Emergence of Modern Human Behavior: Middle Stone Age Engravings from South Africa

Abstract

In the Eurasian Upper Paleolithic after about 35,000 years ago, abstract or depictional images provide evidence for cognitive abilities considered integral to modern human behavior. Here we report on two abstract representations engraved on pieces of red ochre recovered from the Middle Stone Age layers at Blombos Cave in South Africa. A mean date of 77,000 years was obtained for the layers containing the engraved ochres by thermoluminescence dating of burnt lithics, and the stratigraphic integrity was confirmed by an optically stimulated luminescence age of 70,000 years on an overlying dune. These engravings support the emergence of modern human behavior in Africa at least 35,000 years before the start of the Upper Paleolithic.

Full text

Emergence of Modern Human
Behavior: Middle Stone Age
Engravings from South Africa
Christopher S. Henshilwood,
1,2,3
* Francesco d’Errico,
4
Royden Yates,
1
Zenobia Jacobs,
5
Chantal Tribolo,
6
Geoff A. T. Duller,
5
Norbert Mercier,
6
Judith C. Sealy,
7
Helene Valladas,
6
Ian Watts,
1,7
Ann G. Wintle
5
In the Eurasian Upper Paleolithic after about 35,000 years ago, abstract or
depictional images provide evidence for cognitive abilities considered integral
to modern human behavior. Here we report on two abstract representations
engraved on pieces of red ochre recovered from the Middle Stone Age layers
at Blombos Cave in South Africa. A mean date of 77,000 years was obtained
for the layers containing the engraved ochres by thermoluminescence dating
of burnt lithics, and the stratigraphic integrity was confirmed by an optically
stimulated luminescence age of 70,000 years on an overlying dune. These
engravings support the emergence of modern human behavior in Africa at least
35,000 years before the start of the Upper Paleolithic.
Archaeological evidence associated with
modern cognitive abilities provides important
insights into when and where modern human
behavior emerged (1). Two models for the
origins of modern human behavior are cur-
rent: (i) a late and rapid appearance at ⬃ 40 to
50 thousand years ago (ka) associated with
the European Upper Paleolithic and the Later
Stone Age (LSA) of sub-Saharan Africa (2,
3) or (ii) an earlier and more gradual evolu-
tion rooted in the African Middle Stone Age
(MSA; ⬃250 to 40 ka) (4, 5). Evidence for
modern behavior before 40 ka is relatively
rare and often ambiguous (2, 6). However, in
sub-Saharan Africa, archaeological evidence
for changes in technology, economy, and so-
cial organization and the emergence of sym-
bolism in the MSA may support the second
model (4, 5, 7–9). Examples of these changes
include standardized formal lithic tools (5, 8,
10), shaped bone implements (5, 7, 9, 11),
innovative subsistence strategies such as fish-
ing and shellfishing (10 –12), and the system-
atic use of red ochre (10, 13).
Utilized ochre is found in almost all Stone
Age occupations in southern Africa that are
younger than 100 ka (13). The ochre may
have served only utilitarian functions (e.g.,
skin protection or hide tanning) (3)ormay
have been used symbolically as pigment (4,
10, 13). Evidence for the latter is a persistent
use of ochre with saturated red hues to pro-
duce finely honed crayon or pencil forms (10,
13). However, no ochre pieces or other arti-
facts older than ⬃40 ka provide evidence for
abstract or depictional images, which would
indicate modern human behavior (2, 14, 15).
We have recovered two pieces of engraved
ochre from the MSA layers at Blombos Cave,
South Africa. Situated on the southern Cape
shore of the Indian Ocean, the cave is 35 m
above sea level. A 5- to 60-cm layer of aeolian
sand containing no archaeological artifacts
(BBC Hiatus; Fig. 1) separates the LSA from
the MSA occupation layers. The MSA is divid-
ed into three substages (9, 10) (Fig. 1): (i) an
upper series of occupational deposits, BBC M1,
typified by abundant bifacially flaked, lanceo-
late-shaped stone points (Still Bay points) (10);
(ii) a middle series, BBC M2, containing fewer
Still Bay points but relatively abundant in de-
liberately shaped bone awls and points that
were probably hafted (9, 11); and (iii) a lower
BBC M3 series with few retouched pieces but
with blades and flakes typical of the Mossel
Bay/MSA 2b subphase (10). Associated, well-
preserved faunal remains from all layers indi-
cate that subsistence strategies were wide rang-
ing and include terrestrial and marine mam-
mals, shellfish, fish, and reptiles (10, 11).
