Journal of Anthropological Archaeology 25 (2006) 448–465
www.elsevier.com/locate/jaa
Interpreting pachyderm single carcass sites in the African Lower
and Early Middle Pleistocene record: A multidisciplinary
approach to the site of Nadung’a 4 (Kenya)
Anne Delagnes a,¤, Arnaud Lenoble a, Sonia Harmand b, Jean-Philip Brugal c,
Sandrine Prat d, Jean-Jacques Tiercelin e, Hélène Roche b
a
CNRS, UMR 5199, PACEA, Avenue des Facultés, 33405 Talence, France
CNRS, UMR 7055, Maison de l’Archéologie et de l’Ethnologie, 21 allée de l’Université, 92023 Nanterre, France
c
CNRS, UMR 6636, Maison Méditerranéenne des Sciences de l’Homme, 5 rue du Château de l’Horloge, 13094 Aix-en-Provence, France
d
CNRS, UPR 2147, 44 rue de l’Amiral Mouchez, 75014 Paris, France
e
CNRS/UBO, UMR 6538, Institut Universitaire Européen de la Mer, Place Copernic, 29280 Plouzané, France
b
Received 12 January 2006
Available online 19 April 2006
Abstract
Nadung’a 4 is one of the single carcass pachyderm sites recorded in East Africa during the Lower and Early Middle
Pleistocene. The site has yielded an abundant lithic assemblage in close association with the partial carcass of an elephant.
Conjoined pedological, geoarchaeological, spatial, technological, and taphonomical analyses have been carried out to
address the relationship between hominids and elephant. The resulting data are consistent with a non-fortuitous association between both categories of remains. The lithic artefacts do not match a classical Acheulean tool-kit, as would be
expected for the time period ascribed to the site, and the functional patterns inferred from their analysis make this site radically diVerent from other purported butchery sites. The implications of these original features are discussed.
2006 Elsevier Inc. All rights reserved.
Keywords: East Africa; West Turkana; Lower Pleistocene; Early Middle Pleistocene; Butchery site; Pachyderms; Geoarchaeology; Spatial
analysis; Lithic technology; Site function
Hominid processing of pachyderm carcasses is
documented by a set of sites in the African archaeological record, ranging from the beginning of the
Lower Pleistocene to the Middle Pleistocene. The
sites most speciWcally interpreted as butchery sites or
type B sites, according to Isaac’s model (Isaac and
*
Corresponding author.
E-mail address: a.delagnes@ipgq.u-bordeaux1.fr (A. Delagnes).
0278-4165/$ - see front matter 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jaa.2006.03.002
Crader, 1981), usually yield a single, large mammal
skeleton, and a small number of associated stone
tools. This functional interpretation relies upon the
spatial vicinity of both categories of remains and is
rarely supported by direct evidence of hominid
intervention on the carcasses, such as cutmarks,
intentional breakage of the bones or transport of
some body parts from the sites. The site of FLKN-6
at Olduvai is one of the rare occurrences where cutmarks have been observed on elephant bones
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
(Bunn, 1981; Potts and Shipman, 1981). When such
observations are possible, butchering traces are
always too few and too tenuous to suggest even
roughly the nature of the butchery operations carried out. The mode of procurement of these very
large carcasses, either by scavenging or by hunting,
is also almost impossible to infer from the archaeozoological data, a situation that is equally true for
European sites (Villa, 1990; Villa et al., 2004).
The main interest of such occurrences is the
monospeciWc aspect (i.e., butchery) of hominid activities, as implied by the faunal assemblage and
related stone tools. Although criticized as an “oversimpliWed rendering” (Potts, 1988, p. 150), the type B
site model is still commonly used by prehistorians
when referring to this type of site. Is the single-task
interpretation conWrmed when questioning the
whole data set provided by the archaeological context? Conjoined pedological, geoarchaeological, spatial, technological, and taphonomic analyses have
been carried out to deal with this question on the
recently excavated site of Nadung’a 4 (West Turkana). An abundant lithic assemblage has been
recovered from this site in close association with the
partial carcass of an elephant. The unusual quantity
of artefacts, outnumbering any other assemblage
known from similar contexts, provides additional
interest to this research.
Archaeological context
The site of Nadung’a 4 is located on the west bank
of lake Turkana, in the Nachukui formation. This geological formation includes a succession of hominid
occupations dated between 2.4 million years (My) and
700,000 years BP (Prat et al., 2003; Roche and Kibunjia, 1994, 1996; Roche et al., 1992, 2003). At the
beginning of this time interval, south of the formation,
the Lokalalei sites are among the very few known
African Pliocene sites (Delagnes and Roche, 2005;
Kibunjia, 1994; Kibunjia et al., 1992; Roche et al.,
1999). At the top of the sequence and north of the formation, the Nadung’a sites are part of a well-documented series of East African sites related to the end of
the Late Pleistocene, or to the very beginning of the
Middle Pleistocene, generally characterized by Acheulean industries. Some 20 Oldowan and Acheulean sites
are recorded in the same formation between these two
time periods, showing various levels of technological
development (Roche et al., 2003).
The site of Nadung’a 4 was excavated during
three Weld seasons, in 2002, 2003, and 2004, as part
449
of the West Turkana Archaeological Project
co-directed by Hélène Roche and Mzalendo Kibunjia (National Museums of Kenya). The excavation covers an area of 53 m2 (Fig. 2), including the
entire site as it actually exists. Although both eastern
and western edges of the deposit have been truncated by erosion, we deduce from the conWguration
of the archaeological layer that its initial spatial
extension was not much greater. The total thickness
of the deposit, ranging between 0.30 and 1.50 m, has
been excavated. The in situ recovered assemblage
comprises 6797 lithic artefacts made from lavas
(rhyolite, phonolite, trachyte and basalt) and 142
faunal remains, mostly attributed to Proboscideans.
The density of the remains is high, contained in a
single sedimentological horizon, spatially conWned
within a small area. The corresponding horizons
surrounding the site are completely sterile and the
surface material scattered on the erosional slopes
west and east of the site originated from the same
conWned area.
The Nadung’a sites complex spans a series of
small hills cut by gullies. According to Harris’ mapping (Harris et al., 1988), the outcrop deposits are
correlated to the Nariokotome member, which
ranges in age from 1.33 My to about 700,000 years
BP. The seven sites recorded in the Nadung’a complex are all located in the middle part of the Nariokotome member, 17–20 m above the top of the
Lower Nariokotome TuV (Fig. 1). This marker horizon outcrops at a distance of 100 m north-west of
Nadung’a 4. The stratigraphic position of the site
has been established through successive surveys
along an ephemeral stream (Nadung’a) and through
a bed to bed correlation. Nadung’a 4 is the most
recent site of the complex and is situated some 150 m
north of Nadung’a 1 and a few meters above this
site, where pumices have been recorded the composition of which is similar to pumices from the Silbo
dated to 740,000 years BP (Kibunjia et al., 1992).
