The first use of bone tools: a reappraisal of the evidence
from Olduvai Gorge, Tanzania

Lucinda R. Backwell1* & Francesco d’Errico2

1Institute for Human Evolution, School of Geosciences, University of the Witwatersrand,
Private Bag 3, Wits, 2050, Johannesburg, South Africa

2UMR 5199 du CNRS, Institut de Préhistoire et de Géologie du Quaternaire, Avenue des Facultés, 33405, Talence, France, and
Department of Anthropology, The George Washington University, 2110 G Street NW, Washington D.C. 20052, U.S.A.

Received 16 March 2004. Accepted 10 September 2004

INTRODUCTION
Many bone, antler, and ivory tools are reported from

Lower and Middle Palaeolithic sites in Africa and Europe
(Schmidtgen 1929; Bastin 1932; Breuil 1932, 1938; Koby
1943; Dart 1957; Kitching 1963; Breuil & Barral 1955;
Bonifay 1974, 1986; Freeman 1978, 1983; Cahen et al. 1979;
Howell & Freeman 1983; Howell et al. 1995; Gaudzinski
1999); for a review of the evidence see Henshilwood &
Sealy (1997), Villa & d’Errico (1998, 2001), Henshilwood
et al. (2002) and d’Errico & Backwell (2003). These claims
have, however, been repeatedly called into question.
Studies demonstrating that a number of natural processes
occurring during the life of an animal or after its death
can produce pseudo-tools that have been, or may be,
misidentified as intentionally modified or used bones.
Pre-mortem phenomena that produce pseudo-tools or
pseudo-anthropic use-wear include the remodelling of
the bone structure (d’Errico 1996), vascular grooves
(Shipman & Rose 1984; d’Errico & Villa 1997), teeth use-
wear (Gautier 1986), breakage and wear of deer antler
(Olsen 1989) and elephant tusk tips (Haynes 1991; Villa &
d’Errico 2001). Post-mortem processes are more numerous
and include gnawing or digestion by carnivores, rodents
and herbivores (Pei 1938; Sutcliffe 1970, 1973, 1977;
Binford 1981; Haynes 1983; Villa & Bartram 1996), fracture
for marrow consumption by hominids and carnivores
(Bunn 1981; Gifford-Gonzalez 1989, 1991), trampling
(d’Errico et al. 1984; Haynes 1988, 1991) root etching
(Binford 1981; Andrews 1990), weathering (Brain 1967)
and the action of different sedimentary environments
(Brain 1981; Lyman 1984, 1994). As suggested by these and

other authors (Shipman 1988; Shipman & Rose 1988;
Bonnichsen & Sorg 1989; Villa et al. 1999), in order to
distinguish between pseudo-tools and true tools, it is
necessary to adopt an interdisciplinary approach, com-
bining taphonomic analysis of the associated fossil assem-
blages, microscopic studies of possible traces of manufac-
ture and use, and the experimental replication of the pur-
ported tools. It is by applying this approach, for example,
that Dart’s (1957) theory for an early hominid ‘Osteo-
dontokeratic’ culture has strongly been challenged and
largely refuted (Klein 1975; Shipman & Phillips 1976;
Maguire et al. 1980; Brain 1981). Dart’s hypothesis created
the conditions for a receptive environment, one in which
potentially used or manufactured bone could be recog-
nized and its designation as an artefact tested, using more
reliable frames of inference.

The South African evidence
Building on this premise, Robinson and Brain in South

Africa, and Mary Leakey in East Africa, proposed again
that early hominids used bone tools. In 1959 Robinson
published a single bone tool from Sterkfontein Member 5
West (c. 1.7–1.4 Mya) consisting of a pointed metapodial
shaft fragment with evidence of use on the tip. In the
course of 24 years of excavation at Swartkrans, Brain
(Brain et al. 1988; Brain 1989; Brain & Shipman 1993)
identified 68 bones, bovid horn cores and one equid
mandible from Members 1–3 (c. 1.8–1 Mya) bearing similar
modifications. Comparative microscopic analysis of the
wear pattern on the smoothed tips of these bones, and on
modern shaft fragments used experimentally to dig up
tubers and work skins, suggested to Brain and Shipman

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 95

Purported early hominid bone tools from Olduvai Gorge are studied for microscopic traces of use-wear, and evidence of intentional
flaking by knapping. Comparative microscopic analyses of the edges of the purported tools, and areas far from the potential functional
zone, as well as edges of bone pieces from the remainder of the assemblage, show that possible modifications due to utilization are not
distinguishable from features attributed to post-depositional abrasion. Taphonomic analysis of the bone tool collection, a control
sample of bone shaft fragments from the remainder of the Olduvai assemblage, and experimentally broken elephant long bones,
identifies significant differences in the size and type of mammals represented. The bone tool collection records an abundance of large to
very large mammals, while the control sample comprises mostly medium-size bovids. Puncture and cut-marks occur on one third of the
bone tool collection, and on only a few pieces in the control sample, suggesting hominids were the agent responsible for the breakage of
most of the bones previously described as tools. Analysis of the number, location and length of flake scars in the three assemblages,
reveals that a reduced proportion of purported bone tools bear invasive, contiguous, often bifacially arranged removals, not seen in the
control or experimental collections. This makes these specimens good candidates for having been shaped and used by early hominids.
Complete bones with tool-generated puncture-marks, previously interpreted as anvils, are interpreted here as hammers used on
intermediate stone tools.

Keywords: Olduvai Gorge, early hominid, bone tools.

*Author for correspondence. E-mail: backwell@science.pg.wits.ac.za



that the surface modifications were not natural, and
that the activities they tested experimentally were
indeed those in which the Swartkrans tools were in-
volved.

A recent reappraisal of this material confirmed the
anthropic origin of the use-wear (Backwell 2000; Backwell
& d’Errico 2001; d’Errico et al. 2001). Comparison between
the Swartkrans tool wear pattern and that on bones from
35 reference collections, consisting of fauna modified by
10 non-human agents, identified no natural counterpart
for the Swartkrans modifications. These authors also
showed that the wear on the bone tools does not represent
an extreme in variation of a taphonomic process affecting
to a lesser degree the rest of the assemblage. In addition,
analysis of the breakage patterns and size of the bone tools
from this site, compared with the remainder of the faunal
remains, indicated that early hominids selected heavily
weathered, elongated and robust bone fragments for use
as tools.

Quantification of striation width and orientation com-
prising the wear pattern suggested that these tools were
not used to extract tubers or work skins. The wear pattern
more closely fits that created experimentally when bone is
used to excavate in a fine-grained sedimentary environ-
ment, such as that found in the pre-sorted sediment
constituting termite mounds present in the Sterkfontein
area. This led them to propose that the main, if not exclu-
sive function, of the Sterkfontein and Swartkrans bone
tools, and of the similar 23 undescribed specimens from
Drimolen (c. 2–1.5 Mya) (Keyser 2000), was that of extract-
ing termites. In another paper, Backwell & d’Errico (2003)
report 16 additional bone tools from Swartkrans and
show that there are no significant differences between
Members in the type and size of the bone fragments used
as tools, as well as in the length and type of the wear
pattern, indicating that no major changes occurred
through time in the subsistence strategy for which the
tools were used. Previously unrecognized evidence of
intentional shaping through grinding is also identified by
d’Errico & Backwell (2003) on the tips of six horn cores and
an ulna, indicating that southern African early hominids
had the cognitive abilities to modify the functional area of
bone implements with a technique specific to bone mate-
rial, in order to achieve optimal efficiency in digging activ-
ities. No firm evidence exists on who used these bone
tools. Brain (Brain et al. 1988: 835) and Susman (1991, 1994)
suggest they were used by both early humans and robust
australopithecines. Backwell & d’Errico (2003) consider
instead the robust australopithecines as the more probable
modifiers and users of these tools. The reasons they put
forward in support of this scenario are the absence
of Homo remains in Swartkrans Member 3 – where
Paranthropus (Australopithecus) robustus fossils occur in
association with relatively few stone and many bone
tools, together with the virtual absence of diagnostic stone
tools at Drimolen – a site dominated by robust austra-
lopithecine remains, and a substantial collection of similar
bone tools. In addition, no such bone tools are found at
South African sites postdating 1 Mya, the time of the
robust australopithecine extinction.

The East African evidence
Mary Leakey (1971) reports 125 artificially modified

bones and teeth from Olduvai Beds I and II bearing
evidence of intentional flaking, battering and abrasion.
These specimens derive from massive elephant, giraffe
and Libytherium limb bones, and to a lesser extent from
equids and bovids, as well as from hippopotamus and
suid canines. In a comprehensive reappraisal of this mate-
rial, Shipman (1989) correctly points out that Leakey’s
identification of Olduvai bone tools was not based on
explicit criteria, and lacked analogies that would allow the
ruling out of alternative interpretations.

In her reappraisal of the Olduvai material, Shipman
(1984, 1989) uses a control sample consisting of scanning
electron microscope-analysed resin replicas of bones
submitted to a number of natural phenomena (e.g. weath-
ering, chewing, licking, digestion, wind), and experimen-
tal or ethnographic bone tools used for butchering,
digging, grinding, and hide and meat processing.

Microscopic analysis of these collections provided criteria
(Shipman & Phillips-Conroy 1977; Shipman et al. 1984;
Shipman & Rose 1988) to identify the material on which
bone tools were used (hide, meat, soft vegetables), the
kinesis and function (digging, bark-working, grinding
hard grains, butchering), and the duration (brief, moderate,
extensive) for which they were used. Shipman’s ability
to distinguish between unused and used bones, and to
identify their main function, was verified through blind
tests. The control sample also includes experimental
reproduction of wind abrasion through the use of an abra-
sion gun driven by pressurized air. Sedimentary abrasion
was mimicked using a tumbling barrel with different
types of sediments, with and without the addition of
water. According to Shipman (Table 1), utilization pro-
duces differential wear between functional and non-
functional zones of the tool, and at a microscopic scale, be-
tween more exposed and recessed/concave areas, while
aeolian and sedimentary abrasion with no water creates a
pitted or pebbly texture, homogeneously altering the
entire surface. Pits caused by striking harder particles may
occur on areas worn by utilization, but they are irregularly
spaced and sized. Also, experimental abrasion only rarely
creates scratches, while utilization on mixed substances
produces a glassy polish crossed by striations. Shipman
stresses, however, that these criteria are provisional and
that further experimental studies of abrasion are needed
to firmly identify distinctive features.

Application of these criteria to 116 of the 125 pieces
described by Leakey – teeth were excluded from
Shipman’s analysis – led her to conclude that 41 were
utilized by hominids and the remainder bore ambiguous
traces, or evidence of abrasion by sediment. Four of the
tools bearing punctures – a patella, astragalus, femoral
condyle and magnum – are interpreted as anvils due to
the triangular or diamond shape of the impressions,
which are different from those produced by carnivores;
the absence of counter-bites; large size of the bones
difficult to bite; location of the marks consistent with their
proposed use, and their apparent antiquity. Shipman,
following Leakey, proposes that the marks on these tools

96 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158



may have been produced by stone awls, found at the
same localities, used to pierce leather/hide.

Among the remaining 37 specimens diagnosed as imple-
ments, 35 are described as bones broken and shaped by
flaking prior to use. Twenty-six are interpreted as light-duty
implements used on soft substances (hide-working), and
the remaining 11 described as heavy-duty tools utilized
on mixed substances, perhaps in butchering or digging
activities. According to Shipman, wear patterns cannot be
confused with sedimentary abrasion or weathering, since
bone tools show, with the exception of three cases, a low
degree of natural alteration. Variables such as taxon, body
part, breakage (location, orientation, type and number)
and type of surface alteration (weathering, abrasion) were
recorded by Shipman on the 41 tools and on 350 randomly
selected bones from Olduvai and a few other sites.
Comparison of these parameters indicated that the bone
tools had a significantly higher occurrence of flaked frac-
tures, flake scars and punctures, and a lower presence of
stepped, jagged, or smooth fractures, suggesting that the
bone tools were broken shortly after the death of the
animal. It also showed that humeri, scapulae and femora,
particularly from giraffids and elephants – relatively rare
taxa at Olduvai – are over-represented among the bone
tools.

