Content uploaded by Eleanor M L Scerri
Author content
All content in this area was uploaded by Eleanor M L Scerri on May 19, 2021
Content may be subject to copyright.
The Aterian and its place in the North African Middle Stone Age
Eleanor M.L. Scerri
Centre for the Archaeology of Human Origins (CAHO), 65A Avenue Campus, University of Southampton, Southampton SO17 1BF, UK
article info
Article history:
Available online xxx
abstract
The Aterian is a frequently cited manifestation of the Middle Stone Age (MSA) of North Africa, yet its
character and meaning have remained largely opaque, as attention has focused almost exclusively on the
typology of ‘tanged’,or‘pedunculated’, lithics. Observations of technological similarities between the
Aterian and other regional technocomplexes suggest that the Aterian should be considered within the
wider context of the North African MSA and not as an isolated phenomenon. This paper critically reviews
the meaning and history of research of the Aterian. This highlights a number of serious issues with
definitions and interpretations of this technocomplex, ranging from a lack of definitional consensus to
problems with the common view of the Aterian as a ‘desert adaptation’. Following this review, the paper
presents the results of a quantitative study of six North African MSA assemblages (Aterian, Nubian
Complex and ‘MSA’). Correspondence and Principal Components Analyses are applied, which suggest
that the patterns of similarity and difference demonstrated do not simplistically correlate with tradi-
tional divisions between named industries. These similarity patterns are instead structured geographi-
cally and it is suggested that they reflect a population differentiation that cannot be explained by
isolation and distance alone. Particular results include the apparent uniqueness of Haua Fteah compared
to all the other assemblages and the observation that the Aterian in northeast Africa is more similar to
the Nubian in that region than to the Aterian in the Maghreb. The study demonstrates the existence of
population structure in the North African MSA, which has important implications for the evolutionary
dynamics of modern human dispersals.
Ó2012 Elsevier Ltd and INQUA. All rights reserved.
1. Introduction
The African Middle Stone Age (MSA), beginning around 300 ka,
is associated with the transition from generalised lithic industries
to distinct regional traditions and the apparently widespread use of
composite technology (e.g. Barham, 2001;Phillipson, 2005). In the
north of Africa, these technological changes are most often iden-
tified with the highly distinctive tanged tool assemblages of the
middle and later MSA. Collectively referred to as the Aterian, these
assemblages are frequently seen as a proxy for modern human
dispersals out of sub-Saharan Africa (Caton-Thompson, 1946;
Cremaschi et al., 1998;Van Peer, 1998;Kleindienst, 2001), perhaps
into Eurasia (Garcea, 2011,2012) and as even as one of the earliest
manifestations of identify and ethnicity (d’Errico et al., 20 09;Balter,
2011).
Recent research has provided significant new insights into the
Aterian by revealing its antiquity (Richter et al., 2010), emphasising
variability/low technological affinity between Aterian localities
(e.g. Tillet, 1995;Bouzouggar, 1997;Hawkins, 2001;Bouzouggar
et al., 2002;Clark et al., 2008;Nami and Moser, 2010;Garcea,
2011) and highlighting degrees of specific attribute sharing
between the Aterian and other manifestations of the North African
MSA (Gruet, 1954;Clark, 1980,1982;Betrouni, 1997;Cremaschi
et al., 1998;Van Peer, 1998;Van Peer et al., 2003;Mercier et al.,
2007;Van Peer and Vermeersch, 2007;Aouadi-Abdeljaouad and
Belhouchet, 2008;Nami and Moser, 2010;Richter et al., 2010).
Observations of interstratification at several sites between tanged
tool assemblages and those without these types (Betrouni, 1997;
Aouadi-Abdeljaouad and Belhouchet, 2008;Nami and Moser, 2010)
have, however, simultaneously called the Aterian’s chronostrati-
graphic integrity and value as a named technocomplex (sensu
Clarke, 1968) into question (e.g. Richter et al., 2010;Linstädter et al.,
2012).
Given the lack of a regional framework for the lithic variability of
the North African MSA, it has been difficult to contextualise the
recent evidence in terms of regional scale concerns. It is clear that
a comprehensive comparison of primary data from across North
Africa is required. This paper presents a critical review of the
current evidence together with a quantitative pilot study of the
E-mail address: eleanor@soton.ac.uk.
Contents lists available at SciVerse ScienceDirect
Quaternary International
journal homepage: www.elsevier.com/locate/quaint
1040-6182/$ esee front matter Ó2012 Elsevier Ltd and INQUA. All rights reserved.