Nadung’a 4 is slightly younger and can be correlated
to the beginning of the Middle Pleistocene, i.e., circa
700,000 BP.
This time period is crucial for understanding the
place of Homo erectus in human evolution and the
place and mode of origin of Homo sapiens. Despite
abundant lithic and faunal evidence for this time
period in Africa, hominid fossils are rare. They are represented by partial crania from Buia, Eritrea [UA 31,
dated about 1 My (Abbate et al., 1998)], the Daka calvaria, Ethiopia [BOU-VP-2/66, about 1 My (Asfaw
et al., 2002)], Melka Kunture, Ethiopia [Gombore II,
450
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Fig. 1. Lithostratigraphic log established through four sections (A, B, C, D) spaced out along the Nadung’a stream. Nadung’a 4 stratigraphic position is inferred from a bed to bed correlation between the archaelogical deposit (A) and the Lower Nariokotome TuV (D).
about 0.8 My (Chavaillon and Piperno, 2004)], Olorgesailie, Kenya [KNM-OL 45500, between about 0.97
and 0.9 My (Potts et al., 2004)], and Olduvai, Tanzania
[OH 12, OH 22, OH 23, and OH 51]. Unfortunately,
no hominid specimens have been discovered in the
Nachukui Formation for this time period.
Site formation processes
The archaeological deposit at Nadung’a 4 lies in
the lower part of a stratigraphic unit made of brown
clays including lenses (about 1 m thick) of ill-sorted
sands and gravels. The sand and gravel lenses result
from the Wlling of small sinuous channels, as attested
by the high-angle planar cross-bedding of the inWlling (Reineck and Singh, 1980). This feature, associated with high-textural contrasts, indicates a
seasonal stream regime or at least an intermittent
one (Reading, 1996). All these characteristics point
to a sedimentary environment of mud Xats crosscut
by small streams and regularly swamped during the
seasonal Xoods of the lake or tributaries. The sedimentological sequence of the site is homogeneous,
with only slight diVerences in the Wrst 30 cm of
deposits due to the smaller size of the carbonated
concretions and to the presence of desiccation
cracks. The sediment is a massive, sandy silty clay
with a prismatic structure. A noteworthy feature is
the slickensides (glossy and striated surfaces), which
cross the deposit without any preferential orientation. Secondary characteristics are hydromorphic
features, i.e., green punctuations on the slickensides
and dark punctuations inside the deposits. Whitish
nodules of carbonates, the dimensions of which
increase with depth, are also visible. Two other characteristics related to wet/dry cycles can be observed
in thin sections: crack inWllings and cross-striated
b-fabric (Blockuis et al., 1990).
Such features are indicative of a pedogenesis of
vertisol type (Soil Survey StuV, 1999). This diagnosis
is based on the conjunction of three criteria that are
used to deWne a vertisol (Eswaran et al., 1999):
• the presence of cracks developed during dry periods;
• a proportion of clay greater than 30%;
• the development of slickensides with long axes
tilted 10°–60° from the horizontal in the deep
horizon of the soil.
This type of pedogenesis is typical of wet/dry
tropical climates, where soils exhibit swelling upon
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
451
Fig. 2. Class-size frequencies of the lithic remains in two test-squares compared with an experimental Xaking set made on rhyolite (left).
The experimental set has been carried out following the same principles of debitage than in the archaeological assemblage. The interpretation Weld of the experimental set (dashed) has been drawn after Schick (1986) and Lenoble (2005) experiments. Fields indicating the waterXow modiWcations on size distribution (right) are based on the experiments carried out by Lenoble (2005). As seen on this graph, the
hypothesis of Xuvial modiWcation can be refused.
wetting and shrinking when dried. A carbonated
horizon is found when vertisols have been formed
under semi-arid to arid climates (Eswaran et al.,
1999). The Nadung’a 4 vertisol shares similar morphological, structural, and textural features with the
cumulative pedotype Aberegaiya identiWed in
Northern Kenya (Wynn, 2001). Wynn’s paleoenvironmental interpretation of this type of paleosol can
be applied to Nadung’a 4: low topographical positions such as Xoodplains, in semi-arid regions with
an alternation of long dry seasons and short pluvial
events. Such soils usually support a woody savanna
vegetation. These contexts are equally characterized
by frequent seasonal Xoods and subsequent sedimentation episodes, as suggested by the sedimentary
features.
In the case of vertisols on mud Xats, two main
processes are likely to have distorted the archaeological layer: water transport and argilliturbation.
Estimating the impact of such processes is crucial
for assessing whether bone and lithic remains were
primarily associated or rather secondarily brought
together within the same unit. This issue is particularly relevant with regard to the dispersal of the
remains within a thick archaeological layer.
The hypothesis of water transport has been tested
through a series of criteria initially used by Schick
(1986) and by Petraglia and Potts (1994): artefact size
distribution, physical alteration, and the horizontal
distribution of lithic remains. Neither the fabric (ori-
entation and dip) nor the small-scale spatial features
such as shadow eVects (Schick, 1986) are considered
here for the purpose of estimating the impact of
water transport. We assume that such features would
not have been preserved owing to the argilliturbation
inferred from the sedimentary context.
Artefact size distribution
Sediments from two test squares have been
water-screened and all the artefacts thus collected
have been measured, together with the plotted material. The data recovered have been compared with
an experimental assemblage produced using the
same technical principles identiWed in the archaeological assemblage and consisting in the knapping of
one block of rhyolite according to a discoidal
method. The relative proportions of small lithic
artefacts recovered from the two test squares during
screening (between 2 and 10 mm screen mesh) are
very similar (89.9% in B20 and 86.3% in J20) and
consistent with the experimental set (88.9%). The
strong similarity in size distribution between the
experimental and archaeological sets (Fig. 2) makes
the latter compatible with a knapping locality. The
size distribution of the products resulting from the
experimental set is also consistent with the data
recorded for experimental assemblages (Fig. 2) produced by other means and on diVerent rocks
(Lenoble, 2005; Schick, 1986). Such observations
452
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
have made possible a comparison in terms of size
distribution between the archaeological assemblage
and several other experimental assemblages sorted
by various types of water Xow, or non-sorted (Lenoble, 2005). The lithic sample from Nadung’a 4 corresponds to a non-sorted assemblage.
Physical alterations
All of the artefacts were macroscopically examined. No abrasion or any other edge damage related
to mechanical alteration have been observed. As a
whole, the lithic assemblage is extremely fresh. This
is especially true for the rhyolite artefacts, while the
other volcanic rocks sometimes show an in-depth
alteration, which could be the result of a geochemical alteration related to their porosity. The elephant
bones are also badly preserved: desquamated bone
surfaces and in-depth bone alterations, resulting in a
white and chalky structure, are the most common
observable features. Post-burial alteration is
deduced from the sedimentary context; repeated episodes of desiccation and then dampening within the
deposits are suYcient to account for the alteration
of the bones, even if previous weathering, no longer
distinguishable, may also be assumed.