Objectives
In sum, South and East African early hominid sites dated

to between 1.8–1 Mya have yielded what appear to be
very different types of bone tools. The former are charac-
terized by long bone shaft fragments and horn cores of
medium- to large-sized bovids, collected after weather-
ing, and possibly used in specialized digging activities.
Marginal shaping by grinding occasionally involves
robust horn core tips. Those from East Africa mainly
consist of freshly broken or, more rarely, complete irregu-
lar bones from very large mammals, used as such, or
modified by flaking. Irregular bones or epiphyses appear
to have been used as hammers, while the others were
apparently involved in a variety of light- and heavy-duty
activities. What are the reasons for such differences? Were
these bones used by the same or by different hominid
species, if not taxa? If the first applies, do they reflect
different cultural traditions? One may expect, if this is the
case, to find additional differences between these two
regions in other aspects of material culture and adapta-
tion. Although the Oldowan is associated with sites from
both regions, this lithic technology appears to occur in
East Africa at least more than half a million years earlier
than in South Africa (Kibunjia 1994; Kuman 1994, 2003;
Semaw et al. 1997; Kuman & Clarke 2000). This gap may be
due to a time lag in the diffusion of this behaviour,
staggered independent invention, or a scarcity of late
Pliocene deposits in South Africa. Since few studies
(Petraglia & Korisettar 1998) have tried to address this
question by a detailed comparative technological analysis
of contemporaneous lithic assemblages, as currently
conducted by Roche’s team on East African sites (Roche
et al. 1999), it is difficult at present to know whether what is
generally called Oldowan in these two regions corresponds

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 97

Ta
bl

e
1.

Su
m

m
ar

y
of

th
e

cr
ite

ri
a

pr
op

os
ed

by
Sh

ip
m

an
an

d
R

os
e

(1
98

8)
an

d
Sh

ip
m

an
(1

98
9)

to
di

st
in

gu
is

h
na

tu
ra

la
br

as
io

n
fr

om
ut

ili
za

tio
n

an
d

id
en

tif
y

th
e

ta
sk

fo
r

w
hi

ch
bo

ne
to

ol
s

w
er

e
us

ed
.

U
til

iz
at

io
n

Ex
pe

ri
m

en
ta

la
br

as
io

n

So
ft

M
ix

ed
an

d
ha

rd
su

bs
ta

nc
es

A
eo

lia
n

Se
di

m
en

tf
lo

w
H

yd
ra

ul
ic

Ta
sk

H
id

e
w

or
ki

ng
C

ut
tin

g
m

ea
t

Ve
ge

ta
bl

e
pr

oc
es

si
ng

D
ig

gi
ng

so
il

Ba
rk

-w
or

ki
ng

G
ri

nd
in

g
ha

rd
gr

ai
ns

Bu
tc

he
ry

A
br

as
io

n
gu

n
Tu

m
bl

in
g

ba
rr

el
w

ith
se

di
m

en
t(

fr
om

lo
es

s
to

gr
av

el
)

Tu
m

bl
in

g
ba

rr
el

w
ith

se
di

m
en

t
(f

ro
m

lo
es

s
to

gr
av

el
)a

nd
w

at
er

Lo
ca

tio
n

Ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

R
ai

se
d

ar
ea

s
of

th
e

ed
ge

A
ll

ov
er

A
ll

ov
er

A
ll

ov
er

Ed
ge

sh
ap

e
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th
R

ou
nd

an
d

sm
oo

th

Po
lis

h
Fi

ne
gl

as
sy

po
lis

h
Po

lis
h

Fi
ne

gl
as

sy
po

lis
h

Fi
ne

gl
as

sy
po

lis
h

Fi
ne

gl
as

sy
po

lis
h

Fi
ne

gl
as

sy
po

lis
h

Fi
ne

gl
as

sy
po

lis
h

A
bs

en
t

A
bs

en
t

A
bs

en
t

St
ri

at
io

ns
Fe

w
or

no
ne

Fe
w

or
no

ne
Fi

ne
an

d
co

ar
se

st
ri

at
io

ns
Fi

ne
an

d
co

ar
se

st
ri

at
io

ns
Fi

ne
an

d
co

ar
se

st
ri

at
io

ns
Fi

ne
an

d
co

ar
se

st
ri

at
io

ns
Fi

ne
an

d
co

ar
se

st
ri

at
io

ns
A

bs
en

t
Fe

w
Fe

w
or

no
ne

Pi
tt

in
g

Fe
w

or
no

ne
Fe

w
or

no
ne

Ir
re

gu
la

rl
y

sp
ac

ed
an

d
si

ze
d

pi
ts

Ir
re

gu
la

rl
y

sp
ac

ed
an

d
si

ze
d

pi
ts

Ir
re

gu
la

rl
y

sp
ac

ed
an

d
si

ze
d

pi
ts

Ir
re

gu
la

rl
y

sp
ac

ed
an

d
si

ze
d

pi
ts

Ir
re

gu
la

rl
y

sp
ac

ed
an

d
si

ze
d

pi
ts

Ev
en

ly
pi

tt
ed

su
rf

ac
e

Ev
en

ly
an

d
he

av
ily

pi
tt

ed
su

rf
ac

e
A

bs
en

t



to a single cultural tradition, or the expression of distinct
regional trends. In addition, exactly who was responsible
for the Oldowan technology is still a matter of debate.
Since the identification of Homo habilis (Leakey et al. 1964;
Tobias 1965), it has fallen into common usage to consider
this species as the more probable maker of the Oldowan
tools. The hypothesis of a robust australopithecine
authorship, though marginal, has not been abandoned
(Brain 1993; Susman 1994). A date of 2.6 Mya for the oldest
occurrence of stone tools in the Afar, Ethiopia, slightly pre-
dating the oldest evidence for early Homo in East Africa
(2.4–2.3 Mya) (Suwa et al. 1996; Deino & Hill 2002 in
Semaw et al. 2003) brings new interest to the subject, with
Australopithecus garhi proposed as the best candidate for
the first user of stone tools in this region. An attribution to
Australopithecus is further suggested by the relatively
sophisticated stone tool technology and raw material
procurement strategies recorded at the Gona sites. One
may assume that to reach the advanced stage of technical
and gestural competence recorded at Gona, the makers of
the stone tool assemblages had already established a
history of stone working. This may already have been in
place by 2.9 and 2.7 Mya, a period poorly represented in
the sections exposed at Gona, and for which no evidence
of early Homo exists.

In this ongoing debate, bone tools have not received the
attention they deserve. Variability in bone tool manufac-
ture and use may provide a means independent of lithic
technology to address crucial issues such as the character-
ization of early hominid cultural traditions. However, the
artefactual nature of Lower Palaeolithic bone tools and
the reality of the associated behaviours identified must be
verified before we use this evidence to create scenarios of
early hominid cultural evolution and adaptation. In this
respect, the evidence for bone tool use is quite different
from these two regions. Bone tools from South Africa are
documented at a number of cave sites and may now be
regarded as unquestionably utilized, if not modified, by
hominids. Those from East Africa are attested only at
Olduvai Gorge Beds I–II, in spite of the numerous sites
excavated in the region, and their identification is based
on results that may be preliminary. There are various
reasons for this uncertainty. The first problem stems from
the frame of inferences used by Shipman to assess the
artefactual nature of the Olduvai bone tools. Although
criteria are provided to distinguish between experimen-
tally-used and abraded bone, it is uncertain whether such
experiments successfully reproduce the entire range of
post-depositional phenomena that may have affected the
Olduvai bone assemblage, and whether the actual site
formation processes that occurred there produced modi-
fications that may closely mimic experimental traces of
use, and be the source of misinterpretation. This is more
so, considering that other studies postdating Shipman’s
research on Olduvai, including her own work (Olsen &
Shipman 1988), have expanded our knowledge of natural
modifications (e.g. Marshall 1989; Haynes 1991, Dechant
Boaz 1994, Backwell 2000) and produced results that in
some instances challenge her criteria. It has been shown,
for example, that tumbling individual bones or bone

objects with sand in leather bags produces a fairly large
number of striations (d’Errico 1993a). The absence of
striations on the bones tumbled by Shipman is most likely
due to the presence of water during the experiment.
However, no proof exists that water was a constant feature
during the deposition of Beds I and II faunal assemblages
(Potts 1988). Therefore, the presence of striations associated
with polish on the edges of the purported bone tools does
not necessarily result from use. In contrast to Shipman’s
proposition that digging produces a fine glassy polish,
experimental reproduction of this task has produced a
wear pattern dominated by individual striations (Backwell
& d’Errico 2001) associated with a smoothing of the active
zone, but not producing a glassy polish. While scanning
electron microscopy (SEM) provides a useful means to
study microscopic bone surface modification, intimate
use of this tool shows that strong lateral variations often
occur on adjacent areas, making a true documentation of
the entire appearance of the inspected surface a challenging
endeavour. This may be overcome by quantifying the
surface features (e.g. Backwell & d’Errico 1999, 2001; see
González-Urquijo & Ibáñez-Estévez 2003) on bone tools
and/or increasing the number of micrographs presented
to illustrate the variability of the surface features. To date,
the microscopic evidence that documents the use-wear on
the 41 bone tools from Olduvai consists of only 3 SEM
micrographs. No documentation is presented on the
appearance of the surface of the bone tools away from the
area interpreted as utilized, nor on bones from the
remainder of the Olduvai assemblage from where the
bone tools derive, which prevents control comparison
between clearly natural and purportedly anthropic modi-
fications. Recognition of tools on the basis of use-wear
alone may be misleading because tools may have been
shaped for a number of reasons, and subsequently not, or
only marginally, used. Identification of tools based on
use-wear alone may thus result in the discarding of a
number of true tools. Use-wear results should therefore
be crossed with analyses of possible evidence of inten-
tional shaping. Also problematic is the relationship
between specific tools and the tasks Shipman assigns to
them. Perhaps because of the preliminary nature of the
work she published on the Olduvai material, Shipman
only provides percentages of tools used for different
functions and duration, without specifying which tool
was used for what. Correlating tools with tasks is crucial to
evaluating whether tools of particular morphology,
weight, degree of shaping, and size, i.e particular types,
were used for specific tasks. This would also enable
comparison between knapped bone and stone tools, and
the evaluation of degrees of gestural competence in the
working of different raw materials.

Although tools are identified on the basis of their wear
and not on evidence of manufacture, all but two are
described as clearly broken and shaped prior to use. How-
ever, criteria to identify bone tools shaped by flaking are
unclear. Experimental flaking of large- and medium-sized
mammal bones, either to produce blanks, or to shape core
tools, has shown this technique may be used with some
success on bone, though it is conditioned by the unisotropic

98 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158



nature of this material (Bonnichsen 1979; Bonnichsen &
Will 1980; Stanford et al. 1981; Walker 1999). The use of this
technique at Lower Palaeolithic sites is demonstrated by
the discovery of bone hand axes made of Elephas antiquus
limb bones at three Italian sites (Biddittu et al. 1979;
Biddittu & Segre 1982a, b; Biddittu & Bruni 1987). How-
ever, pseudo flaked bone tools may be produced by
anthropic processes, such as bone breakage for marrow
extraction (Peretto et al. 1996), or non-human modifica-
tion, such as tramping by animals (Haynes 1988; 1991) and
gnawing by large carnivores (Binford 1981; Villa &
Bartram 1996). In spite of valuable work conducted in the
last decade to identify firm criteria for distinguishing
between individual percussion marks and carnivore
notches (Blumenschine & Selvaggio 1988; Capaldo &
Blumenschine 1994; Blumenschine 1995; Blumenschine
et al. 1996; Capaldo 1998; Selvaggio 1998), the identifica-
tion of bone tools shaped by flaking, especially those
bearing a low level of modification and no compelling tool
morphology, remains a matter of debate. This uncertainty
affects a number of Lower and Middle Palaeolithic sites
from Europe such as Morin (Freeman 1978, 1983) La
Polledrara (Villa et al. 1999), Casal dei Pazzi (Anzidei et al.
1999), Castel di Guido (Campetti et al. 1989; Radmilli &
Boschian 1996), Vaufrey (Vincent 1993), Torralba (Aguirre
1986), Bilzingsleben (Mania & Weber 1986; Mania 1990,
1995), Rhede (Tromnau 1983), Kulna (Mania 1990) and the
Vallonnet (d’Errico 1988a). It also applies to Palaeo-Indian
sites, such as Lange-Ferguson (Hannus 1990, 1997), where
mammoth epiphiseal fragments and bone flakes have
been interpreted, based on studies of traces of knapping
and use-wear conducted by Shipman, as tools used by
Clovis hunters to butcher mammoth carcasses.

In addition, Leakey and Shipman do not provide a
complete list of the bones identified by the former as tools,
nor give a complete representation of the bones inter-
preted as tools. A photograph of one aspect is given for
some specimens, while others are represented by line-
drawings. This prevents independent evaluation of the
basic features characterizing these objects.

In this paper we provide a complete photographic
record of this collection and reassess both Leakey’s and
Shipman’s arguments for these being tools, using a
multiple approach study based on data provided by
microscopic, taphonomic, and morphometric analysis of
the purported bone tools, faunal material from Olduvai,
and experimentally and naturally modified bone.

