The age of the hominin fossils from Jebel Irhoud, Morocco, and the origins of the Middle Stone Age

Abstract

The timing and location of the emergence of our species and of associated behavioural changes are crucial for our understanding of human evolution. The earliest fossil attributed to a modern form of Homo sapiens comes from eastern Africa and is approximately 195 thousand years old, therefore the emergence of modern human biology is commonly placed at around 200 thousand years ago. The earliest Middle Stone Age assemblages come from eastern and southern Africa but date much earlier. Here we report the ages, determined by thermoluminescence dating, of fire-heated flint artefacts obtained from new excavations at the Middle Stone Age site of Jebel Irhoud, Morocco, which are directly associated with newly discovered remains of H. sapiens. A weighted average age places these Middle Stone Age artefacts and fossils at 315 ± 34 thousand years ago. Support is obtained through the recalculated uranium series with electron spin resonance date of 286 ± 32 thousand years ago for a tooth from the Irhoud 3 hominin mandible. These ages are also consistent with the faunal and microfaunal assemblages and almost double the previous age estimates for the lower part of the deposits. The north African site of Jebel Irhoud contains one of the earliest directly dated Middle Stone Age assemblages, and its associated human remains are the oldest reported for H. sapiens. The emergence of our species and of the Middle Stone Age appear to be close in time, and these data suggest a larger scale, potentially pan-African, origin for both.

Full text

8 JUNE 2017 | VOL 546 | NATURE | 293
LETTER doi:10.1038/nature22335
The age of the hominin fossils from Jebel Irhoud,
Morocco, and the origins of the Middle Stone Age
Daniel Richter1,2,3, Rainer Grün4,5, Renaud Joannes-Boyau4,6, Teresa E. Steele1,7, Fethi Amani8, Mathieu Rué9,10,
Paul Fernandes9,11, Jean-Paul Raynal1,11, Denis Geraads1,12, Abdelouahed Ben-Ncer8, Jean-Jacques Hublin1,13 &
Shannon P. McPherron1
The timing and location of the emergence of our species and of
associated behavioural changes are crucial for our understanding
of human evolution. The earliest fossil attributed to a modern form
of Homo sapiens comes from eastern Africa and is approximately 195
thousand years old
1,2
, therefore the emergence of modern human
biology is commonly placed at around 200 thousand years ago
3,4
.
The earliest Middle Stone Age assemblages come from eastern and
southern Africa but date much earlier
5–7
. Here we report the ages,
determined by thermoluminescence dating, of fire-heated flint
artefacts obtained from new excavations at the Middle Stone Age site
of Jebel Irhoud, Morocco, which are directly associated with newly
discovered remains of H. sapiens
8
. A weighted average age places
these Middle Stone Age artefacts and fossils at 315 ± 34thousand
years ago. Support is obtained through the recalculated uranium
series with electron spin resonance date of 286 ± 32 thousand years
ago for a tooth from the Irhoud 3 hominin mandible. These ages are
also consistent with the faunal and microfaunal9 assemblages and
almost double the previous age estimates for the lower part of the
deposits
10,11
. The north African site of Jebel Irhoud contains one of
the earliest directly dated Middle Stone Age assemblages, and its
associated human remains are the oldest reported for H. sapiens.
The emergence of our species and of the Middle Stone Age appear
to be close in time, and these data suggest a larger scale, potentially
pan-African, origin for both.
Jebel Irhoud (Irhoud), Morocco (Fig. 1 and Extended Data Fig. 1),
contains stratified archaeological deposits (Fig. 2) best known for
yielding abundant late Pleistocene hominin remains associated with a
Levallois-based Middle Stone Age stone-tool assemblage8,12 (Fig. 3 and
Extended Data Figs 4–6). Taxonomically these fossils have generally
been considered to be primitive forms of H. sapiens13, but they have
been dated to a relatively recent age11. However, the uncertain find
location of the key fossils has limited the accuracy of their age estimates.
1Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, Leipzig 04103, Germany. 2Institute of Ecology, Subject Area Landscape Change, Leuphana
University Lüneburg, Scharnhorststrasse 1, 21335 Lüneburg, Germany. 3Freiberg Instruments GmbH, Delfterstrasse 6, 09599 Freiberg, Germany. 4Research School of Earth Sciences, The
Australian National University, Canberra, Australian Capital Territory 0200, Australia. 5Australian Research Centre for Human Evolution, Environmental Futures Research Institute, Griffith University,
Nathan, Queensland 4111, Australia. 6Southern Cross GeoScience, Southern Cross University, Military Road, Lismore, New South Wales 2480, Australia. 7Department of Anthropology, University
of California, Davis, One Shields Avenue, Davis California, USA. 8Institut National des Sciences de L’Archéologie et du Patrimoine, Ministère de la Culture et de la Communication, Hay Riad,
Madinat Al Ifrane, Angle rues 5 et 7, BP 6828, Rabat, Morocco. 9Paléotime, 6173 Rue Jean Séraphin Achard Picard, 38250 Villard-de-Lans, France. 10Archéologie des Sociétés Méditerranéennes,
(ASM, UMR 5140 CNRS), Université Paul-Valéry, Montpellier 3, MCC, Route de Mende, 34199 Montpellier cedex 5, France. 11De la Préhistoire à l’Actuel : Culture, Environnement, Anthropologie
(PACEA, UMR 5199 CNRS), Université de Bordeaux, MCC, Bâtiment B18, Allée Geoffroy Saint-Hilaire, CS 50023, 33615, Pessac, France. 12Centre de Recherches sur la Paléobiodiversité et les
Paléoenvironnements (CR2P, UMR 7207 CNRS), Sorbonne Universités, MNHN, UPMC, CP 38, 8 Rue Buffon, 75231 Paris cedex 05 France. 13Chaire Internationale de Paléoanthropologie, Collège
de France, Paris, France.
c
a
Jebel Irhoud
Marrakesh
Casablanca
Figure 1 | Excavation site and hominin fossils.
a,South view of the site with the inset showing the
location of Irhoud in northwest Africa. The remaining
deposits are located in what was a tunnel-like karstic
feature dipping to the east that was later fully exposed
(at least in part to the left) by quarrying-related
activities. The remaining insitu sediments are to the
right of the blue tarp. The red circle indicates where
the hominin remains shown in b and c were found
(photo taken after the hominins were removed; photo
is a composite image with some distortion noticeable
left and right). b,View showing the partial skull
(Irhoud 1678/Irhoud 10) in the centre foreground
(white arrow) and the femur (Irhoud 2252/Irhoud 13)
in the centre background (yellow arrow). The back
centre and left is the cliff face. The archaeological
material is resting on a large boulder (the 10-cm
scale bar is at the contact). c, Plan view of b, but
after additional excavation. The partial skull (white
arrow) and femur (yellow arrow) are still present.
The mandible (Irhoud 4765/Irhoud 11) is wrapped
around the upper corner of the pyramid-shaped
rock. A portion of the right tooth row is clearly visible
(red arrow). A smaller portion of the left tooth row is
also visible. The photograph was taken after the area
(including the skull and femur) had been covered
with glue before moulding (see Supplementary
Information). Some glue has also been applied to the
mandible to stabilize it before removal.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
294 | NATURE | VOL 546 | 8 JUNE 2017
Electron spin resonance (ESR) dating of three mammal teeth from the
deposit immediately overlying a partial humerus, the only hominin
fossil at the time with a precisely documented find location, has resulted
in an age range of 190 to 90 thousand years (ka)11, while a coupled
U-series(US)/ESR
14
age for a human tooth fragment of 160 ± 16kyr
has been suggested
10
. All these results were, however, based on single
estimates of the external γ -doses obtained from sediment samples of
uncertain stratigraphic context.