Horizontal distribution
A rough examination of the horizontal spatial
distribution of archaeological material (Fig. 3)
shows a non-homogeneous spatial distribution:
highly concentrated areas (less than 1 m in diameter)
adjoin low-density areas. The concentrations have a
circular shape and are clearly distinct from the linear concentrations documented by Schick for deposits reworked by Xuvial Xows (Schick, 1986).
All the criteria tested for evaluating the impact of
water transport are consistent with the absence of
any modiWcation by water Xows related to the Xood
plain system. The eVects of argilliturbation are
investigated through the vertical distribution of the
remains and their fabric (below).
Vertical distribution
The archaeological layer is very dilated, with a
total thickness ranging over 1.5 m in the southern
part of the site (Fig. 3). In detail, the distribution of
the artefacts appears more complex. The slickensides, formed during massive localized slips within
the deep horizons of the vertisol, have distorted the
overall distribution of the pieces (>2 cm) which is
characterized by sharp lower outlines, following
oblique plans ranging between 15° and 35°. The slickensides are also responsible for splitting the dense
area visible in the northern part of the site into two
secondary concentrations, as was observed during
excavation. Their organization results typically in a
bowl structure. Such bowl structures characterize
the in-depth geometry of vertisols associated with a
gilgai topography: involutions of deeply buried
horizons related to slipping along slickensides are
regularly arranged in the polygonal network of dessiccation cracks (Eswaran and Cook, 1988). These
structures explain the wavy limits of the archaeological layer.
Regarding the whole assemblage (i.e., all pieces
>2 mm) from the two test-squares, the vertical distribution of the material, according to the successive
levels established during excavations, follows a normal distribution (Fig. 4). There is a progressive
decrease in the number of recorded artefacts
towards the top and towards the bottom of the
deposit. Such characteristics are consistent with
artefact dispersal from a single archaeological layer.
The vertical dispersal is more important towards the
bottom of the deposit, which is indicative of argilliturbation—when artefacts dropped into the
cracks—as mentioned by DuYeld (1970) and Wood
and Johnson (1978).
Fabric
Considering the fabric, the vector magnitude L
(Curray, 1956) has a low value: 4.6%, which means
that the remains do not follow any preferential orientation (Lenoble and Bertran, 2004). As shown in
Fig. 5, their inclinations are also extremely heterogeneous, indicating an isotropic fabric. The position of
Nadung’a 4 lithic assemblage on Benn’s diagram
(Benn, 1994) is consistent with deformation due to
argilliturbation.
Argilliturbation appears to be the main diagenetic process involved in the formation of the site.
The swell/shrink process is responsible for the presence of cracks and for the bowl structure of the
deposit. Both phenomena have produced up and
down vertical displacements within the sediment,
together with limited vertical size-sorting of the
material. Lateral displacements, on the order of 1 m
or so, may have also occurred as a result of the formation of bowl structures and localized massive
slips of sediments.
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
453
Fig. 3. Distribution of all the remains in the archaeological layer, (A) horizontal distribution, (B) density areas, (C) vertical projection
according to the Y-axis, and (D) partial vertical projections [bands 1 and 2 mapped in (A)].
454
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Fig. 4. Lithic artefacts frequency per levelling.
Fig. 5. Fabric of the assemblage plotted on the triangular Benn’s
diagram (1994) with interpretation Welds from Lenoble and Bertran (2004).
Spatial plotting per category of remain brings
additional information to estimate the extent of
deformations and the homogeneity of the assemblage. For this purpose, artefacts derived from the
same original block of raw material have been
grouped together. Particularly relevant are six
groups corresponding to blocks of rhyolite showing
very distinctive physical features (in terms of their
color and veining). Inside these groups, some products have even been reWtted. Their vertical dispersal
corresponds to the overall dispersal of the material
inside the archaeological layer (Fig. 6, panel 1). Such
distribution is consistent with an homogeneous
assemblage deposited during a single event or at
least a single period of human presence. For each
block, most of the corresponding products are
grouped within one of the concentrations, and only
a few pieces are dispersed some meters away from it.
This conWguration conWrms that these concentrations are anthropic features, and most likely reXect
knapping areas whose spatial distribution was subsequently dilated.
When comparing the vertical dispersal of the elephant remains with the dispersal of the stone tools,
it appears that both categories of remains are closely
associated (Fig. 6, panel 2). The elephant bones are
distributed throughout the archaeological layer.
Owing to their large dimensions, they cannot be
intrusive elements and they clearly form part of this
in situ layer. Most elephant elements were recovered
within the artefact concentrations described above,
scattered all around (Fig. 7), which suggests that
hominids knapped directly around the elephant
bones.
Geoarchaeological and spatial data recovered at
Nadung’a 4 demonstrate that:
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
455
Fig. 6. Vertical projections according to the Y-axis; panel 1, artefacts attributed to the same initial block of rhyolite (one symbol per
block); panel 2, faunal remains (the large triangles correspond to the remains identiWed as elephant while the small black triangles are the
undetermined fragments which are probably mostly alterated elephant remains).
Assuming that the human occupation is contemporaneous with the deposition of the elephant
bones, the functional signiWcance of this uncommon
association between an abundant lithic assemblage
and the elephant remains still has to be discussed.
Lithic technology can provide the key to such questions, by exploring the functional potential of the
stone tools.
Knapping activities : technological and functional
issues
Fig. 7. Elephant scapula surrounded by artefacts.
• Lithic and elephant remains are primarily associated in the archaeological layer, insofar as:
– water Xow has not aVected the lithic remains,
and neither the sedimentary context nor the
spatial conWguration of the scattered bones
suggest that the elephant remains could have
been washed in, and
– the vertical dispersion of the remains derives
from a single original layer subsequently
aVected by argilliturbation;
• A short period of open-air exposure was followed
by the rapid burying of the remains, as suggested
by the sedimentary context, which substantiates
the hypothesis of a single event or at least of a
single period of occupation.
Raw material procurement consisted in the
transport of volcanic rocks available in the close
vicinity of the site. Raw material sampling was carried out in a debris-Xow outcrop, stratigraphically
situated a few meters above the archaeological
layer. This deposit corresponds to the development
of an alluvial fan and is included in the same
regressive sequence as the unit of brown clays containing the archaeological level. We deduce from
this position that such a fan existed when the site
was formed, but was situated a few hundred meters
away, towards the ranges that lie on the western
border of the formation. It is therefore assumed
that this horizon is representative of the raw
material sources that could have been exploited by
the Nadung’a 4 knappers. That makes a comparison between the diVerent kinds of rocks collected
in the archaeological assemblage and in the debrisXow outcrop relevant. This comparison indicates a
clear preference for rhyolite (Fig. 8). Poorly
represented at the raw material source (5%), it is
456
Source
Site
60
35
%
N=2049
70
N=2196
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
50
Rhyolite
Phonolite, basalte,
trachyte
30
40
20
N=1375
30
25
10
N=958
s
20
Si
l.
ro
ck
yt
e
ch
t
Tr
a
Ba
sa
l
no
Ph
o
R
hy
ol
ite
lit
e
0
15
Fig. 8. Raw materials frequencies at Nadung’a 4 in comparison
with the potential source.