MATERIALS AND METHODS

Contextual information
Olduvai Gorge is probably the most famous ensemble of

Palaeolithic sites in the world and certainly the area which
provides the most continuous record of human presence
during the past two thousand millennia. Located in the
eastern Serengeti Plains of northern Tanzania over an
area that measures about 30 miles in length (Fig. 1) , the
sites within the Gorge date from 2.1 My to 15 kya.
Geologically, the formation is divided into seven main
beds or levels (Hay 1976; Potts 1988): Bed I (about 2.1–

1.7 My), Bed II (1.7–1.15 My), Bed III (1.15–0.8 My), Bed IV
(0.8–0.6 My), the Masek Beds (0.6–0.4 My), the Ndutu
Beds (0.4–32 ky), and the Naisiusiu Beds (22–15 ky).
Radiometric dating of the tuff layers has clarified the age
of the various levels within each bed. Comprehensive
geological and palaeoenvironmental analyses (Hay 1976;
Bonnefille & Riollet 1984; Cerling 1986; Kappelman 1984)
have helped greatly in reconstructing the geomorpho-
logical, taphonomic and palaeoclimatic history of the
Gorge. Table 2 summarizes available data on the location,
stratigraphic provenance, age, associated hominid remains,
stone tools, taphonomy, site function and number of
faunal remains at sites yielding purported bone tools
according to Leakey (1971), Hay (1976) Shipman (1989)
and this study.

Mary Leakey defined three industries at Olduvai. She
called these the ‘Oldowan’, ‘Developed Oldowan’ and
‘Early Acheulean’. The Developed Oldowan was further
subdivided into Developed Oldowan A, and Developed
Oldowan B.

Stone tools from Bed I and the base of Bed II, attributed
to the Oldowan, include side, end, two-edged, pointed,
and chisel-edged choppers, polyhedrons, discoids, scrapers,
a few subspheroids and burins. Hammers, utilized
cobbles and flakes, some of them retouched, probably
used in light-duty functions are also present. The
Developed Oldowan A, found at sites from lower Bed II,
differs from the Oldowan for an increase in spheroids and
subspheroids, interpreted as the introduction of missiles
as hunting weapons; light-duty tools are more varied. The
Developed Oldowan B, from Middle and Upper Bed II
contains very few bifaces. Although bifaces are absent in
Bed I, ‘proto-bifaces’ appear in upper Bed I and Lower
Bed II, picks are discovered above the base of Bed II.

Crude choppers and scrapers occur throughout Beds I
and II, and spheroids and sub-spheroids, modified and
battered nodules and blocks increase in frequency in
Bed II. This corresponds to a rise in the number of artefacts
relative to fauna in middle-upper Bed II. This may be due
to better tool-making abilities and accessibility of raw
materials, or to large mammals being common at these
sites. Fewer large animals are needed to subsist, and fewer
bones are recorded at sites if meat is transported (Leakey
1971).

The beginning of the Acheulean is marked by the
appearance of bifaces with cleavers and hand axes, which
appeared in Bed II. Compared to the Acheulean, the
Developed Oldowan tools evidence greater variability
and seem to differ technologically from the more recent
tradition. Indications are that these two traditions coex-
isted. Discovery of human remains attributed to Homo
erectus in association with hand axes in Bed II suggested to
Mary Leakey that this human type and early forms of
Homo sapiens were the makers of the Acheulean.

Early hominid behaviour at Olduvai
Olduvai assemblages have represented for the last four

decades an arena that has challenged hypotheses on early
hominid behaviour and subsistence strategies. Dense
concentrations of animal bones and stone tools from Bed I

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 99



and II sites were in the early phases of Olduvai investiga-
tions interpreted as living sites, home bases or central
foraging places, where hominids processed the meat of
animals using stone tools (Leakey 1971; Isaac 1971, 1978,
1983). In the late 1970s the campsite interpretation, based
on analogy with behaviours recorded among modern
hunter-gatherers, became the object of thorough scrutiny.
Binford (1981) proposed that sites from Beds I and II
simply represented zones where dead animals were
scavenged by carnivores and hominids. Analysis of six
levels from Bed I (FLK North Level 6, FLK Zinjanthropus,
FLKNN Levels 2 and 3, DK Levels 2 and 3) led Potts

(1988) to reject Binford’s interpretation and propose that
although the attraction of carnivores to these sites prohib-
ited their use by hominids as home bases, bone remains
from these sites should be interpreted as hominid accu-
mulations of carcasses obtained by scavenging/hunting,
and stone tools as ‘stone caches’ repeatedly used to
process carcasses and possibly for other activities.
Hunting is discarded by Shipman (1986a) based on her
study of the occurrence and location of cut-marks on 2700
specimens from 10 Bed I contexts (DK I, FLK Zinj. and
other levels, FLKN 1–6, FLKNN, PDK) and comparison
with modern butchery sites from Kenya.

100 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 1. a, Regional topographic map of northern Tanzania, showing extent of Olduvai area in (b). b, Outline map of Olduvai Main and Side gorges
showing faults, topographic features and localities. Faults are shown as heavy hachures. Roads are shown as dashed lines. c, Map showing localities
near the junction of the Main and Side gorges (modified after Hay 1976). Encircled are sites that have yielded bone tools according to Leakey (1971).



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 101

Ta
bl

e
2.

C
on

te
xt

ua
li

nf
or

m
at

io
n

on
O

ld
uv

ai
si

te
s

th
at

ha
ve

yi
el

de
d

im
po

rt
an

ta
rc

he
ol

og
ic

al
fe

at
ur

es
an

d
pu

ta
tiv

e
bo

ne
to

ol
s.

D
at

a
af

te
r

Le
ak

ey
(1

97
1)

,H
ay

(1
97

6)
an

d
Po

tt
s

(1
98

8)
.

Lo
ca

lit
y

A
ge

ra
ng

e
H

om
in

id
s

Lo
c.

C
ul

tu
ra

l
St

on
e

to
ol

s
Bo

ne
to

ol
s

C
ut

-m
ar

ks
Si

te
fu

nc
tio

n
N

o.
st

on
e

N
o.

fa
un

al
at

tr
ib

ut
io

n
to

ol
s

re
m

ai
ns

Be
d

Si
te

ch
pb

bf
pl

ds
sp

no
sh

sl
bu

aw
oe

an
ha

fl
co

L.
S.

Th
is

st
ud

y

II
up

pe
r

BK
1.

7–
1.

15
H

.e
re

ct
us

rw
D

ev
.O

ld
ow

an
B,

10
1

0
38

8
33

19
9

51
2

18
10

5
23

45
37

4
18

53
43

62
7

46
11

14
?

?
68

01
29

57
A

.b
oi

se
i

A
ch

eu
le

an

II
m

id
dl

e
FC

W
es

t
1.

7–
1.

15
sp

.i
nd

et
.

cw
D

ev
.O

ld
ow

an
B

49
0

5
4

4
48

53
11

9
1

2
0

7
23

74
8

15
9

5
2

1
?

O
cc

up
at

io
n

si
te

11
84

12
7

II
m

id
dl

e
M

N
K

1.
65

–1
.5

3
sp

.i
nd

et
.

cw
O

ld
ow

an
,

96
0

9
9

19
15

9
72

24
59

0
7

0
24

64
32

69
44

9
42

15
17

M
ed

iu
m

to
ve

ry
O

cc
up

at
io

n
si

te
43

99
17

23
H

om
o

sp
.

D
ev

.O
ld

ow
an

B,
la

rg
e

m
am

m
al

s
H

.h
ab

ili
s

cf
.

A
ch

eu
le

an
H

.e
re

ct
us

II
m

id
dl

e
SH

K
1.

7–
1.

15
cw

D
ev

.O
ld

ow
an

B
29

3
0

68
21

62
31

8
11

5
13

1
86

11
12

3
26

0
57

7
27

9
5

1
?

O
cc

up
at

io
n

si
te

18
5

?

II
lo

w
er

H
W

K
1.

7–
1.

15
sp

.i
nd

et
.

cw
O

ld
ow

an
41

1
0

0
3

1
14

4
0

2
0

0
0

21
24

0
6

3
1

?
O

cc
up

at
io

n
si

te
15

4
42

5
Ea

st
H

.h
ab

ili
s

D
ev

.O
ld

ow
an

A
le

ve
l1

A
.b

oi
se

i

I
up

pe
r

FL
K

2.
1–

1.
7

sp
.i

nd
et

.
cw

O
ld

ow
an

14
3

7
0

14
16

18
90

18
18

0
0

0
21

10
3

11
26

11
8

0
0

0
Sm

al
lm

am
m

al
s

FL
K

N
1–

5
15

94
76

87
N

or
th

A
us

tr
.s

p.
D

ev
.O

ld
ow

an
A

&
B

in
te

rp
re

te
d

as
Le

ve
ls

H
.h

ab
ili

s
oc

cu
pa

tio
n

flo
or

,
1–

5
A

.b
oi

se
i

ho
m

e
ba

se
s

or
H

.e
re

ct
us

ce
nt

ra
lf

or
ag

in
g

si
te

s

I
up

pe
r

FL
K

2.
1–

1.
7

H
.h

ab
ili

s
cw

O
ld

ow
an

4
1

0
0

1
0

2
0

0
0

0
0

6
4

85
6

0
0

0
La

rg
e

m
am

m
al

s
Bu

tc
he

ri
ng

si
te

13
0

2
25

8
N

or
th

6
D

ev
.O

ld
ow

an
A

&
B

w
ith

ar
te

fa
ct

s
an

d
el

ep
ha

nt
em

be
dd

ed
in

cl
ay

I
m

id
dl

e
FL

K
2.

1–
1.

7
sp

.i
nd

et
.

cw
O

ld
ow

an
2

0
0

1
0

0
19

1
0

0
0

0
0

1
30

8
1

0
0

R
ar

e,
on

ly
on

O
cc

up
at

io
n

si
te

72
22

61
N

N
3

H
.h

ab
ili

s
sm

al
lm

am
m

al
s

I
m

id
dl

e
FL

K
2.

1–
1.

7
H

.h
ab

ili
s

?
D

ev
.O

ld
ow

an
B

2
0

0
1

0
0

4
0

0
0

0
0

2
0

0
2

0
0

0
?

?
11

18
7

le
ve

ls
21

,
A

.b
oi

se
i

17
,1

6,
15

,
13

,1
2,

11
,

10
,7

I
m

id
dl

e
FL

K
Z

in
j.

2.
1–

1.
7

sp
.i

nd
et

.
cw

O
ld

ow
an

17
0

0
9

3
1

40
9

18
4

0
0

5
13

21
93

15
5

6
3

0
La

rg
e

&
sm

al
l

Li
vi

ng
flo

or
,h

om
e

26
47

40
17

2
le

ve
l2

2
A

us
tr

.s
p.

A
ch

eu
le

an
m

am
m

al
s

ba
se

or
ce

nt
ra

l
H

.h
ab

ili
s

fo
ra

gi
ng

si
te

A
.b

oi
se

i

I
lo

w
er

M
K

2.
1–

1.
7

H
.h

ab
ili

s
?

O
ld

ow
an

?
?

?
?

?
?

?
?

?
?

?
?

?
?

?
?

1
0

0
?

?
?

?

I
lo

w
er

D
K

2.
1–

1.
7

H
om

o
sp

.
cw

O
ld

ow
an

47
1

0
32

27
7

79
10

20
3

1
0

3
48

77
6

11
8

2
2

1
La

rg
e

&
sm

al
l

O
cc

up
at

io
n

si
te

.
11

63
78

55
H

.h
ab

ili
s

m
am

m
al

s
St

on
e

ci
rc

le
in

le
ve

l3
?

?
FC

K
II

?
?

?
?

?
?

?
?

?
?

?
?

?
?

?
?

?
?

?
?

5
0

1
?

?
?

?

?:
da

ta
fr

ag
m

en
ta

ry
or

ab
se

nt
.

Lo
c.

:L
oc

at
io

n
rw

:r
ew

or
ke

d
st

re
am

ch
an

el
de

po
si

t;
cw

:c
lo

se
to

w
at

er
(la

ke
sh

or
e/

ri
ve

r/
st

re
am

).
ch

:c
ho

pp
er

s;
pb

:p
ro

to
-b

ifa
ce

s;
bf

:b
ifa

ce
s;

pl
:p

ol
yh

ed
ro

ns
;d

s:
di

sc
oi

ds
;s

p:
sp

er
oi

ds
/s

ub
-s

pe
ro

id
s;

no
:n

od
ul

es
&

bl
oc

ks
;s

h:
sc

ra
pe

rs
,h

ea
vy

du
ty

;
sl

:s
cr

ap
er

s,
lig

ht
du

ty
;b

u:
bu

ri
ns

;a
w

:a
w

ls
;o

e:
ou

til
s

ec
ai

lle
s;

lt:
la

te
ra

lly
tr

im
m

ed
fla

ke
s;

an
:a

nv
ils

;h
a:

ha
m

m
er

st
on

es
;f

l:
fla

ke
;c

o:
co

re
.

