster (slower) than it should. (B) Evolution of the intensity (in 10²²Am²) of the geomagnetic field (6); note the reversed scale. (C) The ¹⁰Be flux (in 10⁶ atoms⋅cm⁻²⋅y⁻¹) measured in Greenland ice cores (7). (D) Atmospheric ∆¹⁴C (in per mille above modern) based on the IntCal20 in red (1) and IntCal13 in green (2). The prominent maximum of the gradient curve (A) centered around 43 ka BP corresponds to the rising phase of the ∆¹⁴C curve (D) and thus predates the ∆¹⁴C and ¹⁰Be flux maxima (C) and the paleomagnetic intensity minimum (B) corresponding to the Laschamp geomagnetic excursion (vertical dashed line).

In parallel to periods when the ¹⁴C clock runs too slowly, there are also specific periods characterized by an increasing atmospheric ¹⁴C/¹²C ratio—also a consequence of solar and carbon cycle changes. Here, the ¹⁴C clock ticks faster than it should.

One major outcome of the recent IntCal20 curve is that the pace of the ¹⁴C clock can be calculated at unprecedented precision. We have computed the evolution of the first derivative of the IntCal20 curve, focusing on its multimillennial component (Fig. 1A). Over most of the past 55 ka, the gradient is below 1, meaning that the ¹⁴C time scale is mostly compressed. However, the gradient is also characterized by a prominent maximum from 48 to 40 ka BP, reaching values up to 1.5 to 2 for the different sliding windows. Over this multimillennial period, there are about twice as many ¹⁴C years as calendar years.

This expanded ¹⁴C time scale was absent, or much less prominent, in former calibration curves, as seen by the comparison with IntCal13 (Fig. 1A). The 48 to 40 ka BP gradient maximum is directly tied to the rising trend of atmospheric ∆¹⁴C, minimal until 48 ka BP before rising by more than 500‰ over a period of only a few millennia, to reach a maximum around 40 ka BP. As shown, the ∆¹⁴C maximum (Fig. 1D) is broadly in phase with the minimum intensity of the geomagnetic field (Fig. 1B) during the Laschamp excursion (6) and the maximum concentration of ¹⁰Be (Fig. 1C) measured in polar ice (7)—¹⁰Be is also formed by cosmic-ray particles. The precise relationships, in phase and amplitude, between ¹⁴C, ¹⁰Be, and paleomagnetic intensity are complex, notably because ¹⁴C atoms are mixed in the global carbon cycle, but can be studied with numerical models (7–10).

The maximum ∆¹⁴C value above modern, ∼700‰ at 41 ka BP, has been known from ¹⁴C and U-Th dating in corals since the late 1990s (8), but the earlier ∆¹⁴C minimum around 50 to 45 ka BP was only evidenced later with marine sediments (9, 11). The ∆¹⁴C record prior to 40 ka BP has been refined with independent data based on counting tree rings in subfossil kauri logs from New Zealand (12), and with ¹⁴C and U-Th dating of stalagmites from the Hulu Cave in China (10). These independent data and their updates confirmed each other and were used collectively with updated statistical techniques (13) in order to construct the new IntCal20 calibration curve (1).

The radiocarbon time dilation over the 48 to 40 ka BP window, occurring just before the ∆¹⁴C maximum, is thus a novel and major feature of the new IntCal20 curve. This time expansion effect has remained unnoticed, even though it was present to a lower extent in IntCal13. The difference between IntCa20 and IntCal13 is mainly linked to new data, notably Hulu stalagmites (10) and kauri trees (12), corrections and screening of existing data (1), and improved statistical modeling (13).

To illustrate the impact of the 48- to 40-ka-BP time dilation, Fig. 2 presents the ¹⁴C and calendar chronologies of a selection of prehistoric sites dated using ¹⁴C from human bone collagen. This includes famous sites occupied by Neanderthals (in red) and by Homo sapiens (in blue). In terms of the radiocarbon clock, the chronological overlap between the oldest H. sapiens remains (Bacho Kiro Cave) and the youngest Neanderthal age (Saint-Césaire) is 6,250 ± 910 ¹⁴C years (uncalibrated). When calibrated against the IntCal13 and IntCal20 curves, this difference is reduced to 5,000 ± 860 and 3,960 ± 710 calendar years, respectively, clearly illustrating how the expanded ¹⁴C time scale is compressed by about 60% after conversion to calendar ages with the new IntCal20 calibration. We note, however, that to investigate possible cultural and genetic exchanges, contact between two populations should be considered at a regional scale for adjacent sites.

Fig. 2.

Comparison between radiocarbon ages (Upper) and corresponding calibrated ages with IntCal13 (Middle) and IntCal20 (Lower) for a selection of human bone samples of Neanderthals (red) and early H. sapiens (blue). The ¹⁴C ages (±1σ) were calibrated using IntCal13 (ref. 2, Middle) and IntCal20 (ref. 1, Lower) in OxCal 4.2 (16). Note that the time axes of the three panels have exactly the same duration (13,000 y). The oblique dashed lines highlight the time dilation effect centered around 43 ka BP. Dataset S1 provides data and sources.

For prehistory and human evolution, the impact of the 48 to 40 ka BP time dilation goes beyond the study of late Neanderthal and early H. sapiens in Europe. Indeed, it will also affect the current discussion on H. sapiens spread across Eurasia and into Australia and help improve the genetic clock with a better calibration of genome mutation rates (e.g., refs. 14 and 15).

The new prominent maximum between 48 and 40 ka BP in the gradient between ¹⁴C and calendar years is important as it enables improved resolution to separate events during this period (e.g., different stratigraphic levels in the same site). In addition to the compression converting ¹⁴C to calendar time, the combined effects of the radiocarbon time dilation and the IntCal data improvements also lead to increased calendar age precision (e.g., the 1$\sigma$ uncertainty for the Les Cottés Neanderthal in Fig. 2 ranges from 270 in ¹⁴C years to 250 and 160 calendar years with IntCal13 and IntCal20, respectively). Determining the relative calendar age ordering of multiple events dated by ¹⁴C in this period is therefore not affected. The structure of IntCal20 beyond 40 ka BP reinforces the need to measure accurately and precisely the small ¹⁴C content of old samples, and in particular the use of updated pretreatment techniques to purify the original carbon fraction in order to eliminate residual contaminations.

Data Availability.

All study data are included in the paper and Dataset S1.