New excavations were initiated in 2004 on the complete, intact
section remaining from the late 1960s excavations12 (Fig. 1b). This sec-
tion rests against the cave wall and atop large blocks that are likely to
represent a roof fall. On the basis of the limited documentation that is
available of the geomorphology before the cave was opened by recent
blasting, the remaining deposits appear to have been located within
the cave. The deposits are poorly stratified and poorly sorted, contain
occasional gravel lenses, and were formed mainly through processes
of debris flow and run-off, including material from the exterior. There
were no major post-depositional disturbances which could have caused
mixing or important disturbances for the dosimetry (Extended Data
Figs 2, 3 and Supplementary Information). Within this deposit, 7 layers
are distinguished and well-correlated with the 22 layers that have
been previously reported (Fig. 2). Layers 1–3, representing the upper
approximately 75 cm of the remaining deposits, contain very little
archaeological material. Layers 4–6 form a thick deposit (175 cm) of
relatively undifferentiated, unsorted, compacted sediments with vary-
ingly sized clasts and include some archaeological material. Layer 7 is
similar, but contains pockets of cemented sediments and the highest
density of archaeo logical finds. We noted in layer 7, as well as at the
contact between layers 5 and 6 in the unexcavated profile, several spa-
tially constrained centimetre-thick accumulations consisting at least
partially of charcoal (see Supplementary Information) that are tenta-
tively interpreted as features related to combustion. These features are
consistent with limited post-depositional disturbance and, along with
the occurrence of heated lithics throughout the sequence, they attest
to the presence of fire at Jebel Irhoud.
The stone tools and most of the faunal remains are concentrated in
the lower portion of the deposit, especially layer 7. Additionally, except
for one tooth (Irhoud-1653), all the newly discovered hominins
8
come
from layer 7 (Fig. 2). Most were excavated from a wedge of layer 7
sediment extending east from the main sediment block along the
bedrock face (Fig. 1) within an area of approximately 40 × 40 cm.
Among the previously discovered fossils, the Irhoud 4 humerus12 and
Irhoud 5 coxal15 can also be securely correlated to layer 7 (Fig. 2).
The fauna are dominated by remains of gazelle (Gazella sp.), but
Equidae (zebras) and Alcelaphini (wildebeest and hartebeest) remains
are also present (Supplementary Information). The faunal assemblage
preserves a diversity of carnivores; leopards (Panthera pardus) are most
common, but there are also remains of lions (Panthera leo) and smaller
cats. While carnivore remains and hyena coprolites are consistently
found in low abundance throughout the sequence, there is only a sin-
gle carnivore-chewed bone (a gazelle rib from layer 6). A primarily
anthropogenic origin for the fauna is confirmed by the majority of
long bones exhibiting green or fresh breaks (at least 61% in layer 4,
69% in layer 5, 54% in layer 6 and 60% in layer 7), by probable stone-
tool cut-marks on gazelle bones from layers 4 (one rib), 6 (one rib)
and 7 (one distal humerus and one fragment of a long bone), by a few
100.00
99.00
98.00
97.00
96.00
95.00
94.00 1002 10061003 10071004 10081005 1009
z
y
1
2
3
4
5
6
7
Backdirt
B
5
6
7
Cliff
Face
Cliff
Face
Unexcavated
Block
Backdirt/B
Empty
Humerus
18 17
14
12
Coprolites
10
9
8
11
6
7
5
4
3
2
1
19
20
21
22
I4
I5
Cliff
Face
Coxal
2670
2565
2561
21782252
4765
3748
2196 3747
2401
2381
2383
1653
Figure 2 | South view of the profile (in m) showing the main
stratigraphic units of our excavations correlated with the previous
stratigraphic profile. The layer designations on the right (1–22) are based
on the previous publication (see ref. 12). The designations in the centre
and on the left are our designations (numbers and letters in the coloured
layers). A strict correspondence between the two is not possible; however,
it is clear that the hominins recovered previously12 (the coxal and humerus
noted in red) correspond to our layer 7. The correlation of the two
stratigraphic profiles is based on field observations including finding nails
from the previous excavations12. Layer B is stratigraphically unconnected
to the rest of the sequence. The viewing angle of this section roughly
corresponds to the view in Extended Data Fig. 1. The clearly visible (in the
preserved profile) continuation of the layers is indicated as dotted lines for
the as yet unexcavated sediment. Our fossil finds are indicated with red
dots. The location of the coxal is estimated based on descriptions in ref.
15. The location of the humerus comes from ref. 12. Heated flints used for
thermoluminescence dating are indicated with yellow dots. Stone tools are
indicated with smaller grey dots.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
8 JUNE 2017 | VOL 546 | NATURE | 295
percussion notches consistent with marrow extraction, and by the rel-
atively high abundance of burnt bones (5% in layer 4, 25% in layer 5,
19% in layer 6 and 24% in layer 7). An analysis of the rodent remains
from the sequence suggests a maximum age of marine isotope stage
(MIS) 10 (374–337kyr)16, with strong indications that the sequence is
not much younger than this9.
The lithics of these new Middle Stone Age assemblages are con-
sistent with previous descriptions of the Jebel Irhoud material12
and are dominated by Levallois technology with a high proportion
of retouched tools, especially pointed forms (Fig. 3, Extended Data
Figs 4–6, Extended Data Table 1 and Supplementary Information). No
characteristic Acheulean or Aterian elements are present. There is little
evidence of post-depositional alteration and taphonomic bias in the
artefact assemblage: 56% of the lithics show no edge damage and 76%
of the blanks are complete. There is an underrepresentation of lithic
material from the screens (5–25 mm), which could indicate selective
water transport of fine materials. Given the low frequency of cores and
of shatter (see Extended Data Table 1), it is more likely indicative of a
lithic production system emphasizing blank and/or tool importation
with limited on-site production using local raw materials.
Although the excavated artefact assemblage is relatively small
(n = 320 artefacts larger than 2.5 cm), the high percentage of visibly
(37%) heated flint artefacts allows thermoluminescence dating of the
archaeological assemblage and associated human remains. External
γ -dose-rates for each layer were measured
17
in situ before excavation
with 47α -Al
2
O
3
:C dosimeters. The γ -dose-rates within each layer have
large dispersions of 8 to 14% (1 standard deviation (σ)) (Extended
Data Table 2), demonstrating the inhomogeneity of the sediments.
Therefore, layer γ -dose -rate averages plus the dating of several objects
per layer are required for dosimetric dating in order to obtain reliable
results18,19. This approach is possible with the new excavations, where
the stratigraphic provenience of the samples is known, unlike previ-
ous dosimetric dating attempts
10,11
with ESR. High-purity germanium
(HpGe) γ -ray spectrometry (Extended Data Table 2) shows that the U
and Th radioactive decay chain activities are consistent with present day
secular equilibrium, which leads us to assume that the external γ-dose-
rates have been constant over the entire burial time. The palaeodoses
were determined with a multiple aliquot additive regeneration slide
approach20–22 (Extended Data Fig. 7).