More than 8000 pieces of ochre, many bear-
ing signs of utilization, have been recovered
from the MSA layers at Blombos Cave (10).
Seven of nine pieces are potentially engraved
and under study. We report here on the two
unequivocally engraved pieces recovered in situ
from layer CC, square E6a and layer CD,
square H6a (Fig. 1) (10) during excavations in
1999 and 2000, respectively. The engraved
ochre piece from layer CC (SAM-AA 8937)
was located adjacent to a small hearth, and that
from layer CD (SAM-AA 8938) was surround-
ed by a number of small, basin-shaped hearths.
Both specimens were located in a matrix of
undisturbed and consolidated mixed ash and
sand. There is no indication of perturbation in
either the overlying 15 to 20 cm of MSA de-
posits or in the blanketing aeolian dune sand
and no sign of intrusion of younger LSA ma-
terials (9, 10). All lithic artifacts in the ochre-
bearing and overlying MSA layers are typolog-
ically MSA (9, 10).
On the 8937 piece (Fig. 2, A and B), both
the flat surfaces and one edge are modified by
scraping and grinding. The edge has two
ground facets, and the larger of these bears a
cross-hatched engraved design. The cross
hatching consists of two sets of six and eight
lines partly intercepted by a longer line. The
engraving on 8938 (Fig. 2, C and D) consists
of a row of cross hatching, bounded top and
bottom by parallel lines and divided through
the middle by a third parallel line that divides
1
Iziko Museums of Cape Town, South African Muse-
um, Post Office Box 61, Cape Town, 8000, South
Africa.
2
Department of Anthropology, State Universi-
ty of New York at Stony Brook, NY, 11794, USA.
3
Centre for Development Studies, University of Ber-
gen, Strømgaten, 54, 5007 Bergen, Norway.
4
Institut
de Pre´histoire et de Ge´ologie du Quaternaire, UMR
5808 du CNRS, Avenue des Faculte´s, 33405, Talence,
France.
5
Luminescence Dating Laboratory, Institute of
Geography and Earth Sciences, University of Wales,
Aberystwyth, SY23 3DB, UK.
6
Laboratoire des Scienc-
es du Climat et de l’Environnement, UMR CEA-CNRS,
Avenue de la Terrasse, 91198 Gif-sur-Yvette cedex,
France.
7
Department of Archaeology, University of
Cape Town, Private Bag, Rondebosch, 7701, Cape
Town, South Africa.
*To whom correspondence should be addressed. E-
mail: chenshilwood@iziko.org.za
BBC Hiatus
CA
CI
CB
CC
CD
CJ
CFB/CFC
CFA
CGAC
CGB
CGAA
BBC M1
BBC M2
BBC M3
BBC LSA
0.5 m
SAM-AA 8938
*
SAM-AA 8937
*
Unexcavated
Fig. 1. Stratigraphy of Blombos Cave (34°25⬘S,
21°13⬘E). Sequence of MSA layers in square H6
showing relative location of engraved ochre
pieces SAM-AA 8937 and SAM-AA 8938, bifa-
cial Still Bay points (lanceolate shape), and
bone tools. Closed and open symbols for bifa-
cial points and bone tools indicate common
and rare occurrences, respectively. The MSA
layers consist principally of sands interlayered
with consolidated beds, lenses, and stringers of
marine shells, organic matter, and wood ash.