N=91
N=65
N=5
5
N=11
the predominant raw material in the archaeological assemblage (64%) thus pointing to highly selective raw material procurement. Rhyolite, mostly
red, sometimes green or more rarely grey, brown or
yellow, is a particularly Wne-grained raw material,
very suitable for obtaining sharp cutting edges, but
less compact and homogeneous, owing to frequent
internal Wssures, than the other available rocks
which include phonolite, basalt and trachyte, ranging mostly from medium to coarse-grained qualities. Unlike rhyolite, which comes as diaclasic
angular blocks, phonolite, basalt and trachyte correspond to rounded compact pebbles.
The artefacts can be segregated into two main
categories: light-duty tools and heavy-duty tools.
Light-duty tools, by far the most abundant, result
from a debitage reduction sequence. They correspond to Xakes and debris smaller than 2 cm,
Xakes greater than 2 cm, some of which were secondarily transformed into formal tools, and
related cores. These products are mainly made
from rhyolite and to a lesser extent from mediumgrained phonolite. Heavy-duty tools correspond
to worked pebbles, which are exclusively shaped
from phonolite, basalt and trachyte. A few hammerstones in quartz, basalt and phonolite are also
present in the assemblage. The high proportion of
small elements (<2 cm), the signiWcant quantity of
cores and the presence of hammerstones (Fig. 9)
are consistent with knapping activities carried out
on the spot.
The cores are made from blocks or pebbles, but
also frequently from large Xakes. Four distinct types
of cores have been identiWed (Fig. 10). The corresponding Xakes are inferred from a small number of
reWts as well as from the morpho-technical analysis
of the Xakes compared with the cores. The core
types include:
N=41
10
0
1
2
3
4
5
6
Fig. 9. Lithic assemblage composition; 1, whole pebbles; 2,
worked pebbles; 3, hammerstones; 4, Xakes and fragments <2 cm;
5, Xakes and fragments >2 cm; 6, cores.
• discoidal cores (Fig. 10: 1 and 2), with one preferential Xaked surface and centripetal removals
extending all around the cores. Platforms are prepared with large deep removals on the other face,
and natural surfaces are often directly used as
striking platforms. The resulting Xakes (Fig.10:
6–11) are mostly short and wide with multidirectional removal negatives on their dorsal face,
wide and thick plain or dihedral butts, at least
one long sharp cutting edge and frequently a
pointed disto-lateral end. Pseudo-Levallois points
are also part of the products obtained from these
cores;
• cores with one preferential Xaked surface and
unidirectional removals (Fig. 10: 3 and 4). They
are produced from a single platform prepared
with one large removal or consisting of a fracture
plane. They usually have a prismatic cross-section. The corresponding Xakes (Fig. 10: 12–14)
are somewhat elongated, with negatives of unidirectional removals on their dorsal face as well as
regular and continuous sharp cutting edges, on at
least one of their lateral edges. The opposite edge
can be backed, resulting from the removal of the
core’s edge (corresponding to an éclat débordant:
Beyries and Boëda, 1983). Their butts are mostly
wide and plain;
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
457
Fig. 10. Cores and Xakes; 1 and 2, discoidal cores; 3 and 4, cores with one preferential Xaked surface and unidirectional removals; 5, cores
with several alternatively Xaked surfaces and multidirectional removals; 6–11, Xakes issuing from discoidal cores; 12–14, Xakes issuing
from cores with one preferential Xaked surface and unidirectional removals.
• cores with several alternatively Xaked surfaces
and multidirectional removals (Fig. 10: 5). No
speciWc platform preparation has been conducted
insofar as each negative serves as a striking plat-
form for the following removal, on a secant face.
The Xakes produced that way have various
shapes, and are frequently short with thick and
asymmetrical cross-sections;
458
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
• cores on Xakes or large multiple notches. This
ambiguous category is characterized by at least
one large and deep removal creating a notch, or
several non-contiguous notches on the edge of
the blank. It is impossible to establish whether
they correspond to cores or to tools (or both) on
Xakes. The resulting products are small and wide,
with a large plain butt and an acute dorsal face/
butt angle. Some Xakes, detached from the ventral face of the blank, are Kombewa Xakes.
The cores are made using the rhyolite and, to a
lesser extent, the phonolite. The reduction sequence
is clearly carried out preferentially on the Wnegrained rocks. The other rocks, mostly of medium
and coarse-grained quality, are exploited according
to the same principles but with less normalization.
On the whole, the Xakes detached from these diVerent types of cores are quite diverse in terms of size,
morphology and elongation. Nonetheless, they
share common morpho-functional features, which
are: long, continuous, and sharp cutting edges, either
in lateral or distal positions, opposed to a thick and
steep edge formed by a lateral back or by the butt of
the Xake, equally suitable for prehension. The sharp
cutting edges may have been used as such, without
any modiWcation. This is suggested by the high ratio
(17 to 1) of unmodiWed Xakes (>2 cm) per core,
which shows that most of these Xakes have been
abandoned on the spot, after being produced and
probably partly used there.
A signiWcant portion of the Xakes and a few cores
have been transformed by retouching (410 pieces,
i.e., 11% of the potential blanks, including Xakes
>2 cm and cores). Most of them are in rhyolite
(77%). They correspond, for a large majority (98%),
to notches and denticulates (Fig. 11: 1–9). The
notches are either single or multiple and the retouch
is similar to that of the denticulates: clactonian, deep
and often invasive. They are mostly located on the
initial sharp cutting edge of the blank, either lateral
or distal, which has the eVect of radically transforming its functional properties. Denticulates can be single, or more seldom double, with two convergent
edges creating a roughly pointed extremity. Notches
as well as denticulates are often situated close to an
unmodiWed narrow or pointed end, resulting in a
kind of bec (Fig. 11: 5–9). The repeated location of
the retouch on the same portion of the blanks, their
depth as well as the freshness of the adjacent edges
strongly suggest that we are dealing with an intentional set of notched tools and not with “tools” acci-
dentally created by mechanical alterations, as
documented in other contexts (Vallin et al., 2001).
With regard to the transformations generated by
retouches and the morpho-technical features thus
produced, we assume that such tools fulWl functional
needs distinct from the unmodiWed Xake component.
Besides whole pebbles (N D 37) in phonolite and
basalt, worked pebbles comprise only a few pieces
(N D 10, Fig. 11: 10 and 11) made from the same
raw material. They proceed from a very simple
shaping reduction sequence, consisting in the
removal of a limited number of large alternate
Xakes on one or several adjacent portions of the
pebbles, creating sinuous and bulky edges. Such
edges are clearly not suitable as cutting edges. The
large dimensions of these tools (maximum length:
10–24 cm; maximum width: 6–12 cm), their compactness and the sturdiness of the edges indicate
that they correspond to blunt implements, in keeping with heavy-duty activities such as breaking or
crushing hard materials.