A total of 14 ages, as determined by thermoluminescence dating,
were obtained (Table 1 and Extended Data Table 2). The apparent ages
range from 240 ± 35ka to 378 ± 30ka (Table 1) for the entire strati-
graphy with dispersions of age results for the individual layers similar to
those observed for the γ -dose-rates. Such a wide age range is typical18,23
when individual artefact ages are based on layer averaged γ -dose-rates
in such heterogeneous or lumpy19 sediments. We are assuming that
the ages and associated individual dose-rates exhibit a quasi-random
dispersion around the mean γ -dose-rate18,19,23. Sample sizes are
sufficient
23
for calculating weighted average ages for layers 6 and 7 and
yield ages of 302 ± 32ka and 315 ± 34ka, respectively.
The published ESR age of the Irhoud 3 fossil10 was based on an
external dose-rate measured on a small sediment sample of imprecise
stratigraphic context. Assuming an origin from a layer-7-like deposit,
it can be recalculated using the in situ γ -dosimetry that is now avail-
able (Extended Data Table 2) and with additional recent insights into
the ESR dose estimation of solid enamel fragments
24
. The combined
US/ESR dating system14 assumes a continuous U-uptake based on
a single-parameter (p value) diffusion equation. Using this system,
US/ESR analysis results in an age of 281 + 37/− 29ka with a p value
of − 0.27. The closed system US/ESR (CSUS/ESR)
25
, which assumes
a fast U-uptake at the time before present corresponding to a closed
U system, yields an age estimate of 286 ± 32ka. The age range of both
ab
c
defg
hij
k
l
m
n
o
l1 cm
Figure 3 | Flint artefacts from layers 6 and 7. a–e, Convergent scrapers
(layer 6). f, g, Levallois flakes (layer 6). h, Transverse scraper (layer 6).
i,Limace (layer 6). j, Déjeté scraper (layer 7). k, Levallois flake (layer 7).
l,Convergent scraper (layer 7). m, Convergent scraper (layer 7). n, Unifacial
point (layer 7). o, Levallois flake (layer 7). Outlines represent profile views.
Table 1 | Individual thermoluminescence and thermoluminescence
context ages (1σ) of heated flint artefacts with percentage
contribution of the stable internal and external γ-dose-rates to the
total dose-rate (stable cosmic dose contribution31 is not given)
EVA-LUM
(Lab.
number) Layer D
. -external
D
. -g-
external Age
Statistical
uncertainty
Total
uncertainty
(% total D
. )(% total D
. )(ka) (ka) (ka)
07/15 421 72 274 20 36
07/16 425 69 374 35 52
07/17 555 41 292 18 27
07/18 522 72 309 12 35
07/19 626 68 240 24 35
07/21 637 58 290 19 32
07/25 630 64 322 13 34
08/03 622 72 307 26 42
Layer 6 weighted average context age 302 16 32
07/28*7A 42 53 378 20 30
08/05 7A 25 69 295 932
08/06 7A 20 74 310 15 38
08/07 7A 32 63 328 10 33
08/08 7A 22 72 332 16 41
08/12 7B 36 59 329 31 44
Layer 7 weighted average context age 315 11 34
*Weighted average of 2 subsamples.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
296 | NATURE | VOL 546 | 8 JUNE 2017
models, around 350–220ka at 2σ, provides an error envelope that
encompasses most possible scenarios of how the uranium diffused
into the dental tissue.
The thermoluminescence data suggest that the Middle Stone Age
industry from layer 7 has a 95% probability (2σ) of dating to 383–247 ka,
an age that is also in agreement with the fauna and microfauna9 of the
site. The stratigraphy shows that the dated artefacts and the hominin
remains originate from the same geological layer, and thus the age range
obtained for the heated flints can be used to estimate the age of the
hominin fossils. Additionally, the direct US/ESR date for the Irhoud
3 specimen is consistent with the thermoluminescence average ages
for layers 6 and 7.
Our ages for Jebel Irhoud overlap with those reported for the
Florisbad partial cranium
26
, suggesting that the hominins from Irhoud
8
and Florisbad represent the earliest known representatives of the
H. sapiens clade. The ages for the heated artefacts from layer 7 are
probably older than the sediments containing a Levallois-based Middle
Stone Age assemblage in the Maghreb, which were dated using optically
stimulated luminescence (OSL) to 254 ± 17ka (ref. 27). In east Africa,
the Middle Stone Age at ETH-72-8B (Gademotta, Ethiopia) occurs
before 275 ± 6 ka (ref. 6) and in the Kapthurin Formation (Baringo,
Kenya) before 284 ± 12 ka (ref. 5) (recalculated to 282 ± 20 ka)6 based
on 40Ar/39Ar-dating of overlying tuffs. In southern Africa, a rede-
posited Middle Stone Age assemblage at Kathu Pan layer 3 dates to
before 291 ± 45 ka (ref. 7) based on OSL ages. Therefore by approxi-
mately 300 ka or soon after MIS 9 (337–300 ka)
16
, Levallois technology
was distributed across a large part of Africa and Eurasia28 and was
in Africa associated with the earliest dated occurrences of H. sapiens.
The Saharan desert was greatly reduced during a series of Middle
Pleistocene ‘green Sahara’ episodes, with an especially marked but short
period around 330 ka (ref. 29). This would have allowed ecological
continuity between north Africa and sub-Saharan Africa. Biological
continuity between east and northwest Africa is also supported by
strong faunal similarities, especially for the Middle Pleistocene,
suggesting at least frequent communication between these regions30
(and see Supplementary Information). Therefore, whether the Jebel
Irhoud data suggest an even earlier origin for the Middle Stone Age that
was directly associated with the emergence of H. sapiens and followed
by a relatively rapid dispersal or whether there were multiple, regionally
specific, but related origins
28
is as yet unclear. Minimally, these behav-
ioural data, along with the associated fossil evidence, suggest a complex
pan-African process before or around 300 ka, a period for which we
still have relatively few data points, and we caution against favouring
one region over another in constructing models to account for these
changes in human behaviour and biology.
Online Content Methods, along with any additional Extended Data display items and
Source Data, are available in the online version of the paper; references unique to
these sections appear only in the online paper.
Received 4 May 2016; accepted 5 April 2017.
1. McDougall, I., Brown, F. H. & Fleagle, J. G. Sapropels and the age of hominins
Omo I and II, Kibish, Ethiopia. J. Hum. Evol. 55, 409–420 (2008).
2. Brown, F. H., McDougall, I. & Fleagle, J. G. Correlation of the KHS Tu of the
Kibish Formation to volcanic ash layers at other sites, and the age of early
Homo sapiens (Omo I and Omo II). J. Hum. Evol. 63, 577–585 (2012).
3. Scheinfeldt, L. B., Soi, S. & Tishko, S. A. Colloquium paper: working
toward a synthesis of archaeological, linguistic, and genetic data for
inferring African population history. Proc. Natl Acad. Sci. USA 107,
8931–8938 (2010).
4. Gronau, I., Hubisz, M. J., Gulko, B., Danko, C. G. & Siepel, A. Bayesian inference
of ancient human demography from individual genome sequences. Nat. Genet.
43, 1031–1034 (2011).