R EPORTS
15 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org1278

the lozenge shapes into triangles. Some of the
lines are well-defined single incisions; others
have parallel tracks along part or all of their
lengths. Much of the parallel tracking may
have resulted from a change in position of the
engraving tool causing simultaneous scoring
from more than one projection. The midline
comprises three marking events. Examination
of the intersections of the cross-hatched lines
indicates that they were not executed as con-
secutive cross hatchings but that lines were
made in first one direction and then another;
the horizontal lines overlie the cross hatch-
ing. The preparation by grinding of the en-
graved surface, situation of the engraving on
this prepared face, engraving technique, and
final design are similar for both pieces, indi-
cating a deliberate sequence of choices. Al-
though the engraving on the 8937 ochre has
fewer markings than the 8938 piece, it indi-
cates that 8938 is not unique; the engraving
on 8938 can be considered a complex geo-
metric motif as the cross-hatched lines are
bisected and framed by horizontals.
Assessing the significance of these engrav-
ings demands an accurate determination of their
age (16 ). The engraved ochres were found
within layers containing bifacially flaked stone
points; in the South African MSA, these stone
point types occur only within or below How-
iesons Poort horizons (10) dated to ⬃65 to 70
ka (17). This association suggests that the en-
gravings are older than 65 ka. To independently
confirm and refine this time frame, we applied
two luminescence-based dating methods to the
Blombos Cave layers. Thermoluminescence
(TL) dates were obtained for five burnt lithic
samples from the MSA phase BBC M1 (Fig. 1)
(18, 19). The mean age for the lithic samples is
77 ⫾ 6ka(20). To confirm the stratigraphic
integrity, we applied optically stimulated lumi-
nescence (OSL) dating to the aeolian dune
(BBC Hiatus) separating the LSA and MSA
layers (Fig. 1). Multiple grain measurements
with a single aliquot regenerative (SAR) proce-
dure (21) yielded a depositional age of 69 ⫾ 5
ka (22). Single-grain SAR measurements (23)
yielded consistent ages (24 ), indicating that the
aggregate samples were not contaminated by
grains of different ages (25). Because only
1.8% of the 1892 grains analyzed yielded re-
producible growth curves, a more representa-
tive approach was also used (26 ), combining
OSL signals from grains to generate synthetic
aliquots. These provide a depositional age of
70 ⫾ 5ka(20) and confirm the antiquity of the
engraved ochres.
Abstract images similar to the Blombos
Cave engravings occur at Upper Paleolithic
sites in Eurasia (15). The Blombos Cave mo-
tifs suggest arbitrary conventions unrelated to
reality-based cognition, as is the case in the
Upper Paleolithic (15), and they may have
been constructed with symbolic intent, the
meaning of which is now unknown. These
finds demonstrate that ochre use in the MSA
was not exclusively utilitarian and, arguably,
the transmission and sharing of the meaning
of the engravings relied on fully syntactical
language (5, 27).
Genetic and fossil evidence suggests that
humans were anatomically near modern in
Africa before 100 ka (5, 28, 29). Key ques-
tions are whether anatomical and behavioral
modernity developed in tandem (5) and what
criteria archaeologists should use to identify
modern behavior (2, 4, 5). For the latter, there
is agreement on one criterion—archaeologi-
cal evidence of abstract or depictional images
indicates modern human behavior (2, 14, 15).
The Blombos Cave engravings are intentional
images. In the light of this evidence, it seems
that, at least in southern Africa, Homo sapi-
ens was behaviorally modern about 77,000
years ago.
References and Notes
1. The term “modern human behavior” as used here has
no chronological implication and means the thoughts
and actions underwritten by minds equivalent to
those of Homo sapiens today. Key among these is the
use of symbols.
2. P. A. Mellars, K. Gibson, Eds., Modelling the Early
Human Mind (McDonald Institute Monographs, Cam-
bridge, 1996).
3. R. G. Klein, The Human Career (Chicago Univ. Press,
Chicago, IL, 1999).
4. H. J. Deacon, J. Deacon, Human Beginnings in South
Africa: Uncovering the Secrets of the Stone Age (Dav-
id Philip, Cape Town, South Africa, 1999).
5. S. McBrearty, A. Brooks, J. Hum. Evol. 38, 453 (2000).
6. F. d’Errico, P. Villa, J. Hum. Evol. 33, 1 (1997).
7. J. Yellen, A. Brooks, E. Cornelissen, M. Mehlman, K.
Stewart, Science 268, 553 (1995).
