## Abstract

Hershkovitz *et al*. (Reports, 26 January 2018, p. 456) interpreted the Misliya-1 fossil maxilla as evidence of the earliest known anatomically modern human outside Africa. However, the fossil’s reported age of 177,000 to 194,000 years relies on flawed interpretations of uranium-series data. We contend that those data support a minimum age of no more than ~60,000 to 70,000 years.

Misliya-1, part of the maxilla of an anatomically modern human from Misliya Cave, Israel, was assigned an age of 177 to 194 thousand years (ky) (*1*). Methods using U-series disequilibrium to date the fossil included combined U-series and electron spin resonance (US-ESR) on tooth enamel, laser-ablation ^{230}Th/U dating of dentine, and ^{230}Th/U dating of soil carbonate encrusting the maxilla. The US-ESR age estimate of 174 ± 20 ky (all errors 2σ) was considered a maximum age because the fossil was irradiated during three computed tomography scans prior to dating (*1*). Laser spots across a section of dentine yielded broadly consistent ^{230}Th/U age estimates between 83 ± 9 and 62 ± 3 ky. A ^{230}Th/U age estimate of 185 ± 8 ky on soil carbonate was interpreted as a minimum age for the fossil; however, that date is not supported by the U-series data in (*1*).

A reported age of 185 ± 8 ky for crust sample Maxilla-6 is crucial to the central claim of Hershkovitz *et al*., but that sample has clearly failed to maintain a closed U-series system. The authors erroneously reported an uncorrected age of “equilibrium” for Maxilla-6 [table S2 of (*1*)]. In fact, Maxilla-6 has an uncorrected ^{230}Th/^{234}U activity ratio greater than 1.0, indicating an excess of ^{230}Th over its parent ^{234}U (*2*). This condition is commonly produced when materials lose uranium relative to thorium after deposition, thereby violating a fundamental premise of radiometric dating. We contend that the reported age for Maxilla-6 was obtained by incorrectly computing detritus-corrected U-series isotope ratios, as described below.

Figure 1 shows measured and detritus-corrected activity ratios (solid ellipses and red crosses, respectively) for all eight maxilla crusts analyzed in (*1*). Maxilla-6 may have originally had a composition similar to other crusts; however, its measured ^{230}Th/^{238}U activity ratio is too high to be produced by closed-system radioactive decay, indicating that the sample has likely lost uranium after crust formation (Fig. 1, thick arrow).

Hershkovitz *et al*. recognized that measured U and Th compositions represent mixtures of authigenic soil carbonate and detrital silicate containing extraneous ^{230}Th, ^{234}U, and ^{238}U proportional to the amount of common Th (^{232}Th) present in the detritus. Using the observed ^{232}Th abundance and a model detritus composition, we derived the composition of the authigenic carbonate by mathematically subtracting U-series isotopes associated with the detritus from measured compositions. In Fig. 1, model detritus (red star) is assumed to have a Th/U value similar to Earth’s average upper crust and U-series isotopic ratios in secular equilibrium (^{230}Th/^{238}U and ^{234}U/^{238}U activity ratios = 1.0) as in (*1*). Subtraction of detritus moves the calculated detritus-corrected compositions radially away from the model detritus by an amount that scales with the level of contamination (solid black arrows) (*3*). Proper correction for detritus in Maxilla-6 likewise moves its composition away from the model detritus, further into the open-system region (dashed arrow). However, the detritus-corrected composition reported for Maxilla-6 in (*1*) is displaced in the opposite direction (Fig. 1, red cross #6). The erroneous corrected composition is misleading because it implies that an age can be calculated for Maxilla-6, which has clearly failed to maintain a closed U-series system. Moreover, the inaccurate corrected composition for Maxilla-6 in (*1*) yields an unreasonable initial ^{234}U/^{238}U activity of ~0.75, unlike other soil carbonates from Misliya Cave and elsewhere that are enriched rather than depleted in ^{234}U relative to ^{238}U (*4*, *5*).