5. Deino, A. L. & McBrearty, S. 40Ar/39Ar dating of the Kapthurin Formation,
Baringo, Kenya. J. Hum. Evol. 42, 185–210 (2002).
6. Sahle, Y., Morgan, L. E., Braun, D. R., Atnafu, B. & Hutchings, W. K. Chronological
and behavioral contexts of the earliest Middle Stone Age in the Gademotta
Formation, Main Ethiopian Rift. Quat. Int. 331, 6–19 (2014).
7. Porat, N. et al. New radiometric ages for the Fauresmith industry from Kathu
Pan, southern Africa: implications for the Earlier to Middle Stone Age
transition. J. Archaeol. Sci. 37, 269–283 (2010).
8. Hublin, J.-J. et al. New fossils from Jebel Irhoud, Morocco and the pan-African
origin of Homo sapiens. Nature http://dx.doi.org/10.1038/nature22336 (2017).
9. Geraads, D. et al. The rodents from the late Middle Pleistocene hominid-
bearing site of J’bel Irhoud, Morocco, and their chronological and
paleoenvironmental implications. Quat. Res. 80, 552–561 (2013).
10. Smith, T. M. et al. Earliest evidence of modern human life history in north
African early Homo sapiens. Proc. Natl Acad. Sci. USA 104, 6128–6133 (2007).
11. Grün, R. & Stringer, C. B. Electron spin resonance dating and the evolution of
modern humans. Archaeometry 33, 153–199 (1991).
12. Hublin, J.-J., Tillier, A. M. & Tixier, J. L’humérus d’enfant moustérien (Homo 4)
du Djebel Irhoud (Maroc) dans son contexte archéologique. Bull. Mem. Soc.
Anthropol. Paris 4, 115–141 (1987).
13. Hublin, J.-J. Recent human evolution in northwestern Africa. Phil. Trans. R. Soc.
Lond. B 337, 185–191 (1992).
14. Grün, R., Schwarcz, H. P. & Chadam, J. ESR dating of tooth enamel: Coupled
correction for U-uptake and U-series disequilibrium. Int. J. Rad. Appl. Instrum. D
14, 237–241 (1988).
15. Tixier, J., Brugal, J., Tillier, A. M., Bruzek, J. & Hublin, J. Irhoud 5, un fragment
d’os coxal non-adulte des niveaux moustériens marocains. Actes J. Natl.
Archéol. Patrim. Maroc 149–153 (2001).
16. Lisiecki, L. E. & Raymo, M. E. A Pliocene–Pleistocene stack of 57 globally
distributed benthic δ
18O records. Paleoceanography 20, PA1003 (2005).
17. Richter, D., Dombrowski, H., Neumaier, S., Guibert, P. & Zink, A. C.
Environmental gamma dosimetry with OSL of α -Al2O3:C for in situ sediment
measurements. Radiat. Prot. Dosimetry 141, 27–35 (2010).
18. Richter, D. & Krbetschek, M. Luminescence dating of the Lower Palaeolithic
occupation at Schöningen. J. Hum. Evol. 89, 46–56 (2015).
19. Schwarcz, H. P. Current challenges to ESR dating. Quat. Sci. Rev. 13, 601–605
(1994).
20. Mercier, N., Valladas, H. & Valladas, G. Observations on palaeodose
determination with burnt ints. Anc. TL 10, 28–32 (1992).
21. Mercier, N. Flint palaeodose determination at the onset of saturation. Int. J. Rad.
Appl. Instrum. D 18, 77–79 (1991).
22. Valladas, G. & Gillot, P. Y. Dating of the Olby lava ow using heated quartz
pebbles: some problems. PACT 2, 141–150 (1978).
23. Richter, D. Advantages and limitations of thermoluminescence dating of
heated int from Paleolithic sites. Geoarchaeology 22, 671–683 (2007).
24. Joannes-Boyau, R. & Grün, R. A comprehensive model for CO2- radicals in
fossil tooth enamel: implications for ESR dating. Quat. Geochronol. 6, 82–97
(2011).
25. Grün, R. An alternative model for open system U-series/ESR age calculations:
(closed system U-series)-ESR, CSUS-ESR. Anc. TL 18, 1–4 (2000).
26. Grün, R. et al. Direct dating of Florisbad hominid. Nature 382, 500–501 (1996).
27. Ramos, J. et al. The Benzú rockshelter: a Middle Palaeolithic site on the north
African coast. Quat. Sci. Rev. 27, 2210–2218 (2008).
28. Adler, D. S. et al. Early Levallois technology and the Lower to Middle Paleolithic
transition in the southern Caucasus. Science 345, 1609–1613 (2014).
29. Larrasoaña, J. in Modern Origins: A North African Perspective (eds Hublin, J.-J.
& McPherron, S. P.) 19–34 (Springer, 2012).
30. Geraads, D. Biogeographic relationships of Pliocene and Pleistocene
north-western African mammals. Quat. Int. 212, 159–168 (2010).
31. Prescott, J. R. & Hutton, J. T. Cosmic ray contributions to dose rates for
luminescence and ESR dating: large depths and long-term time variations.
Radiat. Meas. 23, 497–500 (1994).
Supplementary Information is available in the online version of the paper.
Acknowledgements The Jebel Irhoud project is jointly conducted and
supported by the Moroccan Institut National des Sciences de l’Archéologie
et du Patrimoine and the Department of Human Evolution of the Max Planck
Institute for Evolutionary Anthropology (MPI-EVA). We thank S. Albert (MPI-EVA)
for sample preparation and for measuring the flint samples, E. Pernicka (Curt-
Engelhorn-Zentrum Archäometrie, Mannheim) for neutron activation analysis
and D. Degering (Verein für Kernverfahrenstechnik und Analytik, Rossendorf)
for performing γ -ray spectrometry. The Max Planck Society funded the
fieldwork and the thermoluminescence analysis. V. Aldeias (MPI-EVA) excavated
the partial skull. B. Larmignat illustrated the stone artefacts. Philipp Gunz
commented on the manuscript, and Les Kinsley (RSES, ANU) assisted with laser
ablation measurements. Parts of the US/ESR research were funded by ARC
discovery grants (DP0664144 to R.G.) and (DP140100919 to R.J.-B.)
Author Contributions Thermoluminescence dating was carried out by D.R.;
ESR dating was done by R.G. and R.J.-B.; zooarchaeology and taphonomy
was carried out by F.A., T.E.S. and D.G.; lithics analysis was done by S.P.M.; raw
material analysis was carried out by P.F. and J.P.R.; geology was done by J.-P.R.,
P.F. and M.R.; and A.B.-N., J.-J.H. and S.P.M. excavated the site. D.R. and S.P.M.
wrote the paper with contributions from all authors.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial
interests. Readers are welcome to comment on the online version of the paper.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations. Correspondence and
requests for materials should be addressed to S.P.M. (mcpherron@eva.mpg.de).
Reviewer Information Nature thanks R. G. Klein, R. G. Roberts and the other
anonymous reviewer(s) for their contribution to the peer review of this work.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
METHODS
Thermoluminescence. Thermoluminescence dating of heated flint artefacts is
based on the accumulation of metastable charges (palaeodose) in the crystal lattice
by ionizing radiation since the last heating of the rock32. The method provides an
age estimate of a prehistoric activity and therefore of human behaviour directly
23
.