8. S. Wurz, thesis, Stellenbosch University, Stellenbosch,
South Africa (2000).
9. C. S. Henshilwood, F. d’Errico, C. W. Marean, R. Milo,
R. Yates, J. Hum. Evol. 41, 631 (2001).
10. C. S. Henshilwood et al., J. Archaeol. Sci. 28, 421 (2001).
11. C. S. Henshilwood, J. C. Sealy, Curr. Anthropol. 38,
890 (1997).
12. R. C. Walter et al., Nature 405, 65 (2000).
13. I. Watts, in The Evolution of Culture, R. Dunbar, C.
Knight, C. Power, Eds. (Edinburgh Univ. Press, Edin-
burgh, 1999), pp. 113–146.
14. G. A. Clark, C. M. Willermet, Conceptual Issues in
Modern Human Origins Research (de Gruyter, New
York, 1997).
Fig. 2. Engraved ochres from Blombos Cave. (A) SAM-AA 8937 is a flat piece of shale-like ochre that
grades into silt on the reverse side: weight ⫽ 39.2 g; maximum length ⫽ 53.6 mm; breadth ⫽ 42.6
mm, depth ⫽ 11.7 mm; streak color notation 3060 Y65R (33). (B) Tracing of lines verified as
engraved by study under magnification (scale bar, 5 mm). (C) SAM-AA 8938 is a rectangular slab
of ochreous shale: weight ⫽ 116.6 g; maximum length ⫽ 75.8 mm; breadth ⫽ 34.8 mm; depth ⫽
24.7 mm; streak color notation 4050 Y60R (30). Oblique lighting of specimen accentuates both
engraved lines and irregularities of the surface, some created by grinding before the engraving and
others by the process of engraving. (D) Tracing of lines verified as engraved by study under
magnification, superimposed on flat-bed scan of engraved surface (scale bar, 10 mm).
R EPORTS
www.sciencemag.org SCIENCE VOL 295 15 FEBRUARY 2002 1279

15. P. Bahn, J. Vertut, Journey Through the Ice Age (Wei-
denfeld & Nicolson, London, 1997).
16. A sediment sample from Blombos was included in an
earlier study of coastal sediments, in which it was
found to date to oxygen isotope stage 5 (30). How-
ever, the dates obtained for unetched quartz and
feldspar grains with TL and infrared stimulated lumi-
nescence, respectively, were inconsistent.
17. J. C. Vogel, in Humanity - from African Naissance to
coming Millennia, P. V. Tobias, M. A. Raath, J. Moggi-
Cecchi, G. A. Doyle, Eds. (Univ. of Florence Press,
Florence, Italy, 2000), pp. 261–268.
18. The five lithic specimens (a few cm in size and weighing
between 6 and 35 g) came from the upper phase of
occupational deposits BBC M1: Samples BBC24 and
BBC23 are from CA/CB and BBC15, BBC20, and BBC22
are from CC. Examination of thin sections revealed the
presence of quartz grains embedded in a siliceous ma-
trix. The time since they were burnt was computed
from TL analysis of 100- to 160-␮m grains obtained by
crushing after the samples’ outer surfaces had been
removed with a diamond saw (31). The equivalent dose
(D
e
) was determined with a combined additive and
regenerative dose protocol (19). U, Th, and K concen-
trations of the lithic samples were measured by neutron
activation analysis (32). The total dose rates (20) were
calculated assuming that the quartz grains were free of
radioactive impurities and that all radioisotopes were
confined to and uniformly distributed within the sur-
rounding siliceous matrix. In computing the alpha and
beta dose rates received by the grains, attenuation
factors appropriate for the mean grain size in each
specimen were taken into account. To determine the
gamma dose rates, we buried 24 dosimeters in the cave
deposits for 1 year at points no farther than 1 m from
each previously excavated lithic. The ages combine to
provide a mean age of 77 ⫾ 6 ka, which is consistent
with the OSL age for the overlying dune layer.
19. N. Mercier, H. Valladas, G. Valladas, Ancient TL 10,28
(1992).