The Nadung’a 4 assemblage shows a clear complementarity between:
• the Wne-grained raw materials (rhyolite and to a
lesser extent phonolite) exploited for producing
Xakes and notched tools, as a result of a debitage
reduction sequence, and
• the coarser raw materials (phonolite, basalt) dedicated to the making of heavy-duty tools (worked
pebbles) by means of a simple shaping sequence.
Direct functional inferences are, of course, impossible to propose without micro-wear data (tests are
in progress by J.P. Caspar). Nevertheless, the technological composition of this assemblage and its
functional potential make it fully suitable for processing an elephant carcass: light duty tools, in this
case unmodiWed Xakes, could have been used for
cutting meat, while heavy-duty tools could have
served for breaking bones.
The production system is quite elaborate with
regards to the selective procurement and exploitation of raw materials according to their properties
and in relation to functional needs. The debitage
process itself is not very elaborate; it looks simple
but at the same time it is fully eYcient for the
inferred purpose (i.e., the obtaining of cutting edges).
Discoidal debitage is the predominant method: such
a method, which is found in many diVerent chronological and cultural contexts (Peresani, 2003),
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Fig. 11. 1–5, Formal tools; 9, denticulates; 6–8, notches, 10 and 11, worked pebbles.
459
460
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
appears in East Africa as early as 1.5 My in the
Ugandan site of Nyabusosi (Texier, 1995). Little
attention has been paid to the Xakes and more generally to the light duty tools present in contemporaneous Acheulean assemblages. Direct comparisons
are therefore diYcult to carry out. But it is at least
possible to point out that in many Acheulean assemblages the technical complexity lies with the shaping
process for producing large cutting tools (handaxes
and cleavers), rather than with the debitage component (Roche and Texier, 1996; Texier and Roche,
1995). Flakes and tools on Xakes proceeding from a
debitage reduction sequence are usually presented as
expedient in the Acheulean record; such is not the
case at Nadung’a 4. Moreover, the large cutting
tools, i.e., handaxes and cleavers, which form the
diagnostic features of the Acheulean and which are
present in other sites of this complex, are totally
absent there.
Assemblages without large cutting tools have
already been mentioned and described in East
Africa during the Lower and Middle Pleistocene.
Such industries are usually likened to the Mode 1
(Clark, 1961), or Oldowan techno-complex. They
are documented in the Middle Awash Formation in
Ethiopia (de Heinzelin et al., 2000; Schick and
Clark, 2003), where they are interpreted as a variant,
without handaxes and cleavers, of the local Acheulean. They are correlated in this context with expedient tool production, fulWlling immediate needs by
means of local raw materials during short periods of
occupation of the sites by small groups of hominids.
This interpretative model does not apply to the
Nadung’a 4 assemblage: lithic production cannot be
considered expedient here—rather, it is a structured
activity requiring planning and anticipation. The
Nadung’a 4 assemblage is therefore not consistent
with a Mode 1 or Oldowan techno-complex. Neither
can it be grouped with the Middle Stone Age,
the earliest occurrences of which are much more
recent and signiWcantly diVerent technologically
(McBrearty and Brooks, 2000). Should we consider
such an assemblage as an atypical form of Acheulean, related to other Acheulean assemblages poor
in large cutting tools, or should we assign it to
another techno-complex partly contemporaneous of
the Acheulean? This is still an open issue.
The Nadung’a 4 assemblage supports the
assumption already stated (Clark and Haynes, 1970)
that handaxes are not systematically associated with
the butchering of large animals. A similar archaeological occurrence is found at the Middle Pleistocene
site of Mwanganda’s Village in Malawi (Clark and
Haynes, 1970), interpreted as a butchery site. It has
yielded an elephant carcass associated with a lithic
assemblage composed predominantly of light cutting tools, with a signiWcant proportion of notched
pieces, and few heavy-duty tools. On the other hand,
large cutting tools are present at Olduvai WK
Hippo CliV (Leakey and Roe, 1994), Gombore II-2
(Chavaillon and Berthelet, 2004), and Olorgesailie
Hippo Banda Site (Isaac, 1977). These sites, subcontemporaneous with Nadung’a 4, are interpreted
as pachyderm butchery sites. It seems, therefore,
that the position connecting the lack of the classical
Acheulean tool-kit in some assemblages with an
activity variant is not tenable, at least for this time
period. Nadung’a 4, just like Mwanganda’s Village
assemblage, raises the question of technological
diversity at the end of the Lower Pleistocene and
beginning of the Middle Pleistocene, irrespective of
the function of the sites.
Elephant carcass processing
The elephant bones were found together with a
few other faunal remains corresponding to rare, isolated fragments attributed to bovids and Wsh (Table
1). Owing to their scarcity and their relation to
diVerent paleoenvironmental contexts (savanna versus aquatic environment), such remains must very
likely be interpreted as background fauna, accumulated before or after the occupation of the site. If
these remains are subtracted from the site’s record,
the faunal assemblage consists exclusively of elephant remains, with a signiWcant portion of undetermined fragments (owing to their alteration) mostly
belonging to an elephant carcass. Only a minority of
recognizably elephant bones (N D 19 out of 49) have
been determined anatomically. The body part distribution (Fig. 12) suggests that they came from a single individual, represented by the skull, a scapula
and few long bones. From their size and lamellae
structure (Fig. 13), they can be attributed to a young
adult Elephas recki. Its incomplete state and the high
ratio of undetermined fragments, is probably due to
combined taphonomical processes, such as the compaction/alteration caused by repeated dry/wet episodes (see above) and recent erosive phases. It can
also be envisaged that the bones were initially scattered and weathered which weakened the bone
before to burial in the vertisol, making it more liable
to post-depositional alterations. The lack of small
parts, e.g., basipodials or metapodials, as well as
461
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Table 1
Nadung’a 4 fauna composition
Taxa
Age
Biotope
Surface (Nb)
Excavation (Nb)
Total (Nb)
Elephas recki
Potamochoerus porcus (M3/)
Medium size Bovid
Aves (marabou size)
Crocodile
Silur (catWsh)
Indet.
Young adult
Young adult
—
adult
—
—
—
Woody savanna
Shrubs
Savanna
—
Aquatic
Aquatic
—
16a
1
4
—
1
—
66b
49
—
3
1
—
2
90
65
1
7
1
1
2
156
TOTAL
—
—
88
145
233
a
b
Small pieces of elephant enamel.
Small splinters.
Fig. 13. Elephant tooth lamellae (M/1?) from Nadung’a 4.
Fig. 12. Anatomical distribution of the determinable elephant
remains (grey colored).
most of the axial elements (vertebrae, ribs) is a striking feature of this assemblage which is likely to be
related to the same taphonomic processes. In any
case, an argument for the intentional transport by
hominids would be impossible to sustain. Some long
bones (e.g., tibia and ulna) were broken, but their
state of preservation does not allow us to determine
whether the fragmentation is anthropic or natural,
i.e., a result of pressures generated within the sediments in relation to their cyclical shrinking and
swelling. Similarly, no cutmarks or blows are
observable on the bone surfaces.