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Quaternary International xxx (2012) 1e20
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
spatial organisation of diversity in the North African MSA. The term
‘Aterian’is retained to describe assemblages with tanged tools.
Pending the substantial revaluation of assemblages characterised
by the presence of these tools and related technological features,
the term ‘Aterian’remains a useful referent and organisational
heuristic.
2. Research history
2.1. Discovery and typological characterisation
MSA assemblages with tanged tools currently have a known
distribution from Morocco to the Western Desert of Egypt, and as
far south as the Sahel (Fig. 1). Chronologically, they date from at
least Marine Isotope Stage 6 (MIS), lasting until MIS 3 (Hawkins,
2012), and in the literature are seen simultaneously as an adap-
tation to local conditions (e.g. Hawkins, 2012) and a technological
diversification of sub-Saharan origins (e.g. Caton-Thompson, 1946;
Clark, 1993). Definitions of the Aterian are, however, vague,
contradictory and problematic. As these definitions are inseparable
from the research history of the Aterian, it is necessary to detail
the construction of ‘La Question Aterienne’(Balout, 1955). Lithics
which would later be described as tanged tools were first pub-
lished as specifically Palaeolithic artefacts by Carriere (1886) in
Algeria, who described them simply as ‘Mousterian’. References to
tanged tools were occasionally published after this date (e.g.
Debruge, 1910;De Morgan et al., 1910) however they were first
specifically addressed by Debruge (1912) via a presentation of the
unpublished memoirs of fist name M. Latapie, a gendarme who
pioneered prehistoric research in the Tebessa region of Algeria
(Vaufrey, 1955). In Latapie’s memoir, unifacial foliates and tanged
tools from Ain el Oubira (Tebessa, Algeria) were described, and
were considered by Debruge as important enough to represent
a specific industry. Ultimately Debruge conceded that a ‘Loubiran’
industry was still too ill defined to warrant designation as
a distinct archaeological industry.
Although Debruge offered the first considerations of the
Aterian, it was Latapie’s protégé, Maurice Reygasse, who ulti-
mately brought the Aterian to prominence. Reygasse, as was
typical of his time (Jones, 1997), was a diffusionist who was highly
influenced by Breuil’s(1912)Les Subdivisions du Paléolithique
supérieur et leur signification (Breuil, 1912), which argued that the
Capsian ‘Middle Aurignacian’may have influenced the French
Aurignacian by way of Spain. Reygasse had been working on the
archaeological validation for Breuil’sview(Reygasse, 1920,1922),
when he coined the term ‘Mousterian with tangs’(Reygasse, 1919)
in his review of the prehistory of the Maghreb. In 1919, Reygasse
himself discovered tanged artefacts of Pleistocene age at the
eponymous site of Bir el Ater, in the Tebessa region of Algeria.
Despite recognition that there was not enough data through
which to clearly define it (Debruge, 1912), Reygasse presented the
Aterian in 1922 at the Congrès Annuel de l’Association Française
pour l’Avancement des Sciences in Montpellier (Reygasse, 1922;
Holl, 2005) as a new and advanced facies of the Mousterian. It is
perhaps no coincidence that, like Breuil’s Aurignacian, the Aterian
assumed a parallel role, this time as the progenitor of the
Fig. 1. Map of key Aterian localities illustrating their distribution across North Africa. Comparison of these positions with oases locations and trading depots indicates a high
correlation between these and Aterian site locations, suggesting a possible element of survey bias.
E.M.L. Scerri / Quaternary International xxx (2012) 1e202
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Solutrean’s tanged points and foliates, and also by way of Spain
(Caton-Thompson, 1946;Smith, 1966). Although the lack of data
and any comprehensive description should have warranted
caution, the term was immediately accepted by many French
prehistorians (Holl, 2005).
This ‘interpretation before description’approach illustrates
the pursuit of an ideological goal in which the Maghreb was
centre stage to an Africa at the fringes of Europe rather than the
rest of its own continent (Robertshaw, 1990), echoing the
concerns of the colonialist movement (Sarvan, 1985). Despite the
prominence of the Aterian industry in the Maghreb, there was
asignificant lack of comprehensive or even consistent description
(Balout, 1955)anddefinitions of the Aterian were provided on the
basis of individual experiences of scholars and information from
other geographic contexts (Garcea, 2010;Bouzouggar and Barton,
2012). There were significant numbers of Aterian site descrip-
tions as the enthusiasm generated by the newly hailed Aterian
launched a rush to locate further evidence. El Hank, Cap Blanc, El
Khenzira, Ain Fritissa, Grotte de carrière Anglade and Aïn Meth-
erchem were all Aterian sites of major importance excavated in
the 1930s (Antoine, 1932;Ruhlmann, 1936,1939;Vaufrey, 1955).