20. Supplementary data including a table containing
dose rate, D
e
, and age information for TL and OSL
analyses and a supplementary figure showing a radial
plot of OSL D
e
values for the “synthetic” aliquot data
are available on Science Online at www.sciencemag.
org/cgi/content/full/1067575/DC1.
21. A. S. Murray, A. G. Wintle, Radiat. Meas. 32, 57 (2000).
22. The depositional age of the dune layer was deter-
mined by OSL dating. OSL analyses on Aber/52-ZB-
15 were undertaken on quartz grains to measure the
radiation dose that they had received since their last
exposure to daylight. Their equivalent dose (D
e
) was
determined with the SAR procedure. The total radi-
ation dose rate to the grains (20) was measured with
a combination of thick-source alpha counting, beta
counting, and atomic absorption spectroscopy for
potassium determination, and a water content of
10 ⫾ 5% (weight water / weight dry sediment) was
used, based on current moisture contents in the cave.
The calculated gamma dose rate was consistent with
that measured in the field, and the cosmic ray dose
rate of 45 ␮Gy/year was based on the thickness of
the overlying rock. Quartz grains were extracted after
treatment with 10% hydrochloric acid to remove
carbonates and 30 vol of H
2
O
2
to remove organics.
The sample was sieved to obtain grains from 212 to
250 ␮m in diameter. Feldspars and heavy minerals
were removed by density separation at 2.62 and 2.70
g/cm
3
. The alpha-irradiated outer layer of the grains
was removed by etching in 48% hydrofluoric acid for
45 min. The initial set of luminescence measure-
ments (stimulation at 470 nm with blue diodes) used
48 aliquots, each containing about 500 grains. A
range of thermal pretreatments (preheats) from 160°
to 300°C for 10 s was used. From 200° to 280°C, the
D
e
values from 30 replicate aliquots were reproduc-
ible and showed no systematic trend with tempera-
ture. A weighted mean was calculated, with the
individual D
e
values weighted according to their un-
certainty. The uncertainty in the mean was divided
by
公
N
, where N is the number of independent
estimates of D
e
—in this case 30. D
e
values were
calculated with the package Analyst, which combines
uncertainties due to the counting statistics of each
OSL measurement and the error associated with the
mathematical fitting of the growth curve to the
luminescence data. The 30 aliquots gave a value of
47.9 ⫾ 1.7 Gy, resulting in an age of 69 ⫾ 5 ka.
23. L. Bøtter-Jensen, E. Bulur, G. A. T. Duller, A. S. Murray,
Radiat. Meas. 32, 523 (2000).
24. The single-grain SAR measurements were made in an
automated reader based around a 10-mW, 532-nm
Nd: YVO
4
laser, whose beam can be directed at
individual grains (23). A single preheat at 220°C for
10 s was used, with the main OSL measurement (L)
being followed by measurement of the OSL response
(T) to a test dose as observed after a 160°C cut heat
(21). Grains whose natural signal ratio (L
N
/T
N
) does
not intersect the regeneration growth curve were not
used in subsequent analyses. In addition, grains
thought to contain some feldspar were also rejected.
The presence of feldspar was identified by making
additional measurements of a given regeneration
dose on each grain. The first measurement is under-
taken within the SAR procedure and yields the ratio
L
1
/T
1
used in the growth curve. For each grain, two
additional measurements of L/T were made at the
end of the SAR procedure. The first duplicates the
previous measurement, as a test of the sensitivity
correction. The second uses the same regeneration
dose, but, before preheating, the grains are exposed
to infrared (830 nm) radiation from a 500-mW laser
diode for 100 s. If the grains contain feldspar, then
the infrared exposure will have reduced the magni-
tude of L, and hence the ratio of these last two
measurements of L/T will be substantially less than
unity; for a quartz grain, the ratio will be consistent
with unity. Of the 1892 grains that were measured,
22 were rejected on the basis of these criteria. The
OSL signal from many of the remaining 1870 grains
was close to instrumental background, and only 34
yielded reproducible growth curves.