The high degree of alteration of the bones makes
it diYcult to assess whether the elephant died on the
spot or in the immediate vicinity of the site and
whether it was subsequently transported by hominids. The co-occurrence of lithic and elephant
remains in an environment devoid of any other bone
remains, but nevertheless conducive to fossil preservation, argues in favor of the hypothesis that the site
462
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
corresponds to the place where the animal died.
When referring to actualistic data (Crader, 1983;
Haynes, 1991), the body part representation, the age
of the elephant, and the paleoclimatic context (seasonal contrasts with cyclical dry/wet episodes)
makes an assumption of a natural death very plausible, even if there is no way to prove it directly. On
the other hand, geoarchaeological, spatial, and technological data all demonstrate that the hominid
occupation is at least partly linked to the presence of
the elephant carcass. Processing the carcass was very
likely the primary reason for hominids to settle on
the site, importing adequate raw materials from a
short distance away in order to make the necessary
tools. The practice of knapping on the spot is quite
common for this type of site and is documented at
Barogali (Berthelet, 2001; Chavaillon et al., 1987)
and Mwanganda’s Village (Clark and Haynes,
1970). At Nadung’a 4, such a practice is linked to
selective raw material sampling, which is in keeping
with a non-random food procurement strategy.
Discussion
What is unusual at Nadung’a 4 is the large quantity of artefacts produced when compared with the
small quantities recorded in other pachyderm single
carcass sites (Table 2). In those sites, the lithic component was likely almost entirely used for butchering activities. The Nadung’a 4 lithic assemblage
cannot be the result of a single activity, dedicated
exclusively to butchery, owing to the quantity of
unmodiWed Xakes and to the presence of notches
and denticulates, whose function is diYcult to link
directly to meat processing. Considering the unmodiWed Xake component, it is clear that several thousand Xakes were not necessary to process a single
elephant carcass: a few hundred are more than
enough. Notches and denticulates could have been
multipurpose tools, but functional data are very
poor for this category of tools. It is not even clear
whether the working edge of these tools is the
notched one, which is not suitable for cutting soft
materials in any case. However, such tools are quite
frequent in the pachyderm butchery sites; their proportions are signiWcant at Mwanganda’s Village
(Clark and Haynes, 1970) and they are also documented in the European record, e.g., at Torralba in
Spain (Freeman, 1978). They are also found in many
other contexts, however, and are not speciWcally
associated with one type of site.
What we can infer from the data set recovered
from Nadung’a 4 is that:
• important knapping activities were carried out on
the spot, directly around an elephant carcass;
• the artefacts produced were, in all likelyhood,
partly dedicated to butchery and carcass processing: such is probably the case for the heavy-duty
tools and some unmodiWed Xakes;
• a signiWcant portion of the assemblage (notched
tools and a large proportion of unmodiWed
Xakes) was probably devoted to other tasks, e.g.,
the transformation of non-perennial materials
Table 2
Main pachyderms single carcass sites from the Lower and Middle Pleistocene of East-Africa: related archaeological data
Nb of
associated
artefacts
Sites
Chronological
estimation
Excavated
areas (m2)
Koobi Fora FxJj3
(Kenya)
Olduvai FLKN-6
(Tanzania)
Barogali (Djibouti)
»1.9 my
34,5
122
1.9–1.7 my
32
130
1.6–1.3 my
35
569
Nadung’a 4 (Kenya)
Olduvai WK
Hippo CliV
Gombore II-2
(Ethiopia)
Olorgesailie - Hippo
Banda Site
(Kenya)
Mwanganda’s
Village (Malawi)
1.0–0.7 my
»0.7 my
53
16
6797
51
»0.7 my
26
51
< 0.7 my
?
565
»44
314
Middle Pleist.
Pachyderm taxa MNI (aga)
Butchery References
marks
Hexaprotodon
karumensis
Elephas recki
1
absent
Bunn, 1997
1 (juvenile)
present
Elephas recki
ileretensis
Elephas recki
Hippopotamus
gorgops
Hippopotamus
1 (adult)
absent
Bunn, 1981; Potts and
Shipman, 1981
Chavaillon et al., 1987
1 (young adult)
2 (1 ad., 1 juv.)
absent
absent
—
Leakey and Roe, 1994
2
absent
Hippopotamus
gorgops
1
absent
Chavaillon and
Berthelet, 2004
Isaac, 1977
Elephas
1
absent
Clark and Haynes, 1970
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
such as wood or plants, in relation or not with
meat processing and consumption. It is also possible that the meat was transported to another
place (such as a home base), an activity that
would have required the manufacture of implements for food transport.
These features do not match the current model
of a single carcass butchery site or type B site
(Isaac and Crader, 1981). Looking closely at the
archaeological record, however, few sites show the
expected patterns for this site type, either because
of the diversity and quantity of other faunal taxa
present, e.g., Olduvai FLKN-6, (Potts, 1988) and
Koobi Fora FxJj3 (Bunn, 1997), or because of the
absence of associated artefacts when cutmarks are
observable, as at Koobi Fora (Bunn, 1994) and
Buia, (Fiore et al., 2004), or Wnally because of the
important knapping component abandoned on the
spot, e.g., Nadung’a 4. The sites that Wt the deWnition of butchery sites, and are interpreted as such,
never provide direct evidence of human intervention on bones, e.g., Barogali, (Chavaillon et al.,
1987), Gombore II-2 (Chavaillon and Berthelet,
2004), Olduvai WK Hippo CliV (Leakey and Roe,
1994), Olorgesailie Hippo Banda site (Isaac, 1977),
and Mwanganda’s Village (Clark and Haynes,
1970). This lack of butchering evidence could be
the result of bone alteration (e.g., Nadung’a 4) or
weathering, but it could also be related to the
inherent scarcity of cutmarks on processed pachyderm carcasses, as previously pointed out by
Crader (1983).
What is not debatable is the fact that Lower and
Early Middle Pleistocene hominids, in Africa as well
as in Near East and Europe, exploited pachyderms.
How is this part of their food procurement activities
economically structured? This is still an open issue.
No doubt, the situation is more diversiWed and complex than what the type B site model implies. Owing
to the speciWc features (e.g., scarcity of cutmarks)
and constraints related to the exploitation of pachyderm carcasses, multidisciplinary analysis appears
as the most promising approach to deal with this
question.
To conclude, Nadung’a 4 is a unique occurrence
which can be considered a new type of site among
pachyderm single carcass sites. It also opens interesting new perspectives for discussing the presence of
diVerent techno-complexes in East Africa at the end
of the Lower and very beginning of the Middle
Pleistocene.
463
Acknowledgments
We thank the OYce of the President of Kenya
and the board of governors of the National Museums of Kenya for giving permission to conduct
research in West Turkana area. We are grateful to
the Fyssen Foundation as well as to the West Turkana Archaeological Project, co-directed by Hélène
Roche (Mission Préhistorique au Kenya) and Mzalendo Kibunjia (National Museum of Kenya), for
supporting the Weld work carried out at Nadung’a 4.