An age is obtained by the ratio of the palaeodose, determined with thermolumi-
nescence, to the total effective dose-rate, with the assumption that the dose-rate
was constant over the entire burial time32.
Palaeodose determination. Because the natural luminescence signal of the Jebel
Irhoud samples is in the nonlinear part of the dose–response curve, the palaeodose
on the 90–160 μ m fraction of the crushed and chemically treated flint material
18,23
(after the removal of the outer 2 mm surface with a cooled low-speed saw) was
obtained by a multi-aliquot additive regeneration slide protocol18,20–22 (Extended
Data Fig. 7). Heated flint thermoluminescence dating results were obtained with
this protocol for many other ancient sites
33–35
, where there was good agreement
with other chronometric dates
33,36–38
. The thermoluminescence data are described
by exponential functions and shifted along the dose axis20–22 to obtain the
palaeodoses, similar to slide approaches for other materials39–41. Between 4 and
12 aliquots were used for each of the 3–5 dose points for each dose–response
curve, where the grains used to construct the regeneration dose–response curve
were heated to 360 °C for 90 min in air. This procedure is assumed to induce the
least changes in sensitivity. However, only samples exhibiting unity of the thermo-
luminescence-signal ratios of the two dose–response curves after sliding (Extended
Data Fig. 7) were accepted
20
. Comparable results, with many more samples passing
the quality criteria, are obtained when the regeneration dose–response curves are
scaled
40,41
, an approach that has provided congruent results in comparison to other
luminescence dating38 analyses and/or when thermoluminescence glow peaks
are aligned
42
. This larger sample set includes artefacts from the eastern part of the
section, close to the hominins.
Thermoluminescence was measured with an EMI 9236QA photomultiplier
with detection restricted to the UV-blue-wavelength band by Schott BG25 and
WG5 filters at a heating rate of 5 Kmin−1 to 450 °C on a Risø DA-20 system.
Irradiations were performed with external calibrated sources (β with
90
Y/
90
Sr at
0.26 ± 0.01Gys
−1
and α with
214
Am at 0.178 ± 0.011 μ m
−2
min
−1
) and samples
were stored before measurements (1week at 70 °C or 4weeks at room temperature).
The α sensitivity
43
was determined by comparing the regenerated luminescence
response of 4–11-μ m fine-grained material (heated in air to 500 °C for 30 min) to
single doses of α and β irradiation. The doses were chosen to produce thermolumi-
nescence signals at similar levels, while staying well within a linear dose–response.
Whether the thermoluminescence signal was sufficiently zeroed was analysed for
the 45 flint artefacts that showed macroscopic traces of heating with the heating
plateau test32 (inset in Extended Data Fig. 7). Of these, 14 artefacts eventually
passed the described tests (Table 1 and Extended Data Table 2). The integration
range of the luminescence signal for palaeodose determination was defined by the
heating plateau (Extended Data Table 2).
Dosimetry. The calculations of internal dose-rates44 are based on neutron acti-
vation analysis results (Extended Data Table 2) for U, Th and K of 200 mg sample
material of less than 160 μ m from the material after sawing that was obtained before
the chemical treatment and further sieving.
Dosimetric dating methods are based on the assumption of stability of the
dose-rates over the burial time and that radioactive elements are homo genously
distributed throughout the sample. However, either can be modelled as well45,46.
Heterogeneous internal dose-rates from inhomogeneities in radioactive element
distributions have been shown to be limited to macroscopically visible veins, inclu-
sions and different mineral phases47. These are not observed here and any altered
parts were removed
23
. At Jebel Irhoud, the internal dose-rate provides variable
accounts of the total dose-rate ranging between 20% and 55% (Table 1).
The stability over time of the external γ -dose-rate from the surrounding sedi-
ment was verified by HpGe γ -ray spectrometry (based on an estimated matrix of
95% SiO
2
and 5% BaSO
4
to accommodate for the ubiquitous presence of barite)
on sediment samples (grain fraction less than around 40 mm in particle diameter).
There is no indication of disequilibria for either the U or the Th decay chain
(Extended Data Table 2b) in the entire sediment column. While these analyses do
not reject the possibility of very ancient occurrences of disequilibria, we interpret
the lack of such and the absence of any trends in ratios of analysed isotopes through
the sediment column as indications that the decay chains have always been in
equilibrium. The absence of significant changes in the external γ -dose-rate is also
indicated by the age results of samples with the highest internal dose-rates (Table 1
and Extended Data Table 2), which can be considered to be more reliable23. They
provide results in the upper as well as in the lower range of the age dispersion,
which would not be expected for samples of the same age suffering from fluctuating
γ -dose-rates. We assume, therefore, that the external γ -dose-rate was stable over
the entire burial period.
HpGe γ -ray spectrometry allows analysis of only the small sediment particles,
but Jebel Irhoud contains larger rocks and especially large boulders in the lower
sequence. HpGe γ -ray spectrometry on sediment samples in the laboratory is,
therefore, not representative of the spatial heterogeneity and the site’s dosimetry
has to be considered as ‘lumpy’19.
The external γ -doses were measured with α -Al
2
O
3
:C dosimeters. These were
left buried 30 cm into all of the exposed profiles for one year with 60 cm spacing
between dosimeters to measure all available layers (Extended Data Table 2a). The
dosimeters record the cosmic and γ -dose-rates at their individual positions, the
latter of which are assumed to be similar to those of the excavated flint samples
coming from different positions within a given layer. The γ -dose-rates are obtained
by comparing each crystal’s OSL response (an average of five for each dosimeter)
of the natural one-year exposure to an irradiation by a calibrated 137Cs-source.
The high accuracy and precision of this approach has recently been shown by
comparison with HpGe γ -ray spectrometry
17
. The present day cosmic dose differs
from that of the past due to the recent removal of overlying rock and sediment,
and therefore the measured values have to be adjusted. The nearly vertical, 4 m
high, rock wall at a maximum distance of only around 1.5 m from the dosimeters
provides approximately 50% shielding on one side in addition to the sediment
that covers them. Thus cosmic doses were calculated separately for the full 4π
geometries of 4 m rock overburden, as well as for the present day sediment coverage,
and then adjusted by 50% each and summed.
Analytical uncertainties for dosimeter measurements are typically between 2%
and 5%, but for age calculation a conservative uncertainty of 15% was used for the
averaged external γ -dose-rates, which is larger than the dispersion of the dosimeter
readings. The sublayers of layer 7 (Figs 1, 2) are defined in part by the slight syn- or
post-depositional process of localized weak brecciation, which probably has led to
differences in the external γ -dose-rate related to this geochemical process of fixa-
tion of the sediment of layer 7C. The γ -dosimetry from each sublayer (as defined
by consolidated appearance and not necessarily in superposition) was used for age
calculation. On the basis of archaeological and sedimentological interpretation,
which indicate a common age for all material from layer 7, but not necessarily the
same dosimetry, the average ages are reported as a single context age for layer 7.