A review of the individual reports of these excavations, however,
reveals very little about site comparability and a heavy depen-
dence on tangs, and to a lesser degree, bifacial foliates, as
exclusive cultural markers. These events, together with the
dispersal of the original Bir el Ater assemblage amongst different
institutions meant that by 1939, Marcel Antoine insightfully
described the Aterian as a meaningless name (Antoine, 1939),
despite the fact that there was a recognition that the Aterian was
more than merely a ‘Mousterian with tangs’(cf. Breuil, 1931;
Antoine, 1934).
2.2. Typological perspectives on the Aterian
Following her seminal work with Elinor Gardner at Kharga
Oasis, Egypt, Caton-Thompson (1946, 1952) provided the first
objective and comprehensive techno-typological descriptions of
the Aterian across North Africa (Fig. 2). This landmark study
argued that the Aterian was a separate entity rather than
a‘Mousterian’facies and correlated the Aterian with dry phases,
contra her contemporary, Armand Ruhlmann (Barton et al., 2009).
In doing so, she hailed the Aterian as a ‘desert adaptation’,an
interpretation which remains common in current research (e.g.
Cremaschi et al., 1998;Garcea, 2011,2012). Perhaps most contro-
versially, Caton-Thompson placed the Aterian within a wider
African context. Her observations belie important considerations
about technological convergence as well as cultural diversity.
Unfortunately, her placing of the Aterian within an African
demographic context were not immediately considered, along
with her seminal observations as to what the Aterian may look like
without tangs.
Caton-Thompson’s prescient observations were arguably trun-
cated by the systemic failures of Bordian typological systematics
rather than wilful disregard. Whilst the general shortcomings of the
system are widely discussed (e.g. Débenath and Dibble, 1994;
Bisson, 2000), for the Aterian its application was especially prob-
lematic because Bordes’system subsumes all tanged tools under
two types (57 and 58), giving the tang primacy over the tool
(Hawkins, 2001) and negating the role of variability in form,
manufacture and cutting edge. Furthermore, the system was
adopted without due consideration of the fact that the Aterian
contains types missing from Bordes’list. This meant that key early
observations about variability such as Caton-Thompson’swere
instantly undermined and the ‘tang as index fossil’concept was
reinforced.
Systems such as Tixier’s (1967a, 1967b) attempted to accom-
modate these issues by expanding on the range of types associ-
ated with tanging. By maintaining a focus on tanged tool
variability, the underlying assumption that the Aterian is unique
only due to the presence of tanged tools, and to some extent
bifacial foliates, was again emphasised. Tixier’s approach was
justified by the apparent observation that Aterian assemblages
Fig. 2. Aterian artefacts from Kharga Oasis KO6E (photograph by the author, reproduced with kind permission from the British Museum).
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 3
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
featured a tanged component of around 25%, however this
assertion is now disputed. Tanged components can be as low as
1.3% (Bouzouggar and Barton, 2012) and some assemblages have
been described as Aterian on the basis of a single tanged piece
(e.g. Clark, 1982 contra Gruet, 1954). This somewhat uncomfort-
able designative strategy has led to renewed interest in what the
Aterian may look like without tangs (e.g. Kleindienst, 2001).
Unfortunately this important question has been muddied by
a series of reports on sites described as Aterian on the basis of
proximity to diagnostic Aterian sites (e.g. di Lernia, 1999), or on
the basis of other index fossils such as bifacial foliates (e.g. Robert
et al., 2003). As it is the patterning of tanged tools that has thus
far guided understanding of the Aterian, it is incumbent upon
such researchers to justify this use of the term Aterian when these
types are not present. In the absence of a systematic re-review of
the evidence, it is difficult to evaluate whether the Aterian is
recognisable without tanged tools.
3. Characteristics of the Aterian
Recent research has significantly contributed to understanding
Aterian assemblages by associating them with anatomically
modern human skeletal remains (Ferembach, 1976a,1976b;Hublin,
1992) (Table 1) and in some instances, the traces of apparently
symbolic behaviour, such as intentionally perforated shell beads
(d’Errico et al., 2009) and stones (Tillet, 1978,1983,1984), pigment
use (Nami and Moser, 2010), structures (Tillet, 1983;Clark et al.,
2008) and a bone industry (El Hajraoui, 1994;Nami and Moser,
2010).