Z
in

j.:
Z

in
ja

nt
hr

op
us

.
L.

:L
ea

ke
y;

S.
:S

hi
pm

an



Focusing on skeletal part frequency and cut-mark data
on 60 000 bone specimens from the FLK Zinjanthropus site
(middle Bed I), Bunn & Kroll (1986) found a high limb to
low axial skeleton representation, with cut-marks located
in places consistent with butchering practices. These data
are interpreted as evidence of hominids transporting
selected portions of carcasses to favoured localities in the
landscape (Bunn 1986). In addition, a high proportion of
prime adult mammal remains suggests to these authors
that hominids aggressively scavenged and may even
have hunted large animals. Bunn & Kroll (1986) note
that sharp-edged stone flakes are among the best-known
cutting tools, and numerically constitute the bulk of the
Oldowan assemblages from Olduvai. The meat-cutting
function of the flakes is supported by microwear studies
(Keeley & Toth 1981) demonstrating that unmodified
flakes from Koobi Fora were used to cut meat.

Using a landscape archaeology approach, Blumenschine
& Masao (1991) sampled the spatial distribution, density
and character of archaeological occurrences in a 1 km2

area of the HWKE site, lower Bed II. A high percentage of
core tools and long bone specimens preserving fractures
or percussion marks indicative of hammerstone breakage,
suggests to them an association with marrow extraction.
Trenches near the lake shore preserve a greater proportion
of large mammal long bones showing evidence of
hammerstone breakage. This pattern is consistent with
modern observations of lower levels of competition
among carnivores for bovid carcasses in the near-lake
environs surrounding Lake Ndutu in the Serengeti
(Blumenschine 1987). It suggests hominids had better
opportunities to gain access to whole marrow bones near
the shore of palaeo-Lake Olduvai. Blumenschine & Masao
(1991) argue that the apparent continuous distribution of
artefacts and associated bones away from purported
occupation sites show that repeated visits to particular loci
cannot be proved for Bed II times. A lack of trees in the
palaeo-lake margin zone meant no refuge from predators
while processing carcass parts, which they propose were
most frequently procured through scavenging of preda-
tor kills. Stone tools were transported to the butchering
sites, and the variability in density of bones and artefacts
suggests that hominids at times concentrated carcass
parts for processing, attracted perhaps to isolated patches
of shade or stone caches. The hypothesized paucity of
trees suggests hominid visits were brief, and that their
subsistence and social activities were focused elsewhere
in the Olduvai Basin.

To test the various hypotheses concerning the timing
and nature of hominid and carnivore activities in
Plio-Pleistocene bone assemblages from Olduvai (Bunn
1986; Bunn & Kroll 1986; Leakey 1971; Binford 1981),
Blumenschine (1995) focused his attention on the bone
assemblage from the FLK Zinjanthropus site. Frequencies
of percussion and tooth marks reject the hypothesis that
carnivores had first access to long bone marrow. This
contradicts Binford’s interpretation of hominids as
marginalized scavengers of already heavily ravaged
carcasses. A high proportion of tooth-marked long bone
mid-shaft fragments also rejects the alternative hypothe-

sis, that carnivore access was secondary to butchery and
marrow extraction by hominids (Leakey 1971; Bunn 1986).
Blumenschine (1995) proposes that the sequence of carni-
vore and hominid access to long bones and their marrow
is consistent with scavenging by hominids.

The Olduvai bone tool collection
The Olduvai bone tool collection housed in the Depart-

ment of Archaeology at the National Museums of Kenya
in Nairobi consists of 125 specimens that were analysed by
us in October 2001. These include some pieces that were
not designated as tools by Leakey (1971), since seven
specimens interpreted as tools by her and later by
Shipman (1989: 323), could not be located in the museum
(HWKEII 368; HWKEII 886; MNKII 23369; MNKII 1099;
BKII 2494; BKII 068-6688; BKII 3240). Annotated line-
drawings, comprising two to four aspects of each speci-
men were made. These recorded the location of macro-
and microscopic modifications such as original or post-
depositional breakage, flake removals, punctures, carni-
vore traces, cut-marks, trampling and polish. Recorded
variables also included taxon, body part, bone region
involved, dimensions of each specimen, the weathering
stage according to Behrensmeyer (1978), and location,
number, association and length of flake scars according to
fracture axis. While some of these variables have already
been recorded by Shipman, others such as the number,
location on the bone flake, occurrence on the periosteal
versus medullar face, and dimension of removals, possibly
due to intentional shaping, were recorded in the frame-
work of the present study for the first time. The term
‘flake’ is used here to describe pieces detached from long
bones, and may be taken to encompass fragments. Long
bone ends or shaft pieces are described as such, or referred
to as ‘pieces’.

The same variables were recorded on a control sample of
86 randomly-selected limb bone shaft fragments from the
FLKI, FLKNI, FLKII, BKII, MNKII and DKI Olduvai sites.
This was to establish whether the modifications recorded
on the purported bone tools did not represent an extreme
in variation, affecting to a lesser degree the remainder of
the Olduvai assemblage. Colour slides and digital images
of two to four aspects of each piece were also taken in
order to document the collection.

Using high-resolution dental impression material
(Coltene President), 76 replicas were made from different
areas of the purported tools and the control sample,
which consisted of shaft fragments from the FLKI, FLKII
and MNKII Olduvai sites. Cast areas included the edges of
the tools, whether described by Shipman as utilized or
not, regions located away from the purported functional
zones, and similar areas on the control specimens. All
puncture marks and some cut-marks were also moulded.
Transparent replicas made with RBS resin (T2L Chimie,
France) were cast from these moulds. All were examined
in transmitted light using an optical microscope (Wild
M3C) equipped with a digital camera, and 300 digital
micrographs were captured. Forty-one replicas were
analysed with a scanning electron microscope (840A Jeol)
(Bromage 1987; d’Errico 1988b) and 380 SEM micrographs

102 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158



were taken at ×15 to ×350 magnification. The presence of
striations (either single or multiple, parallel or intersect-
ing) and evidence of smoothing, polishing, pitting, and
possible residues was recorded.

Comparative collections
Thirty-five non-human reference collections of known

taphonomic history were examined and studied using the
same microscopic techniques described above (Backwell
2000; Backwell & d’Errico 2001). These represent nine
damage categories derived from both modern and fossil
contexts, including animals (hyaena, dog, leopard, cheetah,
porcupine) and geological processes (river gravel, spring,
flood plain, wind, trampling).

Experimental material
Nine modern elephant limb bones (Table 3) were experi-

mentally broken by 26 university students of mixed
gender. Ranging between nine and 22 kg each in weight,
eight of the bones originated from a young adult c. 20
years old that had died five months before the experi-
ment. Only one bone originated from a teenage individ-
ual and was weathered. The experiment was conducted at
Plovers Lake in the Sterkfontein Valley, South Africa. The
students were asked to work in groups of three to five in
order to break the bones and produce flakes, employing
only resources available in the environment. Knapping of
bone flakes was attempted by one of us (F.D.) using elon-
gated pebbles to replicate the flake removals recorded on
the Olduvai purported bone tool collection. Un-retouched
flakes were used for flaying and cutting the fresh meat
from an adult male eland, working fresh hides with the
addition of sand, dry hides with the addition of salt, and
digging in soil to extract tubers and grubs, as well as
removing bark from trees.

RESULTS

Microscopic analysis

Olduvai
Edges or tips of the bone specimens described by

Shipman as probable tools, show at microscopic scale a
great deal of variation in their appearance (Table 4). The
large majority are characterized by smoothing associated
with or without either parallel or intersecting single or
multiple striations (Fig. 2a–i).

This pattern is more pronounced on some pieces or
areas of a single specimen, where it may in places
completely obliterate the anatomical structure of the bone
(Fig. 2b–c). The smoothing often decreases from the edge
toward the inside of the object and in one case (Fig. 2d), a
clear worn band of 1 mm wide appears on the edge. A
minority of these bones present edges covered by a
more-or-less glossy polish associated with no or very few
striations (Fig. 2j–l). Although features that appear as
micro-pits are common on most pieces (Fig. 2b,o), it is
often difficult to distinguish between concavities pro-
duced by impact or pressure of sedimentary particles, and
damaged bone structures such as vascular openings and
Haversian canals (Fig. 2m,l).

Interpreting these wear patterns as evidence of tool use
is problematic. Comparable wear is identified on areas of
the purported tools located a considerable distance from
the worn edge unsuitable for use (Fig. 2m–n). Microscopic
analysis of the edges of the specimens described by
Leakey as tools, but rejected by Shipman, also cautions
against an anthropic interpretation, since a number of
them (e.g. BK 3122, BK 201, MNKII 848) record wear that
falls within the three categories described above on the
Shipman tools (Fig. 2o–r). Examination of the control
sample also reveals the same range of surface features
seen on Shipman’s and Leakey’s tools (Fig. 3a–f), and
consists of randomly selected shaft fragments from
Olduvai Beds I–II, bearing no apparent traces of anthropic
modification.

In addition to these observations, we have found on
three specimens (two purported tools and one bone from
the control sample) micro-crystals of calcite growing pre-
dominantly on vascular openings and cracks (Fig. 4).
These crystals were probably created by the slow dehy-
dration of bone containing water rich in calcium carbon-
ate. The excellent state of preservation of the crystals and
the fact that they clearly overlie the traces of abrasion, sug-
gest that they appeared only after the abrasion process
took place and that their growth represents the final
taphonomic event that affected part of the bone assem-
blage. These crystals are easily affected by mechanical and
chemical alteration and would not have survived in such
a good condition if the bone had undergone even to a
small degree (e.g. simple re-hydration) one of these pro-
cesses.

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 103

Table 3. Elephant bone breakage and flake use experiments

Tasks

Skeletal Bone Weight Preservation Breakage technique Gender of Flaying Working Working Digging Removing
element No. (kg) breakers fresh hide dry hide soil bark

Femur 1 19 Semi fresh Struck against rock m 2 2 1 2 1
Femur 2 22 Semi fresh Rock on bone bridge f – 3 1 – –
Ulna 3 11 Semi fresh Struck against rock f – – – – –
Ulna 4 11 Semi fresh Thrown against a rock f – – 1 – –
Femur 5 13 Weathered Rock thrown on bone m 3 2 – 1 2
Humerus 6 9 Semi fresh Struck against rock m – – – – –
Humerus 7 22 Semi fresh Struck against rock m 1 – – – –
Tibia 8 9 Semi fresh Rock on bone bridge m – – – – –
Humerus 9 12 Semi fresh Thrown against a rock m – – 1 – –

Total 6 7 4 3 3



104 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Ta
bl

e
4.

Su
m

m
ar

y
of

su
rf

ac
e

m
od

ifi
ca

tio
ns

re
co

gn
iz

ed
by

m
ic

ro
sc

op
ic

an
al

ys
is

on
sp

ec
im

en
s

de
sc

ri
be

d
as

bo
ne

to
ol

s
by

Le
ak

ey
an

d
by

Sh
ip

m
an

,a
nd

on
a

co
nt

ro
ls

am
pl

e
fr

om
O

ld
uv

ai
Be

d
I–

II
.