Calculations using Isoplot (*6*), an application widely used to reduce U-series data, produce an inaccurate detritus-corrected composition and age for Maxilla-6 similar to those given in (*1*). This is apparently a consequence of the unusual composition of Maxilla-6, which has a higher Th/U value than the model detritus, making it unsuitable for U-series dating. We infer that the ^{232}Th-rich composition of Maxilla-6 lies outside the range of applicability of the algorithm used by Isoplot for calculating detritus-corrected compositions.

Detritus-corrected ^{230}Th/U age estimates of 70 to 19 ky were reported for seven additional maxilla crust samples (*1*). All are moderately to highly contaminated by silicate detritus. Assuming a detritus composition like that in (*1*) and uncertainty propagation as in (*5*, *7*), we calculate ^{230}Th/U ages of 60 ± 20 and 61.7 ± 3.8 ky for the two least contaminated samples, Maxilla-7 and -8. Use of a three-dimensional ^{230}Th-^{232}Th-^{234}U-^{238}U isochron approach, which requires no assumptions regarding the isotopic composition of detritus, shows that Maxilla-7 and -8 are approximately collinear with the model detritus composition (Fig. 2). A regression using all measured data, save Maxilla-6, yields an estimated isochron age of 60.6 ± 8.4 ky, albeit with a high degree of scatter. Finite age estimates can be calculated for the more severely contaminated samples Maxilla-1 through -5; however, our results yield age uncertainties ranging from 56 to >350%. Much smaller errors for corrected U-series ages are reported in Hershkovitz *et al*. [table S2 of (*1*)] as a result of failing to propagate uncertainty arising from the large detritus corrections. We contend that such errors severely underestimate actual uncertainties.

We note that compositions and age estimates of the most suitable crusts, Maxilla-7 and -8, are broadly consistent with results from laser-ablation U-series analyses of fossil dentine (Fig. 1, open ellipses) with a mean age of 70.2 ± 1.6 ky (*1*).

In sum, most samples of carbonate encrusting Misliya-1 analyzed in Hershkovitz *et al*. are unsuitable for U-series age interpretation. Only two carbonate crust samples are sufficiently pure to provide reliable U-series dates. If they have maintained closed U-series systems, results for Maxilla-7 and -8 support a minimum age of ~60 ky for the fossil, similar to values determined by laser ablation on dentine (*1*). We contend that there are no reliable U-series dates for Misliya-1 older than ~70 ky. Previously known fossils of anatomically modern humans in the Levant are as old as 90 to 120 ky (*8*–*10*). Thus, the claim that Misliya-1 is the earliest such fossil is not supported by U-series dating in Hershkovitz *et al*. Published thermoluminescence dates from burnt lithic artifacts from Misliya Cave suggest ages of ~140 to 212 ky for associated sediments (*1*, *11*), but the available dates on the human fossil cannot rule out the possibility that the fossil dates to a younger interval.

This is an article distributed under the terms of the Science Journals Default License.

## References and Notes

- ↵
- ↵The decay rate or activity of a nuclide is given by λ
*n*, where λ is the decay constant and*n*is the number of atoms. U-series nuclides in a material that remains undisturbed for a few million years reach a state of secular equilibrium where their activities become equal and their activity ratios reach unity. - ↵We were unable to reproduce the detritus-corrected composition of crust sample Maxilla-2 from the uncorrected ratios in table S2 of (
*1*); therefore, we show our result for the detritus-corrected composition of Maxilla-2 in Fig. 1. - ↵
- ↵
- ↵
- ↵Assumed activity ratios of detritus:
^{232}Th/^{238}U = 1.2 ± 0.6,^{230}Th/^{238}U = 1.0 ± 0.25, and^{234}U/^{238}U = 1.0 ± 0.25. Decay constants used are those in (*12*). - ↵
- ↵
- ↵
- ↵