Excavation and mining activities removed the roof and most of the sediments,
exposing the present day sediments for several decades. Prior to the 1960s, the
site was an almost completely filled and sealed cave, and it is likely that sediments
were moister in this context. Therefore, present day γ -dose-rates are likely to be
overestimated and thus the resulting ages slightly too young. Because there is little
data to estimate the moisture for the entire burial history, we conservatively use the
present day moisture (Extended Data Table 2), which is included in the dosimeter
readings, and thus we consider our thermoluminescence ages to be minimum
estimates. The influence of moisture content on the calculated ages is, however,
not large. For example, an increase in moisture by 20% would result, on average, in
an increase in the ages by only about 10%. As expected, the γ -dose-rates deduced
from in situ dosimeters are between 8% and 30% lower compared to HpGe γ -ray
spectrometry data that was corrected for the moisture as measured from sealed
sediment samples (Extended Data Table 2c).
To estimate the past cosmic dose -ate we used a section drawing48, which pro-
vides a scale and section drawings
49,50
, where the positions of the human skulls
from the previous excavations are marked, as well as historical photos showing
that the recent excavation is located to the south of the previous excavations. Our
link to the Tixier excavation
12
was supported in part by a nail still present in one
of the profiles. In order to calculate the cosmic dose-rates (Extended Data Table 2)
the overburden of sediment and rock (assumed densities of 1.9 and 2.4 gcm
−3
,
respectively) were estimated by connecting the surface of the rock coverage to the
north and to the south (rock that has since been removed by human activities). This
results in a minimum rock thickness of 6 m for the roof and 3–4 m of sediment in
the cave, where approximately 3 m of sediment covered the recent hominin finds.
Estimation of the age of layers 6 and 7. Analysis of the thermoluminescence ages
of layers 4 and 5 is hampered by the low number of samples. For layers 6 and 7 the
relative standard deviations of the ages (9% and 6%, respectively) are similar to the
relative standard deviations of dosimeter readings (8% and 12%, respectively), and
the same applies to the relative average deviations from the means (for layers 6 and
7 these are 9% and 5% for ages; 6% and 11% for dosimeter results, respectively) as
a measure of the variability of the data.
For dosimetric dating of objects (thermoluminescence of heated flint artefacts
or ESR on teeth) from heterogeneous sediments the external γ -dose-rate for each
individual sample cannot be known per se, because the sample is removed from
its context during excavation and the measurement of the 4π external γ -dose-rate
is not possible at each sample’s original location. γ -dose-rates, therefore, have to be
either reconstructed
51–53
or averaged results must be used
18,19,23,34,35,54
. For hetero-
geneous sediments the average value of dosimeter measurements is unlikely to
be correct for any of the samples18. However, for the above reasons, the nearest
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
neighbouring dosimeter will not necessarily provide a better estimate of the exter-
nal γ -dose-rate either, because it might not have been exposed in the exact same
geometry. Nevertheless, provided that the samples were heated at roughly the same
time, an age estimate close to the true mean can be obtained based on averaged
external γ -dosimetry. It is assumed that dosimeter positions are similar to the
ones occupied by the samples in the sediment and, therefore, either geometry can
be considered as random, because no selection is possible for either. This means
that individual dose-rates and ages should be randomly dispersed about the mean
γ -dose-rate
18,19,23
. As a consequence, this lack of knowledge of the true external
γ -dose-rate leads to a large spread in individual apparent thermoluminescence age
data (Table 1), but the mean age from several samples then provides a meaningful
age estimate of the last heating18,19,23.
Basic statistical analysis is only possible for layers 6 and 7. Shapiro–Wilk tests
(software package Origin 8.5) showed that the two datasets (W values of 0.92 and
0.91 for layers 6 and 7, respectively) are samples from normal distributions at
the 0.05 probability level (W threshold values of 0.52 and 0.41, respectively). We
treat each of the major geologically defined stratigraphic layers as analytical units,
therefore providing the minimum resolution, and we calculate weighted average
ages for each. The last heating of the sampled flint artefacts is estimated to have
occurred 302 ± 32 ka for layer 6 and 315 ± 34 ka for layer 7 at the 1σ probability
level. These weighted average ages were calculated with the individual statistical
uncertainties, deriving mainly from the luminescence measurements, as weights.
The uncertainties considered as mainly systematic, for example, source calibration,
were subsequently added and the uncertainty estimate of the weighted average
is derived from both uncertainties summed in quadrature. The two population
means for layers 6 and 7 are not different at a level of 0.05 with a two sample t-test
(P values of 0.09 and 0.12 for equal and unequal assumed variance, respectively,
from software package Origin 8.5). The thermoluminescence determinations
provide nominal age ranges at 95% probability of 366–238kyr for layer 6 and
383–247kyr for layer 7, which contains the hominin remains. Analysing the same
data with an extrapolation approach
32
for both dose–response curves provides
very similar palaeodoses, resulting in weighted mean ages which are different by
only a few per cent.
It can be concluded that the sequence of sediments containing the archaeological
and palaeoanthropological material is about 300kyr old.
ESR. A fragment of a single tooth from the Irhoud 3 mandible was dated by cou-
pled US/ESR to 160 ± 16 ka (ref. 10). This method has been shown to provide
comparable results when critically appraised against other dating results and veri-
fied independent chronologies37,55–57. The dosimetry was based on radionuclide
analysis of a sediment sample argued to be from the same brecciated layers from
which the fossils were thought to derive. Here the age is recalculated due to recent
insights into the dose estimation of enamel fragments24 and the availability of more
detailed dosimetric data.
Dose determination. The ESR dating signal consists of several types of CO
2−
radi-
cals, two anisotropic, orientated (axial and orthorhombic) and one non-oriented
radical. When the enamel fragment is rotated between ESR measurements, the
intensities of the orientated CO
2−
radicals change, whereas the intensity of the
non-oriented radical remains constant
24
. The natural ESR signal of Irhoud con-
tains about 90% and 10% of anisotropic and non-oriented CO2− radicals, respec-
tively (Extended Data Fig. 8). By contrast, the laboratory-generated ESR intensity
contains 40 ± 2% non-oriented CO
2−
radicals. Unfortunately, some of these are
thermally unstable. Different enamel domains with significantly different dose–
responses (figure 6 in ref. 24), show that some of the laboratory induced, non-
oriented CO
2−
radicals are stable (around 10%) and around 35% are likely to con-
vert into stable anisotropic CO2− radicals over time.
The ESR signals of the enamel piece from Irhoud 3 were decomposed following
ref. 24. Extended Data Fig. 8 shows the assessment of the percentage of non-
oriented CO2− radicals in the overall signal intensity of the natural and irradiated
samples. The decomposition results have relatively large individual errors (around
5% for each value) but the data can be fitted with a single saturating function,
which then provides the amount of non-oriented CO
2−
radicals for each radiation
step. Fitting only the anisotropic CO
2−
radicals yields a dose value of 383 ± 8Gy.
However, because of the probable transfer of non-oriented to stable anisotropic
CO2− radicals, the dose is somewhat smaller. The samples from Holon24 and Irhoud
have similar ages and also similar amounts of non-oriented CO
2−
radicals in the
laboratory-generated ESR signal. Assuming that 45 ± 5% of the laboratory-generated
non-oriented CO2− radicals are either stable or are converted into anisotropic CO2−
radicals, a dose of 326 ± 16Gy is obtained (Extended Data Table 3), which is about
26% higher than previously reported10. The fragment was scanned with quadru-
pole laser ablation ICP-MS in order to obtain the spatial uranium distribution in
enamel and dentine
58
. Uranium concentrations average at 0.07 ± 0.04 p.p.m. and
3.76 ± 0.42 p.p.m. in enamel and dentine, respectively. The U-concentrations in
the enamel are too low for in situ U-series analysis. Using the multi-collector laser
ablation ICP-MS system
59,60
, the dentine was analysed for U-series isotopes and
yielded ratios of 1.5287 ± 0.0088 for
234
U/
238
U and 0.7186 ± 0.0114 for
230
Th/
234
U.