Pr
ev

io
us

st
ud

ie
s

Pr
es

en
ts

tu
dy

Si
te

Sp
ec

im
en

an
d

To
ol

s
M

od
ifi

ca
tio

n
Lo

ca
tio

n
Su

rf
ac

e
m

od
ifi

ca
tio

ns
re

pl
ic

a
*

**
**

**
Ti

p
Ed

ge
Bo

dy
Sm

oo
th

ed
Po

lis
he

d
R

an
do

m
ly

or
ie

nt
ed

Se
ts

of
pa

ra
lle

l
Se

ts
of

pa
ra

lle
l

Sh
al

lo
w

Pi
tt

in
g

C
ut

-m
ar

k
si

ng
le

st
ri

at
io

ns
st

ri
at

io
ns

in
te

rs
ec

tin
g

st
ri

at
io

ns
gr

oo
ve

s

D
K

I
06

7/
42

59
1,

2
W

ea
r

–
■

–
■

–
–

■
■

–
■

–
FL

K
II

88
4a

1,
2

Pu
nc

tu
re

s
–

–
■

■
–

■
■

■
–

■
–

FL
K

II
88

4b
1,

2
Pu

nc
tu

re
s

–
–

■
–

–
–

–
■

–
–

–
FL

K
II

88
4c

1,
2

Pu
nc

tu
re

s
–

–
■

–
–

■
–

–
–

–
–

FL
K

II
88

4d
1,

2
Pu

nc
tu

re
s

–
–

■
■

–
■

–
–

–
–

■
(x

)
FL

K
II

88
4e

1,
2

Pu
nc

tu
re

s
–

–
■

–
–

■
–

–
–

–
–

FL
K

II
88

4f
1,

2
Pu

nc
tu

re
s

–
–

■
–

–
■

–
■

–
■

■
(x

)
FL

K
II

88
4g

1,
2

Pu
nc

tu
re

s
–

–
■

–
–

■
–

■
–

–
–

FL
K

II
88

4h
1,

2
Pu

nc
tu

re
s

–
–

■
–

–
–

–
■

–
–

–
FL

K
II

88
4i

1,
2

Pu
nc

tu
re

s
–

–
■

–
–

–
–

■
–

■
–

FL
K

II
sp

it
5+

a
1,

2
W

ea
r

–
■

–
■

–
■

■
■

–
■

–
FL

K
II

sp
it

5+
b

1,
2

W
ea

r
–

■
–

■
–

■
–

■
–

■
–

FL
K

II
sp

it
5+

c
1,

2
W

ea
r

–
■

–
■

–
■

–
■

–
■

–
H

W
K

EI
I

06
8/

66
90

1,
2

W
ea

r
■

–
–

■
–

■
■

■
–

■
–

H
W

K
EI

I
/3

(h
ea

d)
1

W
ea

r
■

–
–

■
■

■
■

■
■

■
–

FC
II

06
8–

66
79

1,
2

W
ea

r
■

–
–

■
■

■
■

■
–

■
–

M
N

K
II

47
5

1,
2

W
ea

r
■

–
–

■
–

■
■

■
–

■
–

M
N

K
II

06
8/

66
76

1,
2

W
ea

r
–

■
–

■
–

■
–

■
–

■
–

M
N

K
II

84
8

1
W

ea
r

■
–

–
■

–
■

■
■

–
■

–
M

N
K

II
92

5
1

C
ut

-m
ar

k
–

–
■

■
–

–
–

–
–

–
■

M
N

K
II

10
51

a
1

W
ea

r
–

■
–

■
–

–
–

–
–

–
–

M
N

K
II

10
51

b
1

Pu
nc

tu
re

s
–

–
■

■
–

■
–

■
–

■
–

M
N

K
II

28
89

a
1,

2
W

ea
r

–
■

–
■

–
–

■
–

–
■

–
M

N
K

II
28

89
b

1,
2

W
ea

r
■

–
–

■
–

–
■

–
–

■
–

M
N

K
II

24
74

a
1,

2
W

ea
r

–
■

–
■

–
■

■
■

■
■

–
M

N
K

II
24

74
b

1,
2

W
ea

r
■

–
–

■
–

■
■

■
–

■
■

M
N

K
II

73
8

1,
2

W
ea

r
–

■
–

■
–

–
–

■
–

■
–

M
N

K
II

17
31

a
1,

2
W

ea
r

–
■

–
■

■
■

■
–

–
–

–
M

N
K

II
17

31
b

1,
2

W
ea

r
–

■
–

■
–

■
■

■
–

■
–

M
N

K
II

29
03

a
1,

2
W

ea
r

–
■

–
■

–
■

■
■

–
■

–
M

N
K

II
29

03
b

1,
2

W
ea

r
–

■
–

■
–

■
■

■
–

■
–

M
N

K
II

10
46

a
1,

2
C

ut
-m

ar
k

–
–

■
■

–
–

■
■

–
■

■

M
N

K
II

10
46

b
1,

2
C

ut
-m

ar
k

–
–

■
■

–
■

■
■

–
–

■

M
N

K
II

17
41

a
1,

2
W

ea
r

–
■

–
■

–
–

–
■

–
■

–
M

N
K

II
17

41
b

1,
2

W
ea

r
–

■
–

■
–

–
–

–
–

–
–

M
N

K
II

17
41

c
1,

2
W

ea
r

–
■

–
■

–
–

–
■

–
■

–
M

N
K

II
47

1a
1

W
ea

r
■

–
–

■
–

■
■

■
–

■
–

M
N

K
II

47
1b

1
W

ea
r

–
■

–
■

–
■

■
■

–
■

–
M

N
K

II
74

4
1

W
ea

r
–

■
–

■
–

–
–

–
–

–
–

M
N

K
II

11
16

1
W

ea
r

–
■

–
■

–
–

–
–

–
–

–
SH

K
II

06
8/

66
84

1,
2

W
ea

r
■

–
–

■
–

–
–

–
–

–
–

SH
K

II
06

8/
66

88
1,

2
W

ea
r

–
■

–
■

■
■

–
–

–
■

–

C
on

tin
ue

d
on

p.
10

5



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 105

Ta
bl

e
4

(c
on

tin
ue

d)

Pr
ev

io
us

st
ud

ie
s

Pr
es

en
ts

tu
dy

Si
te

Sp
ec

im
en

an
d

To
ol

s
M

od
ifi

ca
tio

n
Lo

ca
tio

n
Su

rf
ac

e
m

od
ifi

ca
tio

ns
re

pl
ic

a
*

**
**

**
Ti

p
Ed

ge
Bo

dy
Sm

oo
th

ed
Po

lis
he

d
R

an
do

m
ly

or
ie

nt
ed

Se
ts

of
pa

ra
lle

l
Se

ts
of

pa
ra

lle
l

Sh
al

lo
w

Pi
tt

in
g

C
ut

-m
ar

k
si

ng
le

st
ri

at
io

ns
st

ri
at

io
ns

in
te

rs
ec

tin
g

st
ri

at
io

ns
gr

oo
ve

s

SH
K

II
06

8/
66

87
1,

2
W

ea
r

–
■

–
■

–
■

■
–

–
■

–
SH

K
–E

II
06

8/
66

81
1

W
ea

r
–

–
■

■
–

■
■

■
–

■
–

SH
K

II
06

8/
66

85
1

W
ea

r
■

–
–

■
–

–
–

–
–

■
–

BK
II

31
22

a
1

W
ea

r
■

–
–

■
–

–
–

–
–

–
–

BK
II

31
22

b
1

W
ea

r
■

–
–

■
■

■
■

■
–

■
–

BK
II

06
8/

66
80

a
1

C
ut

-m
ar

k
–

–
■

■
–

■
–

–
–

–
■

BK
II

06
8/

66
80

b
1

C
ut

-m
ar

k
–

–
■

■
–

■
–

–
–

–
■

BK
II

19
38

a
1,

2
W

ea
r

–
■

–
■

–
■

■
■

–
■

–
BK

II
19

38
b

1,
2

W
ea

r
–

■
–

■
–

■
–

–
–

■
–

BK
II

19
38

c
1,

2
W

ea
r

–
–

■
■

–
■

■
■

–
–

–
BK

II
53

–9
ex

19
53

a
1

W
ea

r
■

–
–

■
–

■
■

■
–

■
–

BK
II

53
–9

ex
19

53
b

1
W

ea
r

–
–

■
■

■
■

■
■

–
■

–
BK

II
53

–9
ex

19
53

c
1

W
ea

r
–

■
–

■
–

–
■

■
–

■
–

BK
II

20
1

1
W

ea
r

■
–

–
■

–
■

■
■

–
■

–
BK

II
29

33
a

1,
2

Pu
nc

tu
re

s
–

–
■

■
–

–
–

■
–

■
–

BK
II

29
33

b
1,

2
Pu

nc
tu

re
s

–
–

■
■

–
–

–
■

–
–

–
BK

II
29

33
c

1,
2

Pu
nc

tu
re

s
–

–
■

■
–

■
–

■
–

–
–

BK
II

06
8/

66
78

1,
2

W
ea

r
■

–
–

■
–

■
■

■
–

■
–

BK
II

06
8/

66
66

1
W

ea
r

■
–

–
■

–
■

■
■

–
■

–
BK

II
06

8/
66

70
1

W
ea

r
–

■
–

■
–

■
■

–
■

–
–

BK
II

93
3

1
W

ea
r

■
–

–
■

–
–

–
–

–
–

–
BK

II
18

7a
1

C
ut

-m
ar

k
–

–
■

■
–

–
–

–
–

–
■

BK
II

18
7b

1
C

ut
-m

ar
k

–
–

■
■

–
–

–
■

–
–

■

BK
II

18
7c

1
C

ut
-m

ar
k

–
–

■
■

–
■

■
■

–
■

■

BK
II

06
8/

66
68

1
W

ea
r

■
–

–
■

–
■

■
■

–
■

–
FC

K
II

06
8/

66
82

z
1

W
ea

r
–

■
–

■
–

–
■

–
–

–
–

FL
K

I
34

0a
**

*
**

*
–

–
■

■
–

–
–

■
–

–
■

FL
K

I
34

0b
**

*
**

*
–

–
■

■
–

–
■

■
–

■
■

FL
K

I
11

2
**

*
**

*
–

■
–

■
–

–
–

–
–

–
–

FL
K

I
34

1
**

*
**

*
–

–
■

■
–

■
–

■
–

■
■

FL
K

II
33

9a
**

*
**

*
■

–
–

■
–

■
■

■
–

■
–

FL
K

II
33

9b
**

*
**

*
■

–
–

■
–

■
■

■
–

■
–

FL
K

II
78

0
**

*
**

*
■

–
–

■
–

■
■

■
–

■
–

M
N

K
II

27
70

**
*

**
*

■
–

–
■

–
■

■
■

–
■

–
FL

K
II

TR
I+

**
*

**
*

■
–

–
■

–
■

■
■

–
■

–

*T
he

lo
w

er
ca

se
le

tt
er

id
en

tif
ie

s
th

e
re

pl
ic

a
w

he
n

m
or

e
th

an
on

e
re

pl
ic

a
pe

r
sp

ec
im

en
w

as
an

al
ys

ed
;*

*
to

ol
s

af
te

r
Le

ak
ey

(1
)a

nd
Sh

ip
m

an
(2

);
**

*
co

nt
ro

ls
am

pl
e;

**
**

m
od

ifi
ca

tio
n

af
te

r
Le

ak
ey

an
d

Sh
ip

m
an

;(
x)

cu
t-

m
ar

k-
lik

e
gr

oo
ve

s
as

so
ci

at
ed

w
ith

pu
nc

tu
re

s.



106 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 2. a–n, Wear pattern on bone fragments interpreted as tools by Shipman (a–d: smoothing; e–f: smoothing associated with sub-parallel
striations; g–h: parallel striations; i: sets of intersecting parallel striations; j–l: polish with no striations); m–n: wear pattern on areas of the objects
interpreted as tools, located away from the purported functional area (m: smoothing, n: sub–parallel striations on a smoothed area); o–r, pieces
interpreted by Leakey as tools but rejected by Shipman (o: smoothing, p–q: polish, r: parallel striations). a: BKII 1938b, b: FCIIS 068–6679, c: HWKEII
068–6690, d: FLKII spit5+a, e: BKII 1938a, f: MNKII 1741c, g: FCII 068–6679, h: FLKII spit5+, i: HWKEII 6690, j: MNKII 1731a, k: FCIIS 068–6679,
l: SHKII 068–6688, m: BKII 53–9b, n: BKII 1938c, o: BKII 201, p: BKII 1953b, q: BKII 1953b, r: BKII 3122b.



Comparative collections
Damage inflicted on bones by non-human agents may

also closely match, at microscopic scale, the surface
features observed on the edges of the purported tools.
Comparable features are observed on bones collected by
Brain at the Homeb Hottentot water hole in Namibia,
where they were subjected to trampling by goats
(Fig. 3g–i). The edges of these pieces record significant
smoothing of the more elevated regions often associated
with individual and sets of parallel striations. Bone
pseudo-tools from the Bacon Hole fossil hyaena den cave
site (Stringer 1977) also record comparable modifications,
including smoothing of elevated areas (Fig. 3j) and zones
covered with broad or fine parallel striations (Fig. 3k–l). In
both comparative collections, where these modifications
affect the bone surface to a lesser degree, only more

exposed areas such as ridges, edges and tips develop a
detectable wear pattern that may not appear on the
remainder of the object, thus producing a pattern that
might be erroneously interpreted as differential wear due
to anthropic use.