Such a project would not have been possible without
the precious help on the Weld of a whole team, Wrst
and foremost Boniface Kimeu, Sébastien Manem
and the very professional excavators of the WTAP
(Francis Emekwi, Patrick Kahinju, Bernard
Kimolo, David Massika, Raphael Kioko Massika,
Ali Mwanza, Frederick Mwanza, and Benjamin
Silwa). We are also pleased to thank the reviewer of
this article, the volume editor: Ariane Burke and
Jehanne Féblot-Augustins who helped greatly to
better the english writing, as well as Françoise
Lagarde for her drawing contribution.
References
Abbate, E., Albianelli, A., Azzaroli, A., Benvenuti, M., Tesfamariam, B., Bruni, P., Cipriani, N., Clarke, R.J., Ficcarelli, G.,
Macchiarelli, R., Napoleone, G., Papini, M., Rook, L., Sagri,
M., Tecle, T.M., Torre, D., Villa, I., 1998. A one-million-yearold Homo cranium from the Danakil (Afar) depression of Eritrea. Nature 393, 458–460.
Asfaw, B., Gilbert, W.H., Beyene, Y., Hart, W.K., Renne, P.R.,
Woldegabriel, G., Vrba, E.S., White, T.D., 2002. Remains of
Homo erectus from Bouri, Middle Awash, Ethiopia. Nature
416, 317–320.
Benn, D.I., 1994. Fabric shape and the interpretation of sedimentary fabric data. J. Sediment. Res. 64, 910–915.
Berthelet, A., 2001. L’outillage lithique du site de dépeçage à Elephas recki ileretensis de Barogali (république de Djibouti). C.
R. Acad. Sci. Paris 332, 411–416.
Beyries, S., Boëda, E., 1983. Etude technologique et traces
d’utilisation des “éclats débordants” de Corbehem, Pas-deCalais. Bulletin de la Société Préhistorique Française 80,
275–279.
Blockuis, W.A., Kooistra, M.J., Wilding, L.P., 1990. Micromorphology of cracking clayey soils (Vertisols). In: Douglas, L.A.
(Ed.), Soil Micromorphology : A Basic and Applied Science.
Elsevier, Amsterdam, pp. 123–148.
Bunn, H., 1981. Archaeological evidence for meat-eating by PlioPleistocene hominids from Koobi Fora and Olduvai Gorge.
Nature 291, 574–577.
Bunn, H., 1994. Early Pleistocene hominid foraging strategies
along the ancestral Omo River at Koobi Fora, Kenya. J. Hum.
Evol. 27, 247–266.
Bunn, H., 1997. The bone assemblages from the excavated sites.
In: Isaac, G., Isaac, B. (Eds.), Koobi Fora Research Project,
464
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Plio-Pleistocene Archaeology, vol. 5. Clarendon, Oxford, pp.
402–458.
Chavaillon, J., Berthelet, A., 2004. The archaeological sites of
Melka Kunture. In: Chavaillon, J., Piperno, M. (Eds.), Studies
on the Early Paleolithic site of Melka Kunture, Ethiopia. Istituto Italiano di Preistoria e Protostoria, Florence, pp. 25–82.
Chavaillon, J., Boisaubert, J.L., Faure, M., Guérin, C., Ma, J.L.,
Nickel, B., Poupeau, G., Rey, P., Warsama, S.A., 1987. Le site
de dépeçage pléistocène à Elephas recki de Barogali (République de Djibouti) : nouveaux résultats et datation. C. R. Acad.
Sci. Paris 305, 1259–1266.
Chavaillon, J., Piperno, M. (Eds.), 2004. Studies on the early
Paleolithic site of Melka Kunture, Ethiopia. Istituto italiano
di preistoria e protostoria, Florence.
Clark, G., 1961. World Prehistory : In New Perspective. Cambridge University Press, Cambridge.
Clark, J.D., Haynes, C.V., 1970. An elephant butchery site at
Mwanganda’s Village, Karonga, Malawi, and its relevance for
Palaeolithic archaeology. World Archaeol. 1, 390–411.
Crader, D.C., 1983. Recent single-carcass bone scatters and the
problem of “butchery” sites in the archaeological record. In:
Clutton-Brock, J., Grigson, C. (Eds.), Animals and Archaeology, 1. Hunters and their Prey. BAR, Oxford, pp. 107–141.
Curray, J.R., 1956. The analysis of two-dimensional orientation
data. J. Geol. 64, 117–131.
de Heinzelin, J., Clark, J.D., Schick, K., Gilbert, W.H.E., 2000. The
Acheulean and the Plio-Pleistocene Deposits of the Middle
Awash Valley, Ethiopia. Musée Royal de l’Afrique Centrale,
Tervuren.
Delagnes, A., Roche, H., 2005. Late Pliocene hominid knapping
skills : The case of Lokalalei 2C, West Turkana, Kenya. J.
Hum. Evol. 48, 435–472.
DuYeld, L.F., 1970. Vertisols and their implications for archeological research. Am. Anthropol. 72, 1055–1062.
Eswaran, H., Beinroth, F.H., Reich, P.F., Quandt, L.A. (Eds.),
1999. Vertisols : Their Properties, ClassiWcation, Distribution
and Management. Soil Survey Division. USDA Natural
Resource Conservation Service, Washington.
Eswaran, H., Cook, T., 1988. ClassiWcation and managementrelated properties of vertisols. In: Jutzi, S.C., Haque, I., McIntire, J., Stares, J. (Eds.), Management of Vertisols in Sub-Saharan Africa. International Livestock Centre for Africa, Addis
Ababa, pp. 64–84.
Fiore, I., Bondioli, L., Coppa, A., Macchiarelli, R., Russom, R.,
Kashay, H., Solomon, T., Rook, L., Libsekal, Y., 2004. Taphonomic analysis of the Late Early Pleistocene bone remains from
Buia (Dandiero Basin, Danakil Depression, Eritrea): evidence
for large mammal and reptile butchering. In: Abbate, E., Woldehaimanot, Y., Libsekal, Y., Tecle, T.M., Rook, L. (Eds.), A Step
Towards Human Origins: The BuiaHomo One-Million-Years
Ago in the Eritrean Danakil Depression (East Africa). Dipartimento di Science della Terra, Milano, pp. 89–97.
Freeman, L.G., 1978. The analysis of some occupation Xoors distributions from Earlier and Middle Paleolithic. In: Freeman,
L.G. (Ed.), Views of the Past. Mouton, The Hague, pp. 57–93.
Harris, J.R., Brown, F., Leakey, M.G., 1988. Stratigraphy and
paleontology of pliocene and pleistocene localities west of
lake Tuakana, Kenya. Nat. Hist. Mus. Los Angeles County
Contrib. Sci. 399, 1–128.