Dosimetry. We assume here that the mandible originates from sediments equi-
valent to layer 7 based on published accounts50,61 indicating that the hominin
remains, including the mandible, originated from the lower part of the section.
Although the original position and thus precise sedimentological association is
unknown, assuming an origin from any of the other sediment layers would not
provide significantly different age results because of the similar dosimetry of all lay-
ers (Extended Data Table 2). Dosimetry based on dosimeters is preferred because
the heterogeneity (lumpiness) of the sediment is accounted for, which was previ-
ously not the case10. The average γ -dose-rate measured in situ by the dosimeters
(Extended Data Table 2a) for layer 7 provides 807 ± 107μ Gya
−1
. To account for the
interaction of shielding and dosing of the mandible62, the external γ -dose-rate was
adjusted by 7%, resulting in an effective external γ -dose-rate of 751 ± 100μ Gya
−1
.
An additional cosmic dose-rate component of 70 ± 7 μ Gy a−1 has to be
considered. Total radionuclide concentrations from Hp-Ge γ –spectrometry of the
sediment yielded 2.37 ± 0.67 p.p.m.U; 6.77 ± 0.47 p.p.m.Th and 2.32 ± 0.22%K
(calculated from the average of layer 7 from Extended Data Table 2b). These data
are different from those previously used
10
, where the relative distribution of Th
and U was 1.6 ± 0.1 p.p.m. and 6.98 ± 0.1 p.p.m., respectively. This seems to point
to a mix up of the Th and U values in the previous study
10
. The measured values for
layer 7 were used for the water content of the sediments (Extended Data Table 2c),
and 5% water in the dentine is assumed. The enamel thickness varied along the
fragment and was on average 1,000 ± 200μ m.
Since no external layer was removed from the enamel fragment, the external
α -dose-rate has to be considered. The average nuclide concentrations from HpGe
γ -ray spectrometry (Extended Data Table 2b) of layer 7 combined with an assumed
0.13 ± 0.02 α efficiency63 and attenuation factors64 provide an average external
α -dose-rate of 6 ± 1μ Gya
−1
from U and Th (Extended Data Table 3). The dentine
generates a dose in the enamel of less than 0.5 Gy, which was not considered
further.
Sample size. No statistical methods were used to predetermine sample size.
Data availability. The authors declare that the data supporting the findings of this
study are available within the article and its Supplementary Information.
32. Aitken, M. J. Thermoluminescence Dating (Academic, 1985).
33. Mercier, N. & Valladas, H. Reassessment of TL age estimates of burnt ints
from the Paleolithic site of Tabun Cave, Israel. J. Hum. Evol. 45, 401–409
(2003).
34. Valladas, H. et al. Thermoluminescence dates for the Neanderthal burial site at
Kebara in Israel. Nature 330, 159–160 (1987).
35. Valladas, H. et al. Thermoluminescence dating of mousterian ‘Troto-Cro-
Magnon’ remains from Israel and the origin of modern man. Nature 331,
614–616 (1988).
36. Rink, W. J., Schwarcz, H. P., Ronen, A. & Tsatskin, A. Conrmation of a near
400ka age for the Yabrudian industry at Tabun Cave, Israel. J. Archaeol. Sci.
31, 15–20 (2004).
37. Mercier, N. et al. New datings of Amudian layers at Qesem Cave (Israel): results
of TL applied to burnt ints and ESR/U-series to teeth. J. Archaeol. Sci. 40,
3011–3020 (2013).
38. Frouin, M. et al. Chronology of the Middle Palaeolithic open-air site of Combe
Brune 2 (Dordogne, France): a multi luminescence dating approach.
J. Archaeol. Sci. 52, 524–534 (2014).
39. Prescott, J. R., Huntley, D. J. & Hutton, J. T. Estimation of equivalent dose in
thermoluminescence dating — the Australian slide method. Anc. TL 11, 1–5
(1993).
40. Sanzelle, S., Miallier, D., Pilleyre, T., Faïn, J. & Montret, M. A new slide technique
for regressing TL/ESR dose response curves—intercomparison with other
techniques. Radiat. Meas. 26, 631–638 (1996).
41. Guibert, P., Vartanian, E., Bechtel, F. & Schvoerer, M. Non linear approach
of TL response to dose: polynomial approximation. Anc. TL 14, 7–14
(1996).
42. Berger, G. W. Thermoluminescence dating studies of rapidly deposited silts
from south-central British Columbia. Can. J. Earth Sci. 22, 704–710 (1985).
43. Valladas, H. & Valladas, G. Eect de l’irradiation alpha sur des grain de quartz.
PACT 6, 171–179 (1982).
44. Adamiec, G. & Aitken, M. Dose-rate conversion factors: update. Anc. TL 16,
37–50 (1998).
45. Guibert, P., Lahaye, C. & Bechtel, F. The importance of U-series disequilibrium
of sediments in luminescence dating: a case study at the Roc de Marsal Cave
(Dordogne, France). Radiat. Meas. 44, 223–231 (2009).
46. Tribolo, C. et al. TL dating of burnt lithics from Blombos Cave (South Africa):
further evidence for the antiquity of modern human behaviour. Archaeometry
48, 341–357 (2006).
47. Schmidt, C., Rufer, D., Preusser, F., Krbetschek, M. & Hilgers, A. The assessment
of radionuclide distribution in silex by autoradiography in the context of dose
rate determination for thermoluminescence dating. Archaeometry 55,
407–422 (2013).
48. Ennouchi, E. Essai de datation du gisement de Jebel Irhoud (Maroc). Comptes
Rendus Somm. Séances Société. Geol. Fr. 10, 405–406 (1966).
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
49. Arambourg, C. Le gisement moustérien et l’homme du Jebel Irhoud.
Quaternaria 7, 1–7 (1965).
50. Ennouchi, E. Le site du Jebel Irhoud (Maroc). in Actas del V Congreso
Panafricano de Prehistoria y de Estudio del Cuaternario (Santa Cruz de Tenerife,
1963) 6, 53–60 (1966).
51. Guibert, P., Bechtel, F., Schvoerer, M., Müller, P. & Balescu, S. A new method for
gamma dose-rate estimation of heterogenous media in TL dating. Radiat. Meas.
29, 561–572 (1998).
52. Guérin, G. Modélisation et Simulations Numérique des Eets Dosimétriques dans
les Sediments Quaternaires: Application aux Méthodes de Datation par
Luminescence. PhD thesis, Univ. Bordeaux (2012).
53. Guérin, G. & Mercier, N. Field gamma spectrometry, Monte Carlo simulations
and potential of non-invasive measurements. Geochronometria 39, 40–47
(2012).
54. Mercier, N., Valladas, H., Valladas, G. & Reyss, J.-L. TL dates of burnt ints from
Jelinek’s excavations at Tabun and their implications. J. Archaeol. Sci. 22,
495–509 (1995).
55. Falguères, C. et al. Combined ESR/U-series chronology of Acheulian
hominid-bearing layers at Trinchera Galería site, Atapuerca, Spain. J. Hum. Evol.