Wear patterns on experimentally used bone tools
Experimental use of unmodified shaft fragments

applied to different tasks produced distinct wear patterns
that, in part, overlap with those described by other
authors (MacGregor 1975; Campana 1980; Peltier 1986;
Shipman 1988, 1989; Brain & Shipman 1993; d’Errico
1993b; Lemoine 1997; Backwell & d’Errico 2001). The
working of fresh hides with sand by maintaining the tool
perpendicular to the hide, flattens the edge and smoothes
a 2–3 mm wide adjacent band (Fig. 5a–e). The edge is

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 107

Figure 3. a–f, Edges and tips of shaft fragments from the remainder of the Olduvai Beds I–II assemblage not considered as tools by Shipman and by
Leakey, and bearing no apparent anthropic traces (a: smoothing of elevated areas, b: smoothing with randomly oriented striations, c–e: smoothing
with sets of parallel striations. e: close–up view of (d) showing a bone islet resulting from abrasion; f: polished area crossed by individual striations);
g–i, edges of bone pieces from the Homeb Hottentot water hole bearing traces of smoothing and striations; j–l, edges of bone pieces from the Bacon
Hole hyaena den showing smoothing and areas covered by parallel striations.



intersected by relatively broad superficial striations, and
the adjacent band by narrow parallel striations perpen-
dicular to the tool edge. Flaying with a bone tool produces
a smoothing of the edge associated with polishing of
prominent areas (Fig. 5f). Individual sub-parallel grooves
develop on flat underside areas (Fig. 5g). This activity also
chemically alters the bone surface, differentially etching
the bone structure (Fig. 5h). Bark removal creates a
smoothed surface covered by individual sub-parallel
striations. Digging in soil produces an association of single
and bar-code-like composite broad striations, oblique or
parallel to the main axis of the tool (Fig. 5k–l).

Punctured bones
Only two of the four pieces interpreted by Leakey and

by Shipman as anvils, a giraffe astragalus (BKII 2933,
Fig. 7) and an elephant patella (FLKII 884, Fig. 8), were lo-
cated in the National Museums of Kenya. Our reappraisal
of these pieces has taken into account criteria proposed by
other authors for identifying the causes of impressions on
bone, as well as observations made on our experimentally
broken elephant limb bones. Tooth pits, crushes and
punctures produced by carnivores are well known features
commonly described in the taphonomic literature
(Binford 1981; Haynes 1983; Lyman 1994; Fisher 1995).
Tooth pits are superficial, roughly circular markings
producing no inward crushing of the bone cortex.
Crushes are roughly circular depressions of cortical bone
nested in underlying cancellous bone, often in the vicinity
of epiphyses. Punctures are roughly circular holes in
cortical bone with irregular edges, depressed margins
and flaking of the outer wall of the bone pushed into the
depression. More regular edges are seen on punctures
made on thin cortical bone.

Percussion pits, impact marks, chop-marks, crushing
and percussion striae are terms used to describe the alter-
ations created by a hammer-stone striking a bone surface
(Binford 1981; Blumenshine & Selvaggio 1988; White
1992; Oliver 1994; Fisher 1995). Though referring to the
same phenomenon – the mark inflicted on a bone by a
hammer – the first three terms indicate impact marks of
variable shape. This shape is determined by the morphology
of the contact area of the stone hammer. Tools with

knapped edges typically produce deep v-shaped marks
(chop-marks), while angular-edged hammer-stones
produce irregularly shaped depressions with complex
internal morphologies, and fine-grained pebbles result in
more uniform superficial depressions. Crushing caused
by stone tools is defined as the result of an impact in which
thin cortical bone is nested in underlying cancellous bone.
These features are often associated with micro-striations
resulting from the contact of the hammer-stone tip with
the bone surface before or after the impact.

It may be difficult distinguishing between carnivore-
and hominid-induced punctures when micro-striations
are absent, and the mark produced by the hammer-stone
is circular, rounded, and does not display internal features
reflecting the irregular morphology of the hammer-stone
tip. These features may be mistaken for carnivore activity.
Poor surface preservation can also mask diagnostic char-
acters and make the identification of the agent problem-
atic.

Our bone breakage experiments, conducted using
dolomite and quartzite blocks as hammers or anvils,
confirm the criteria described in the literature for stone
tool-generated puncture marks. In our experiments, this
activity produced three main features that may or may
not be found together (Fig. 6). These are irregular depres-
sions, the morphology of which is determined by the
shape of the tool tip penetrating the bone (Fig. 6a), the
lifting or detachment of micro-flakes adjacent to the
impression (Fig. 6a,c), and the production of broad com-
posite striations visible inside or close to the puncture
(Fig. 6b).

The astragalus from Olduvai (Fig. 7) bears on its dorsal
face a cluster of overlapping punctures of consistent trian-
gular shape and orientation. Apart from the features
already described by Shipman, indicating that these
punctures were made by the same stone tip repeatedly
striking the object, we have identified striations within
two peripheral punctures (Fig. 7c–d), as well as inside the
main area of percussion. Although calculating the precise
number of punctures is difficult due to overprinting,
microscopic analysis suggests that at least 14 blows were
inflicted.

The patella records, on the left half of the articular

108 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 4. a–b, Examples of micro-crystals that developed on an already abraded bone surface (specimen MNKII 2474). Notice the rhombohedral
cleavage of the crystal at top left in (b) demonstrating the calcitic nature of the crystals.



surface, nine scattered punctures (Fig. 8) bearing ambigu-
ous features. Our analysis of these marks (Fig. 8a–i)
indicates that in spite of some degree of morphological
variability, these punctures were probably made by the
same agent striking the surface in a single session, as
demonstrated by their similar internal morphology and
the consistent orientation of their spindle-like shape. The
rounded/oval morphology of a number of them, macro-
scopically similar to carnivore punctures, and the absence
of clean angular edges, makes it difficult to securely
attribute them to hominin agency. Macroscopic analysis
of marks made on a similar bone by various carnivore
taxa, and use of this bone as a hammer on material such as
wood, is necessary before reaching a definite conclusion.
This experiment might also explain, if the carnivore
hypothesis is retained, why no other carnivore damage is
present on the specimen, as would be expected if all these

marks were made by a large carnivore.
In sum, our analysis confirms Leakey’s and Shipman’s

diagnosis of these bones as anthropically modified. We
believe, however, that an interpretation of these objects as
hammers used on intermediate stone tools, rather than
anvils on which to pierce skins, fits the evidence better.
Experimental piercing of leather (d’Errico et al. 2003)
shows that a rotating motion is needed to effectively
perforate this material and leave a suitable non-tearing
hole. If exerted against a bone surface, this motion results
in circular or semicircular impressions with curved inter-
nal striations, not seen on the Olduvai specimens. Also,
striking motions are unsuitable for piercing skin at precise
locations, as generally required by this activity. Piercing a
skin by striking a pointed stone tool against a bone anvil
requires a relatively large and stable bone. Neither of the
bones appears large enough, and the patella is very unsta-

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 109

Figure 5. SEM micrographs of the edges of modern elephant bone shaft fragments used to work hide with sand (a–e), flay an eland carcass (f–h),
remove bark from trees (i–j), and dig in soil in search of grubs and tubers (k–l).



ble. The dispersed location of the punctures on the patella
and the location of some impressions near the edge also
cast doubt on the anvil interpretation, since the bone
would have been destabilized by the striking force. We
argue for now that the astragalus, and perhaps the patella,
were instead used in single-session hammering tasks,
most likely on intermediate stone wedges used to split
bones, fruit or wood. Future research will include a wider
range of actualistic studies, including observations of
carnivore-generated bite marks and the impressions

produced by a number of stone types and tip shapes.
The presence of crude choppers used as hammerstones

and evidence of cut-marks and hammerstone-induced
fractures on bones from the oldest levels of Bed I (DK),
and their persistence through Bed II evidence the regular
use of hammers in the subsistence activities carried out at
Olduvai. In contrast, anvils occur only in Bed I (DK, FLK,
FKLK North), while awls are recorded only in Bed II,
appearing first in level 2 at HWK East, and later at SHK
and BK, the reworked stream channel deposit.

110 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 6. Puncture marks produced experimentally on fresh elephant bone; a, impressions of the stone tool tip associated (right) with chipping of the
outer bone surface; b, impressions with composite striations; c, concentration of punctures with lifting of primary bone lamellae. Scale bars = 1 cm.

Figure 7. a–b, Astragalus from Olduvai (BKII 2933) with puncture marks, c–d, close up view showing striations associated with punctures (arrows).



Taphonomic and morphometric analysis

Experimental results
Four techniques were used by the students to break the

bones (Table 3). The first involved lifting the bone while
holding one epiphysis and repeatedly striking the opposite
epiphysis against a rock. This produced in one case a
mid-shaft break with no flakes, and in three other cases,
breakage with numerous flakes. The second technique
entailed throwing the bone repeatedly against a rock. This

was used for two bones, which both produced flakes. The
third, employed by females, involved throwing a rock at
the bone. To avoid the absorption of the striking force by
the ground, two bones were stabilized between two rock
outcrops, and another inclined against a separate rock.
The break of the former produced no flakes in one case
and many in the other, while that of the latter resulted in
numerous flakes. The fourth technique, attempted by
females, consisting of striking the bone with a hand-held
rock, was unsuccessful in bone breakage. Once the bones

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 111

Figure 8. Elephant patella from Olduvai (FLKII 884) with punctures on the articular surface.; a–i, close up views of the puncture morphology.



were broken, students attempted to detach more flakes
from the epiphyses by throwing the bones against rocks,
rocks against the bones or by using stone wedges. This,
however, resulted in the production of very few addi-
tional flakes. Shaping the flakes by knapping was also
attempted.

Percussion and scrape marks
Half of the flakes bear multiple large percussion marks

(Fig. 6), and 23% bear chop marks. Virtually all the pieces
with chop marks also have punctures, while only a third
of the punctured flakes are associated with chop marks.
The epiphyses of the bones struck against rocks show
distinct parallel scrape marks resulting from their tangen-
tial impact caused by unskilled striking motions (Fig. 9a).
The experimentally-produced scrape mark is characterized
by parallel grooves with a spindle-like shape, and clear
internal striations. A number of Olduvai epiphyses show
traces of carnivore scoring which may be confused with
strike marks (Fig. 9b). Carnivore marks, however, are
seldom parallel, may be curved, do not have clear internal
striations, and generally terminate abruptly at one end.

Flake analysis
Breakage of the nine elephant limb bones produced 134

flakes. Data were recorded on 107 of these pieces. Two
bones produced no flakes, while the remainder produced
between six and 29 flakes each, the highest figure deriving
from the weathered bone.

Flakes can be divided into two broad categories; those
made exclusively of compact bone (Fig. 10) and those
retaining spongy bone (Figs 11 & 12). The compact bone
includes slivers resulting from the detachment of primary
cortical bone (Fig. 10: Nos 6–10, 14), spindle-like thicker
splinters (Fig. 10: Nos 4–5), and rectangular blanks that are
flat on the ventral and dorsal aspects (Fig. 10: Nos 2, 3,
11,12). One sliver-like flake is noteworthy (Fig. 10: No. 7)
in that it demonstrates that very large blanks made of
compact bone (26 × 10 cm) can result from the breakage of
very large mammal bone when struck against rocks for
the exclusive purpose of marrow extraction. The flakes
with spongy bone include large elongated shaft frag-

ments with rounded or pointed ends that retain between
half and one third of the shaft section (Fig. 11), and irregu-
larly-shaped chunks with a considerable proportion of
cancellous bone (Fig. 12).

Data on the dimensions of the flakes are given in Tables
5 & 6 and Fig. 13. Results indicate that the frequency of
flakes in the different size classes remains constant in all
bones, with the exception of the weathered bone (Table 6,
no. 5), which shows an over-representation of elongated
flakes. This difference does not correspond to a change
in the general size of the flakes, as demonstrated by the
scattergram (Fig. 14) correlating the flake width and
length. This scattergram also shows that the size of the
flakes is not determined by the bone type or by the break-
age technique.

Pseudo-retouch
Fifteen flakes (14%) show removals/flake scars that may

be taken as evidence for deliberate shaping (Table 7). Six
bear a single removal, seven bear between two and five
removals, and only two pieces have eight removals. Re-
movals occur more often on the periosteal surface (eight
cases) than on the medullar (three cases). Four pieces
have removals on both periosteal and medullar surfaces.
Isolated removals generally do not exceed one per speci-
men, while multiple removals are in almost all cases found
in association. Results also show that removals occur more
often on the ends of flakes than on their sides. At close
inspection, however, only a few of these removals may be
taken as negatives of flakes produced by knapping. This is
because they lack features that would indicate that per-
cussion was applied. One piece, for example, presents a

112 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 9. a, Head of a modern elephant femur showing scrape marks produced by striking the epiphysis against a rock; b, femoral head from Olduvai
specimen BKII 2230 bearing weathered carnivore tooth notches, scores and pits as well as a modern scrape mark (arrow).