25. R. G. Roberts et al., Nature 393, 358 (1998).
26. Single-grain OSL measurements were made with 19
aluminum discs, with up to 100 grains on each disc.
For each disc, the OSL signals from the unrejected
grains were combined to generate “synthetic” ali-
quots consisting of between 93 and 100 grains. The
D
e
values for 18 “synthetic” aliquots were combined
to give a weighted mean of 48.5 ⫾ 1.2 Gy, giving a
depositional age of 70 ⫾ 5ka(20).
27. L. Aiello, N. G. Jablonski, Eds., The Origin and Diver-
sification of Language (Memoirs of the California
Academy of Sciences, San Francisco, 1998).
28. C. B. Stringer, in Contemporary Issues in Human Evo-
lution, W. Meikle, N. Jablonski, Eds. (California Acad-
emy of Sciences, San Francisco, 1996), pp. 115–134.
29. M. Ingman, K. Kaessmann, S. Pa¨a¨bo, U. Gyllensten,
Nature 408, 708 (2000).
30. J. C. Vogel, A. G. Wintle, S. M. Woodborne, J. Ar-
chaeol. Sci. 26, 729 (1999).
31. H. Valladas, Quat. Sci. Rev. 11, 1 (1992).
32. J.-L. Joron, thesis, Universite´ Paris-Sud (1974).
33. “Natural Color System Index” (Scandinavian Colour
Institute, Stockholm, 1999).
34. This work was supported by grants to C.H. from the
Anglo American Chairman’s Fund, Centre National de
la Recherche Scientifique OHLL, the Leakey Founda-
tion, the National Geographic Society, NSF, the
South African National Research Foundation, and the
Wenner-Gren Foundation; to F.D. from CNRS Origine
de l’Homme, du Langage et des Langues (OHLL) and
the Service Culturel of the French Embassy in South
Africa; to Z.J. from the Sir Henry Strakosch Memorial
Trust and an Overseas Research Student award; to
C.T. from CNRS OHLL; to G.D. from the Natural
Environment Research Council; to J.S. from the South
African National Research Foundation and the Uni-
versity of Cape Town; and to I.W. from the British
Academy. We thank G. Avery at Iziko Museums of
Cape Town, the South African Museum, and K. van
Niekerk.
30 October 2001; accepted 28 December 2001
Published online 10 January 2002;
10.1126/science.1067575
Include this information when citing this paper.
Marine Biodiversity Hotspots
and Conservation Priorities for
Tropical Reefs
Callum M. Roberts,
1,2
* Colin J. McClean,
2
John E. N. Veron,
3
Julie P. Hawkins,
2
Gerald R. Allen,
4,8
Don E. McAllister,
5
†
Cristina G. Mittermeier,
4
Frederick W. Schueler,
6
Mark Spalding,
7
Fred Wells,
8
Carly Vynne,
4
Timothy B. Werner
4
Coral reefs are the most biologically diverse of shallow water marine ecosystems
but are being degraded worldwide by human activities and climate warming.
Analyses of the geographic ranges of 3235 species of reef fish, corals, snails, and
lobsters revealed that between 7.2% and 53.6% of each taxon have highly re-
stricted ranges, rendering them vulnerable to extinction. Restricted-range species
are clustered into centers of endemism, like those described for terrestrial taxa. The
10 richest centers of endemism cover 15.8% of the world’s coral reefs (0.012% of
the oceans) but include between 44.8 and 54.2% of the restricted-range species.
Many occur in regions where reefs are being severely affected by people, potentially
leading to numerous extinctions. Threatened centers of endemism are major biodi-
versity hotspots, and conservation efforts targeted toward them could help avert
the loss of tropical reef biodiversity.
Coral reefs fringe one-sixth of the world’s
coastlines (1) and support hundreds of thou-
sands of animal and plant species (2). Fifty-
eight percent of the world’s reefs are reported to
be threatened by human activities (3). Terres-
trial agriculture, deforestation, and develop-
ment are introducing large quantities of sedi-
ment, nutrients, and other pollutants into coastal
R EPORTS
15 FEBRUARY 2002 VOL 295 SCIENCE www.sciencemag.org1280