Haynes, G., 1991. Mammoths, Mastodonts, and Elephants: Biology, Behavior, and the Fossil Record. Cambridge University
Press, Cambridge.
Isaac, G.L., 1977. Olorgesailie : Archaeological Studies of a Middle Pleistocene Lake Basin in Kenya. The University of Chicago Press, Chicago and London.
Isaac, G.L., Crader, D.C., 1981. To What Extent Were Early
Hominids Carnivorous ? An Archaeological Perspective. In:
Hardinger, R.S.O., Teleki, G. (Eds.), Omnivorous Primates.
Columbia University Press, New York, pp. 37–103.
Kibunjia, M., 1994. Pliocene archaeological occurrences in the
Lake Turkana basin. J. Hum. Evol. 27, 159–171.
Kibunjia, M., Roche, H., Brown, F., Leakey, R.E., 1992. Pliocene
and pleistocene archaeological sites from West of Lake Turkana, Kenya. J. Hum. Evol. 23, 431–438.
Leakey, M.D., Roe, D.A., 1994. Olduvai Gorge: Excavations in
Beds III, IV and the Masek Beds, 1968–1971. Cambridge University Press, Cambridge.
Lenoble, A., 2005. Ruissellement et formation des sites préhistoriques: référentiel actualiste et exemples d’application au fossile.
British Archaeological Report, Oxford.
Lenoble, A., Bertran, P., 2004. Fabric of Palaeolithic levels: methods and implications for site formation processes. J. Archaeol.
Sci. 31, 457–469.
McBrearty, S., Brooks, A., 2000. The revolution that wasn’t: a
new interpretation of the origin of modern human behavior. J.
Hum. Evol. 39, 453–563.
Peresani, M. (Ed.), 2003. Discoid Lithic Technology. Archaeopress, Oxford.
Petraglia, M.D., Potts, R., 1994. Water Xow and the formation of
early pleistocene artifacts sites in Olduvai Gorge, Tanzania. J.
Anthropol. Archaeol. 13, 228–254.
Potts, R., 1988. Early Hominid Activities at Olduvai. Aldine de
Gruyter, New-York.
Potts, R., Behrensmeyer, A.K., Deino, A., DitchWeld, P., Clark, J.,
2004. Small mid-Pleistocene hominin associated with East
Acheulean technology. Science 305, 75–78.
Potts, R., Shipman, P., 1981. Cutmarks made by stone tools on
bones from Olduvai Gorge, Tanzania. Nature 291,
577–580.
Prat, S., Brugal, J.P., Roche, H., Texier, P.J., 2003. Nouvelles
découvertes de dents d’hominidés dans le membre Kaitio de la
formation de Nachukui (1.65–1.9 Ma) Ouest du lac Turkana.
C. R. Palevol 2, 685–693.
Reading, H.G., 1996. Sedimentary environments: processes, facies
and stratigraphy, third ed. Blackwell ScientiWc Publications,
Oxford.
Reineck, H.E., Singh, I.B., 1980. Depositional Sedimentary Environments. Springer-Verlag, Berlin.
Roche, H., Brugal, J.P., Delagnes, A., Feibel, C., Harmand, S.,
Kibunjia, M., Prat, S., Texier, P.J., 2003. Les sites archéologiques plio-pléistocènes de la formation de Nachukui, OuestTurkana, Kenya : bilan synthétique 1997–2001. C.R. Palévol
2, 663–673.
Roche, H., Delagnes, A., Brugal, J.P., Feibel, C., Kibunjia, M.,
Moure, V., Texier, P.J., 1999. Early hominid stone tool production and technical skill 2.34 Myr ago in West Turkana,
Kenya. Nature 399, 57–60.
Roche, H., Kibunjia, M., 1994. Les sites archéologiques Plio-Pléistocènes de la formation de Nachukui, West Turkana, Kenya.
C. R. Acad. Sci. Paris, série II, 1145–1151.
Roche, H., Kibunjia, M., 1996. Contribution of the West Turkana
sites to the Archaeology of the Lower Omo/Turkana Basin.
Paper presented at Four millions years of hominid evolution
in East Africa: an international conference in honour of Dr.
A. Delagnes et al. / Journal of Anthropological Archaeology 25 (2006) 448–465
Mary Leakey’s outstanding contribution in paleoanthropology, Arusha, Tanzanie.
Roche, H., Kibunjia, M., Brown, F., Leakey, R., 1992. Pliocene
and pleistocene sites from West of Lake Turkana, Kenya. J.
Hum. Evol. 23, 432–438.
Roche, H., Texier, P.J., 1996. Evaluation of technical competence
of Homo erectus in East Africa during the Middle Pleistocene.
In: Bower, J.R.F., Sartono, S. (Eds.), Evolution and Ecology
of Homo erectus. Royal Netherlands Academy of Arts and
Science, Leiden, pp. 153–167.
Schick, K., 1986. Stone Age in the making: experiments in the formation and transformation of archaeological occurrences.
British Archaeological Reports, Oxford.
Schick, K., Clark, J.D., 2003. Biface technological development and variability in the Acheulean industrial complex
in the Middle Awash Region of the Afar Rift, Ethiopia. In:
Soressi, M., Dibble, H. (Eds.), Multiple Approaches to the
Study of Bifacial Technologies. University of Pennsylvania
Museum of Archaeology and Anthropology, Philadelphia,
pp. 1–30.
Soil Survey StuV, 1999. Soil taxonomy—A basic System of Soil
ClassiWcation for Making and Interpreting Soil Survey.
United States Department of Agriculture, Washington.
465
Texier, P.J., 1995. The Oldowan assemblage from NY 18 site at
Nyabusosi (Toro-Uganda). Compte rendu de l’Académie des
Sciences 320, 647–653.
Texier, P.J., Roche, H., 1995. The impact of predetermination on
the development of some acheulean chaˆnes opératoires. In:
Bermudez de Castro, J.M. (Ed.), Human evolution in Europe
and the Atapuerca evidence. Junta de Castilla y Leon, Valladolid, pp. 403–420.
Vallin, L., Masson, B., Caspar, J.P., 2001. Taphonomy at Hermies,
France: a mousterian knapping site in a Loessic context. J.
Field Archaeol. 28, 419–436.
Villa, P., 1990. Torralba and Aridos: elephant exploitation in
Middle Pleistocene Spain. J. Hum. Evol. 19, 299–309.
Villa, P., Soto, E., Santonja, M., Pérez-Gonzàlez, A., Mora, R.,
Parcerisas, J., Sesé, C., 2004. New data from Ambrona: closing
the hunting versus scavenging debate. Quatern. Int. 126–128,
223–250.
Wood, W.R., Johnson, D.L., 1978. A Survey of Disturbance Processes in Archaeological Site Formation. Adv. Archaeol.
Method Theory 1, 315–381.
Wynn, J.G., 2001. Paleosols, stable carbon isotopes, and paleoenvironments of Hominid evolution in the neogene Turkana
basin, Northern Kenya. PhD Thesis, University of Oregon.