65, 168–184 (2013).
56. Michel, V., Delanghe-Sabatier, D., Bard, E. & Ruiz, C. B. U-series, ESR and 14C
studies of the fossil remains from the Mousterian levels of Zafarraya Cave
(Spain): a revised chronology of Neandertal presence. Quat. Geochronol. 15,
20–33 (2013).
57. Wagner, G. A. et al. Radiometric dating of the type-site for Homo
heidelbergensis at Mauer, Germany. Proc. Natl Acad. Sci. USA 107, 19726–19730
(2010).
58. Eggins, S., Grün, R., Pike, A. W. G., Shelley, M. & Taylor, L. 238U, 232Th proling
and U-series isotope analysis of fossil teeth by laser ablation-ICPMS. Quat. Sci.
Rev. 22, 1373–1382 (2003).
59. Grün, R., Aubert, M., Joannes-Boyau, R. & Moncel, M.-H. High resolution
analysis of uranium and thorium concentration as well as U-series
isotope distributions in a Neanderthal tooth from Payre (Ardèche, France)
using laser ablation ICP-MS. Geochim. Cosmochim. Acta 72, 5278–5290
(2008).
60. Grün, R., Eggins, S., Kinsley, L., Moseley, H. & Sambridge, M. Laser ablation
U-series analysis of fossil bones and teeth. Palaeogeogr. Palaeoclimatol.
Palaeoecol. 416, 150–167 (2014).
61. Ennouchi, E. Le deuxième crane de l’homme d’Irhoud. Ann. Paleontol. LIV,
117–128 (1968).
62. Nathan, R. & Grün, R. Gamma dosing and shielding of a human tooth by a
mandible and skull cap: Monte Carlo simulations and implications for the
accuracy of ESR dating of tooth enamel. Anc. TL 21, 79–84 (2003).
63. Grün, R. & Katzenberger-Apel, O. An alpha irradiator for ESR dating. Anc. TL
12, 35–38 (1994).
64. Grün, R. Alpha dose attenuation in thin layers. Anc. TL 5, 6–8 (1987).
65. Valladas, H. Datation par la Thermoluminescence de Gisements Moustériens du
Sud de la France PhD Thesis, Univ. Paris VI (1985).
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Figure 1 | View south of the remaining sediments at
the start of excavations in 2004. The approximate area of the main fossil
concentration (not actually visible in this initial photograph taken before
our excavations) is circled in red and detailed in Fig. 1b, c. The stacked
rocks around the base of the sediments and ramping up to the sediments
on the left were placed there for protection of the remaining deposits. The
white tags mark dosimeter locations. The scale is correct for the section
with the tags.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
Extended Data Figure 2 | Non-polarized light photomicrographs
from thin-sections. a, Layer 4, thin-section 608M, showing the good
preservation of the sediment owing to overlying cave lithoclasts. b, Layer 4,
thin-section 608M, clasts are oriented with unit dip. c, Layer 7 upper part,
thin-section 712T, indicating a run-off deposit. d, Layer 7, thin-section
609T, bone micro-fragments in an isotropic fabric microfacies. e, Lower
part of layer 7, thin-section 716, with a high density of micro-charcoal, soil
aggregates, bone fragments and heated lithoclasts. f, Trampled surface in
layer 7, thin-section 712B (thin sections by M. El Graoui). Photos by M.R.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Figure 3 | Cross-polarized and plane-polarized
photomicrographs from thin-section of micromorphology sample
717 (layer 7). a, Scanned thin section. Squares with letters in a refer to the
areas in b–e, each area provided as plane- (PPL) and cross-polarized (XPL)
images. Scale bar, 0.5 mm. Bio indicates bioturbation and the numbers
refer to the sub-units as indicated by dotted lines. ST refers to structure.
b,Black coatings against a biogallery wall. c, Micro-bedded carbon
products preserved under a schisteous clast. d, Carbon aggregates that coat
the bottom of ST1. e, bed of carbon micro-particles in the filling of ST1.
Photos by M.R.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
Extended Data Figure 4 | Flint artefacts. a, b, e, Unifacial points (layer 6). c, d, Convergent scrapers (layer 6). f, Déjeté scraper (layer 6). g, h, Convergent
scrapers (layer 7). i, Unifacial point (layer 7). j, Levallois Flake (layer 7). k, m, Double scrapers (layer 7). l, Déjeté scraper (layer 7). n, Single scraper (layer 7).
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Figure 5 | Stone artefacts from layer 7. a, b, Quartz flakes. c, m, Flint Levallois flakes. d, i, Silicified limestone flakes. e, g, h, Flint flakes
with some edge damage. f, Flint flake. j, n, Silicified limestone flakes with some edge damage. k, l, Flint Levallois flakes with some edge damage.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
Extended Data Figure 6 | Stone artefacts from layer 7. a, c, Single scrapers. b, Double scraper with some edge damage. d, Notch on silicified limestone.
e, Single scraper with some edge damage on a Levallois flake. f, Convergent denticulate (Tayac Point). g, Double scraper. h, Déjeté scraper. i, Unifacial
point. All artefacts are flint unless noted otherwise.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Figure 7 | Dose–response curves of the exponentially
fitted thermoluminescence temperature integrals, where the
regeneration dose–response curves were shifted along the dose axis to
obtain the palaeodoses. The similarity (homothety) of the dose–response
curves is given by the ratios of the thermoluminescence integrals of the
additive and shifted regeneration dose–response curves at the additive
dose points. The inset depicts the glow curves and the heating plateau for
300–600Gy additive β -irradiations.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
Extended Data Figure 8 | Estimation and fitting of non-oriented CO2- radicals (ESR) .
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Table 1 | Lithic data
a, Counts by layer (from the recent excavations) for stone artefact classes. Retouched point corresponds to a Mousterian Point in European Middle Palaeolithic terminology. Platform akes include
complete and proximal akes. b, Platform types. c, Blank technology. Normal means that no particular technology could be identied.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter reSeArCH
Extended Data Table 2 | Dosimetric and thermoluminescence data
a, Layers, locations and γ -dose-rates for individual dosimeters and average context dose-rates. b, Total activities from HpGe γ -ray spectrometry (Bq kg−1 at 2σ) on dry samples of all sediment particles
smaller than 4 cm, based on an estimated 95% SiO2 and 5% BaSO4 composition. For the 238U-series the γ -lines from 234Th were used for 238U; for 226Ra from 214Pb and 214Bi; for 210Pb from 210Pb. For
the 232Th-series the estimates for 228Ra are based on 228Ac and for 228Th on 212Pb, 212Bi and 208Tl. c, Comparison of γ -dose-rates obtained with in situ α -Al2O3:C dosimeters and HpGe γ -ray spectrome-
try on dry sediment, with the latter corrected for measured moisture content. d, Thermoluminescence sample identiers, provenience and analytical results for heated int samples from Jebel Irhoud
(a, b, indicate independent subsamples from a single artefact, for which a weighted average age was calculated). The eective external γ -dose-rates account for the shape and weight of samples65. All
uncertainties at 1σ, with calculations following ref. 32.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

letter
reSeArCH
Extended Data Table 3 | ESR age calculation
For comparison with previously published ESR results, age calculations were also carried out for the parametric early and linear U-uptake models.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.