Table 5. Dimensions of flakes produced during the experimental break-
age of elephant limb bones.

Flake dimensions Mean S.D. Minimum Maximum
n = 107 mm mm mm

Length 139 76.6 23 385
Width 50 25.5 10 125
Maximum thickness 21 12.1 2 52
Compact bone thickness 16.5 9.2 2 44



wedge-like breakage with opposing flat scars that lack a
negative bulb of percussion (Fig. 15a). Another, resulting
from the breakage of the weathered bone, has scars due to
the lifting of primary bone lamellae opposite to irregularly
shaped scars (Fig. 15b). This morphology, difficult to
accept as evidence of purposeful modification, is as to be

expected, the result of the state of preservation of the
bone. Two pieces require special attention. One is a large
flake with a pointed end bearing overlapping removals
that mimic the tip of a dihedral burin (Fig. 15c). The other
is a large flake with a remarkable hand axe-like morphol-
ogy with contiguous pseudo-removals on both ends that

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 113

Figure 10. Cortical bone flakes produced by the experimental breakage of elephant limb bones.

Table 6. Length of the flakes from experimentally broken elephant bones.

Bone No. Flake length (cm)

0–5 5–10 10–15 15–20 20–25 25–30 30–35 35–40 Total

1 1 9 5 2 4 2 0 0 23
2 2 8 1 1 1 0 2 1 16
3 0 2 4 0 0 1 0 0 7
4 0 5 7 3 1 1 1 0 18
5 1 9 8 6 1 4 0 0 29
6 0 0 0 0 0 0 0 0 0
7 0 3 0 2 0 1 0 0 6
8 0 0 0 0 0 0 0 0 0
9 2 2 2 1 1 0 0 0 8

Total 6 38 27 15 8 9 3 1 107



mimic pseudo bifacial shaping at its base (Fig. 16). In spite
of its general resemblance to an Acheulean stone hand axe
or to one of the Acheulean elephant bone hand axes from
the Italian sites (Radmilli 1985; Radmilli & Boschian 1996),
this piece has no invasive contiguous bifacial scars.

Experimental shaping by knapping
Knapping of the elephant bone flakes using quartzite

and dolomite blocks, attempted by the students shortly
after the bone breakage, was unsuccessful. The students

were, however, unskilled knappers and the available
hammers seemed to be unsuitable for the task. Subse-
quent knapping, made by one of us (F.D.) using elongated
quartzite pebbles weighing c. 500 g, took place one year
after the breakage experiment, when the bone flakes had
dried considerably. Flakes from bones fresh at the time of
breakage experiments were difficult to knap, and split
longitudinally. The results obtained after soaking them in
water for two days were no better. Knapping was success-
ful on shaft fragments resulting from bone weathered at

114 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 11. Large bone flakes produced by the experimental breakage of elephant limb bones with a large proportion of spongy bone.

Figure 12. Irregularly shaped chunks of experimentally broken elephant limb bone retaining a high proportion of cancellous bone.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 115

Figure 13. Histograms of the dimensions of flakes produced during the experimental breakage of elephant bone.

Figure 14. Length/width correlation of the flakes produced during the experimental breakage of elephant bone. Symbols indicate flakes from the
same bone (see Table 3 for the bone type and the breakage technique used).

Table 7. Number, association and location of removals on the periosteal and medullar surfaces of flakes produced by the experimental breakage of
elephant limb bones.

No. rem. No. of flakes Association Location
per flake with removals Periosteal Medullar Periosteal Medullar

Total Peri. Med. Isolated Cont. Isolated Cont. End End+side Side End End+side Side

0 92 95 100 103 99 101 106 97 105 106 103 106 105
1 6 4 5 4 0 5 0 3 0 1 3 0 2
2 2 3 0 0 3 0 0 2 1 0 0 0 0
3 1 0 1 0 0 1 0 0 0 0 0 1 0
4 2 3 1 0 3 0 1 2 1 0 1 0 0
5 2 1 0 0 1 0 0 1 0 0 0 0 0
8 2 1 0 0 1 0 0 1 0 0 0 0 0

rem: removals; Peri: periosteal; Med: medulla; Cont: contiguous.



the time of the experimental breakage (Fig. 17). The large
invasive removals obtained exhibit, however, a rough
surface (Fig. 17b–c) different from the fine-grained texture
recorded on the scars on many Olduvai purported bone
tools. Also, the fracture surface is different, in that after
producing a concoidal-like negative, it progresses inward,

with its orientation determined by that of the bone
lamellae. This gives a portion of the scar a flat appearance
not seen at Olduvai.

Olduvai

Leakey bone tool collection and comparative sample
composition

The purported bone tool collection we analyse here
comprises 106 specimens. Teeth and tusks, as well as
complete bones, are not the subject of this paper and
complete bones bearing punctures have been analysed
above. The photographs of the analysed material are
presented in Figs 18–46, by site in alphabetical and numer-
ical order. Most of the analysed pieces are limb bone flakes
and shaft fragments (62%). The remainder consists of
epiphyses with or without a portion of the diaphysis.
Nearly 80% of the pieces come from large to very large
mammals (Fig. 47). Owing to their fragmentary nature,
most of them are difficult to identify to taxon level. The
most commonly identified animals in order of abundance
are giraffids, equids, bovids, elephantids, hippopotamids,
and rhinocerotids. Our comparative sample is instead
mostly composed of flakes and shaft fragments from
medium-sized mammals, particularly bovids (Fig. 47).
This is because outside of the bone tool collection, the
Olduvai assemblage comprises very few large limb bone
fragments, making impossible the construction of a more
appropriate comparative sample.

Bone preservation and surface modifications
The bone tool collection is generally characterized by an

excellent state of preservation. The large majority of the
pieces (Table 8) are either fresh (44%) or slightly abraded
(40%). Post-depositional breaks and scars, mostly due to
excavation and handling, account for 23% and 13% of the
collection, respectively. Almost all of the bones under-
went a rapid burial process; 48% show no weathering,
and 45% show a weathering stage 1 (Behrensmeyer 1978).
A moderate degree of trampling and polishing affects c.
21% of the pieces. The majority of the pieces have spiral
fractures, indicating that the bone was fresh when broken.
Hominids are the agent most likely to have been responsi-
ble for the breakage, considering that 30% of the bone tool
collection bears either cut-marks or clearly lithic-derived
punctures (Fig. 48a–e), and an additional 3% bears a
combination of both modifications (Table 9).

Carnivore marks, although occurring on 25% of the
collection, only consist of superficial tooth scores and pits.
No evidence of destruction of epiphyses by gnawing,
crenulated edges or a combination of these features

116 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 15. Pseudo-removals resulting from experimental breakage of
elephant bones.

Table 8. Bone surface modifications on Olduvai bones.

Collection No. pieces Surface modifications Weathering stage Degree of abrasion Post deposition

Carn. Cut-mark Perc. Tramp. Polish 0 1 2 Fresh Slight Mod. Heavy Break Rem.

Control 82 2 10 7 28 3 58 23 1 41 32 8 1 17 2
Leakey/Ship. 106 26 15 22 24 22 51 48 7 46 42 14 3 25 14

Ship: Shipman; Carn: carnivore tooth marks; Perc: percussion marks; Tramp: trampling: Mod: moderate; Rem: removal



with impact notches is found, suggesting that carnivore
involvement was limited and post-dated bone breakage.
Figure 48f illustrates the most extreme example of carni-
vore damage recorded. Carnivore-, percussion-, and
cut-marks occur on all bone types and mammal size

classes. Their absence from the bones of small mammals is
certainly due to the very low proportion of such bones in
the sample.

The above surface modifications occur in the Olduvai
control sample in proportions that are not significantly

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 117

Figure 16. Bone flake resulting from experimental breakage of elephant limb bones with a hand axe-like morphology.

Figure 17. a, Experimental knapping of a shaft fragment from a weathered elephant bone; b–c, close up view of the scars.



118 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 18. Olduvai bone tools proposed by Leakey. *BKII 068–6666 (top), BKII 068–6667 (centre) and *BKII 068–6668 (bottom). Scale bars = 1 cm. An
asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 119

Figure 19. Olduvai bone tools proposed by Leakey. BKII 068–6669 (top), *BKII 068–6670 (centre) and BKII 068–6671 (bottom). Scale bars = 1cm. An
asterisk indicates a bone tool according to this study.



120 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 20. Olduvai bone tools proposed by Leakey. BKII 068–6672 (top), BKII 068–6673 (centre) and BKII 068–6674 (bottom). Scale bars = 1cm.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 121

Figure 21. Olduvai bone tools proposed by Leakey, BKII 068–6678 (top), BKII 068–6674 (bottom), and by Leakey and by Shipman BKII 068–6673
(centre). Scale bars = 1 cm.



122 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 22. Olduvai bone tools proposed by Leakey, *BKII 068–6686 (top), BKII 1053 (centre), and by Leakey and by Shipman, BKII 1605 (bottom). Scale
bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 123

Figure 23. Olduvai bone tools proposed by Leakey. BKII 187 (top), *BKII 1938 (centre), and *BKII 200 (bottom). Scale bars = 1 cm. An asterisk indicates
a bone tool according to this study.



124 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 24. Olduvai bone tools proposed by Leakey. *BKII 201 (top), BKII 2230 (centre), and *BKII 2382 (bottom). Scale bars = 1 cm. An asterisk
indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 125

Figure 25. Olduvai bone tools proposed by Leakey, *BKII 2715 (top) and BKII 2959 (bottom), and by Leakey and by Shipman, BKII 2870 (centre). Scale
bars = 1 cm. An asterisk indicates a bone tool according to this study.



126 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 26. Olduvai bone tools proposed by Leakey, BKII 3118 (top) and BKII 3122 (centre), and by Leakey and by Shipman, *BKII 3155 (bottom). Scale
bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 127

Figure 27. Olduvai bone tools proposed by Leakey, BKII 505 (top), *BKII 1953 or 53–9 (centre), and by Leakey and by Shipman, BKII 869 (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



128 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 28. Olduvai bone tools proposed by Leakey, *BKII 933 (top), *DKI 067–4259 (centre), and by Leakey and by Shipman, DKI 4200 (bottom). Scale
bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 129

Figure 29. Olduvai bone tools proposed by Leakey, FCII 803 (top), FCII 068–6675 (centre), and by Leakey and by Shipman, *FCII 068–6679 (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



130 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 30. Olduvai bone tools proposed by Leakey, *FCKII 068–6682 (top), FLKII 45 (centre), and by Leakey and by Shipman, FLKII spit 5+ (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 131

Figure 31. Olduvai bone tools proposed by Leakey and by Shipman, HWKEII 068–6690 (top), and by Leakey, *HWKEII 249 (centre), HWKEII 3a
(bottom). Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



132 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 32. Olduvai bone tools proposed by Leakey, HWKEII 3b (top), HWKEII 4021 (centre) and HWKEII 866 (bottom). Scale bars = 1 cm.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 133

Figure 33. Olduvai bone tools proposed by Leakey and by Shipman, *MNKII 068–6676 (top), and by Leakey, MNKII 1051 (centre), MNKII 1053
(bottom). Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



134 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 34. Olduvai bone tools proposed by Leakey, MNKII 1059 (top), MNKII 1090 (centre) and *MNKII 1116 (bottom). Scale bars = 1 cm. An asterisk
indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 135

Figure 35. Olduvai bone tools proposed by Leakey, *MNKII 1117 (top), *MNKII 1133 (bottom), and by Leakey and by Shipman, MNKII 1123 (centre).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



136 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 36. Olduvai bone tools proposed by Leakey, MNKII 1269 (top), MNKII 1304 (centre), *MNKII 1496 (bottom). Scale bars = 1 cm. An asterisk
indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 137

Figure 37. Olduvai bone tools proposed by Leakey, MNKII 1563 (top), MNKII 1711 (centre), and by Leakey and by Shipman, *MNKII 1731 (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



138 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 38. Olduvai bone tools proposed by Leakey, *MNKII 1786 (centre), MNKII 2052 (bottom), and by Leakey and by Shipman, *MNKII 1741 (top).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 139

Figure 39. Olduvai bone tools proposed by Leakey, MNKII 2093 (top), MNKII 2355 (centre), MNKII 2360 (bottom). Scale bars = 1 cm.



140 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 40. Olduvai bone tools proposed by Leakey, MNKII 2369 (top), *MNKII 2464 (centre), and by Leakey and by Shipman, MNKII 2474 (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 141

Figure 41. Olduvai bone tools proposed by Leakey and by Shipman, *MNKII 2889 (top), MNKII 2903 (centre), MNKII 3243 (bottom). Scale bars =
1 cm. An asterisk indicates a bone tool according to this study.



142 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 42. Olduvai bone tools proposed by Leakey, MNKII 3335 (top), *MNKII 471 (centre), and by Leakey and by Shipman *MNKII 475 (bottom).
Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 143

Figure 43. Olduvai bone tools proposed by Leakey, *MNKII 502 (centre), and by Leakey and by Shipman MNKII 485 (top), MNKII 738 (bottom). Scale
bars = 1 cm. An asterisk indicates a bone tool according to this study.



144 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 44. Olduvai bone tools proposed by Leakey, *MNKII 744 (top), MNKII 888 (centre), *MNKII 923 (bottom). Scale bars = 1 cm. An asterisk
indicates a bone tool according to this study.



ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 145

Figure 45. Olduvai bone tools proposed by Leakey, MNKII 925 (top), and by Leakey and by Shipman SHKII 068–6677 (centre), SHKII 068–6684
(bottom). Scale bars = 1 cm.



146 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 46. Olduvai bone tools proposed by Leakey, SHKII 068–6685 (top), and by Leakey and by Shipman SHKII 068–6687 (centre), *SHKII 068–6688
(bottom). Scale bars = 1 cm. An asterisk indicates a bone tool according to this study.



different from those observed on the Leakey collection,
with the exception of carnivore traces and percussion
marks less represented in the control sample (Table 8). The
lower proportion of percussion marks in the control
sample is to be expected, considering the smaller size and
resistance of the original bones, which broke readily
under percussion instead of recording the impact marks.
The lower count for carnivore marks in the control sample

may be due to specimen size, as smaller bones submitted
to the action of carnivores have less chance of survival.

Removals
Bone flakes and shaft fragments described by Leakey

and by Shipman as tools, record a significantly higher
number of removals suggestive of intentional knapping,
than do the Olduvai control sample and the experimental
assemblage (Tables 7, 10 & 11). While the Leakey/
Shipman collection may have up to 20 removals per piece,
no more than four and eight flake scars were observed on
the control sample and the experimental flakes, respec-
tively. The Leakey/Shipman collection has an average of
four removals per piece, the control sample has 1.2 remov-
als and the experimental collection has only 0.4. In partic-
ular, the frequency distribution of the number of removals
per piece reveals in the purported tools a marked bimodal
trend absent in the other collections (Fig. 49). The second
peak in the Leakey/Shipman distribution, composed
of pieces having between five and 22 removals each,
accounts for nearly half of the specimens. All these pieces
come from large- to very large-sized mammals.

Removals do not occur with the same frequency on the
periosteal and medullar surfaces of the pieces from the
two Olduvai collections (Table 10–11). On specimens from
the control sample, flake scars are four times more abun-
dant on medullar than on periosteal surfaces, while on the
Leakey/Shipman specimens, a significantly higher pro-
portion of pieces with removals on the medullar surface
are only observed on pieces with a single removal. An
even more remarkable difference appears when compar-
ing the occurrence of isolated and contiguous removals in
the three collections. The Leakey/Shipman collection
bears a consistent record of specimens with a considerable
number of contiguous removals on both periosteal and
medullar surfaces and very few pieces with multiple iso-
lated flake scars. The other two collections have compara-
tively few pieces bearing contiguous removals, and in the
case of the Olduvai control sample, a high proportion of
pieces with single removals on the medullary aspect. The
number of pieces presenting overlapping removals
mimicking stepped retouch is also much higher in the
Leakey/Shipman collection than in the other samples
(Table 12). Interestingly, only in the Leakey/Shipman
collection do a consistent number of flakes (17%) have re-
movals on the same edge and on opposite aspects of the
bone flake, creating a bifacial arrangement (Table 12).
Moreover, all these pieces belong to large to very large
mammal size classes. No such pieces were found in the
control sample and only two (2%) among the experimen-

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 147

Figure 47. Proportions of mammal size classes (top) and taxa (bottom)
represented in the Olduvai bone tool and comparative collections.

Table 9. Olduvai purported tools with carnivore and hominid modifications.

Carnivore traces MNKII 1046, BKII 187, MNKII 3243, MNKII 1046, BKII 068/6680, HWKEII/3, MNKII 925, MNKII 2355, BKII 2959, MNKII
1053, BKII 068/6669, BKII 949, BKII 1053, SHKII 068/6677, MNKII 1059, FLKII 323

Cutmarks BKII 949, BKII 068/6669, MNKII 1053, BKII 1053/315, FLKII 323, MNKII 1059, SHKII 068/6677, BKII 2959, MNKII 1046,
MNKII 3243, BKII 187, BKII 068/6680, MNKII 2355, MNKII 925, HWKEII/3

Percussion marks BKII 068/6683, BKII 949, BKII 1053/315, BKII 068/6678, BKII 068/6674, BKII 2715, BKII 1605, BKII 068/6666, FLKII 45, FLKII
spit 5+, MNKII 1741, MNKII 1053, MNKII 475, MNKII 888, MNKII 1115, MNKII 1133, MNKII 1051, MNKII 1304, MNKII
1496, MNKII 502, HWKEII 4021, SHKEII 068/6681



tal flakes. The location of the removals is not significantly
different between the samples.

Additional noteworthy differences appear when analys-
ing the length of the removals. Removals exceeding
40 mm are only present on the Leakey/Shipman and

experimental flakes, and those of more than 80 mm occur
only in the former sample (Fig. 50). Also, the large majority
of the removals on the control sample are less than 10 mm
in length (Fig. 50), while those of the same size constitute
40% in the Leakey/Shipman sample and less than 10%

148 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Figure 48. a–b, Percussion marks on Olduvai purported bone tools BKII 068/6666 (a) and MNKII 1133 (b); c–d, multiple cut-marks on specimen MNKII
925; e, cut-marks on specimen BKII 068/6680; f, scoring and pits on specimen MNKII 2093. Scale bars = 1 cm.

Table 10. Number, association and location of removals on the periosteal and medullar surfaces of Olduvai shaft fragments described as tools.

No. rem. No. flakes Association Location

Periosteal Medullar Periosteal Medullar

Peri. Med. Isolated Cont. Isolated Cont. End End + side Side End End + side Side

0 24 16 55 35 45 36 37 62 57 45 59 45
1 8 17 8 0 17 0 5 0 3 10 0 6
2 9 11 1 8 0 8 8 0 1 8 0 3
3 9 6 2 7 0 6 9 0 0 1 1 4
4 3 6 0 3 0 6 2 0 1 2 0 4
5 4 3 0 4 0 3 1 1 2 0 2 1
6 3 2 0 3 0 2 1 0 2 0 1 1
7 2 1 0 2 0 1 1 1 0 0 1 0
8 1 2 0 1 0 2 1 0 0 0 1 1
9 1 0 0 1 0 0 0 1 0 0 0 0

10 1 1 0 1 0 1 1 0 0 0 0 0
14 1 0 0 1 0 0 0 1 0 0 0 0
20 0 1 0 1 0 1 0 0 0 0 0 1

rem: removals; Peri.: periosteal; Med.: medullar; Cont: contiguous.



among the experimental flakes. One might argue from
this evidence, considering the composition of the three
collections, that the length of the removals is a function of
the bone size. Analysis of removal size according to
mammal size classes reveals, however, that those from the
Leakey/Shipman collection are in all size classes signifi-

cantly longer than those on the Olduvai control sample
(Table 13). The high standard deviation observed on the
purported tools from large and very large animals, in
contrast with the low values in the control sample, is due
to the fact that these pieces record a succession of large
and small, often contiguous and overlapping flake scars.
This is supported by the frequency distribution of the
lengths of primary and secondary removals, indicating
that the putative bone tools are the only sample that
records a clear variation in size between first and second

ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158 149

Table 11. Number, association and location of removals on the periosteal and medullar surfaces of a shaft fragment control sample from Olduvai.

No. rem. No. flakes Association Location

Periosteal Medullar Periosteal Medullar

Peri. Med. Isolated Cont. Isolated Cont. End Side End End + side Side

0 66 39 72 74 54 65 69 77 78 79 52
1 9 23 8 1 23 0 7 2 10 0 13
2 2 8 0 2 3 5 2 0 1 1 6
3 2 9 0 2 0 9 1 1 0 0 9
4 1 1 0 1 0 1 1 0 1 0 0

rem: removals; Peri.: periosteal; Med.: medullar; Cont: contiguous.

Figure 49. Frequency distribution of the number of removals per piece in
the Olduvai control sample (top), flakes from the experimental breakage
of elephant bones (centre), and Leakey/Shipman collection (bottom).

Figure 50. Length of the removals from the experimental flakes (top), the
Olduvai control sample (centre), and the purported bone tools (bottom).

Table 12. Frequency of primary/secondary and monofacial/ bifacial removals on the three samples analysed.

Collection Succession Arrangement

First generation Second generation Monofacial Bifacial

Experimental 34 (71%) 14 (29 %) 12 2
Olduvai control 86 (89%) 11 (11%) 81 (100%) 0
Leakey/Shipman 312 (59%) 216 (41%) 84 (83%) 17 (17%)



generation flake scars (Fig. 51). The anomalous size of the
removals on the Leakey/Shipman sample is also indicated
by the correlation of the longest removal with the flake
size index, showing that a number of pieces from this
collection have removals that exceed the size expected on
the basis of the removal/size ratio observed in the other
collections (Fig. 52).

According to flake removal data, we identify a reduced
number of bone tools (n= 36) in Leakey’s (1971) pur-
ported bone tool collection of 123 specimens. Shipman
(1989) identified 41 of these as true bone tools, and even
though we identify fewer, some of these pieces are inter-
estingly not considered as tools by Shipman. Bone tools
identified by us derive from seven sites (DK, MK, HWK
East, SHK, MNK, FC West, BK and FCKII), though site
FCKII is not recorded in Leakey’s 1971 monograph. We
concur that the majority of bone tools occur in MNK and
BK, in middle-upper Bed II (Table 2).

In sum, our results seem to identify within the purported
tools, a cluster of pieces that appear idiosyncratic when
compared with the available non-artefactual analogues.
They consist of fresh bone shaft fragments and epiphyseal
pieces from large and very large mammals, bearing five or
more flake scars, some of which are contiguous, with one
or more anomalously invasive primary removals. Table 14
lists the 37 specimens from the Leakey collection that have
more than five removals. Most of them share the features
mentioned above and reveal a particularly high propor-
tion of bifacially arranged removals, accounting for 14 of
the 17 cases recorded in this collection. Importantly, they
are virtually unaffected by carnivore damage. The anthropic
origin of the removals on many of the specimens belong-
ing to this sample is supported by the few pieces on which
the removals are the likely result of carnivore activity
because of their close proximity to typical carnivore
damage (BKII 869, MNKII 2093, MNKII 2360, MNKII 2369,
MNKII 3335, SHKII 068/6687). The removals on these

150 ISSN 0078-8554 Palaeont. afr. (December 2004) 40: 95–158

Fi
gu

re
51

.L
en

gt
h

of
th

e
pr

im
ar

y
an

d
se

co
nd

ar
y

re
m

ov
al

s
re

sp
ec

tiv
el

y,
fr

om
th

e
O

ld
uv

ai
co

nt
ro

ls
am

pl
e

(A
–B

),
ex

pe
ri

m
en

ta
lf

la
ke

s
(C

–D
),

an
d

th
e

pu
rp

or
te

d
bo

ne
to

ol
s

(E
–F

).

Figure 52. Correlation between the longest flake scar and the shaft
fragment size index.



pieces may be contiguous but are rarely invasive.
Clearly, a number of these pieces are difficult to interpret

as bone tools. This is certainly the case for the distal
epiphyses of humeri (e.g. BKII 2382, BKII 200, MNKII 475)
that do not seem to differ from similar fragments found at
Olduvai and in numerous other collections.

Although bearing a striking amount of invasive contigu-
ous removals, which could suggest their intentional
shaping, some other pieces may also be explained as the
outcome of bone breakage for marrow extraction (e.g.
BKII 068-6668, MNKII 1133, DKI 067-4259 and perhaps
HWKEII 249, MNKII 923, BKII 068-6674 and MNKII 1117,

all from sites considered as occupation floors by Leakey).
These flakes may have been detached as a consequence of
repeated percussion made with a hammer or against an
anvil, inflicted obliquely on the broken end of a shaft
piece. In one specimen (MNKII 1133), these percussions
were followed by a strike inflicted on the cortical bone
close to the broken end