The first chronometric age estimates (e.g. Débenath et al.,
1982;Débenath et al., 1986;Texier et al., 1988) are now known
to be contaminated and therefore incorrect (Hawkins, 2001;
Barton et al., 2009). More recent thermoluminescence (TL),
optically simulated luminescence (OSL) and uranium series
(U-Series) dates, however, demonstrate the appearance of the
Aterian in northwest Africa by late Marine Isotope Stage (MIS) 6
(Table 1).
Although most of these dates are specific to the Maghreb, Jacobs
et al. (2012) have proposed two phases for the Aterian: an early
Aterian stage supported by the MIS 6-5b dates, and a later Aterian
in MIS 5a-4. Importantly, these Aterian phases are themselves
bracketed by an early MSA/‘Mousterian’dating to 171 12 ka at Ifri
n’Ammar (Richter et al., 2010), 122 5 ka at Contrebandiers and
112 8 ka at Dar es Soltan (Jacobs et al., 2011,2012) and a later
MSA/‘Mousterian’dated to w60e50 ka at El Harhoura 2 and Dar es
Soltan I (Barton et al., 2009;Jacobs et al., 2012).
MIS 6 is typically regarded as being an arid period in North
Africa, but it has recently been shown to have included periodic
wetter episodes in at least certain areas (Drake et al., 2011;Drake
et al., in this volume). MIS 5, with which most Aterian sites are
associated, is an interglacial period usually associated with
climatic amelioration in the Sahara (Blome et al., 2012). MIS 4,
however, is correlated with the onset of hyper arid conditions in
much of the Sahara, despite apparently being associated with
some Aterian sites in Libya and Egypt (Cremaschi et al., 1998;
Hawkins, 2012). Whilst it is clear that the resolution of the
Palaeoenvironmental and archaeological records is still poor,
a number of inferences have been made regarding the Aterian on
the basis of climate. The association of Aterian lithics with a rich
symbolic material culture has for example led to important
questions being asked regarding the conditions for such behav-
ioural changes in human history (e.g. d’Errico et al., 2009). The
wide distribution of the Aterian requires a continuing focus on
lithic technology, however, as the most abundant category of
evidence for this technocomplex (and indeed for the Palaeolithic
in general). On these terms, foremost amongst considerations of
‘what the Aterian is’are the numerous references to the Aterian as
a‘desert adaptation’(e.g. Caton-Thompson, 1946;Clark, 1980;
Cremaschi et al., 1998;Garcea, 2004,2012), presumably in contrast
to the non-Aterian MSA.
Table 1
Human skeletal remains associated with the Aterian at various Moroccan cave sites in the Temara region.
Skeletal discovery Site Year of discovery Reference
Juvenile maxilla fragment
Adult teeth
Mugharet el Aliya 1939 Senyurek, 1940
Mandible and cranial fragments
Juvenile skill and partial skeleton
Contrebandiers/Temara 1. 1956
2. 1975
3. 2010
1e2. Roche and Texier 1976
3. Balter, 2011
Cranial fragments of three individuals Dar es Soltan II 1974 Débenath, 1975
Canine and mandible Grotte Zourah/El Harhoura 1 1976 Débenath, 1980
Parietal fragment Taforalt 1951 Roche, 1953
Table 2
Sample of recent chronometric age estimates mentioned in the text.
Site name Location Years (BCE) MIS stage Method Reference
Ifri n’Ammar Morocco 145 9kae83 6 ka 6-5a TL Richter et al., 2010
Adrar Bous Niger w150 and w45 ka 6-3 (range) Comparative stratigraphy Williams, 2008
Dar es Soltan I Morocco 110 ka þ5d OSL Barton et al., 2009
Rhafas Morocco 70e80 ka 5a TL/OSL Mercier et al., 2007
Taforalt Morocco 91e73 ka 5b-5a OSL/TL/U-Series Bouzouggar et al., 2007
Schwenninger et al., 2010
El Mnasra Morocco 109e95 ka 5c-5b Single-grain OSL Jacobs et al., 2012
Oued el Akarit Tunisia 90 ka 5b TL Roset, 2005
Uan Tabu Libya 61 10 ka 4 TL/OSL Cremaschi et al., 1998
Kharga Oasis Egypt 90e40 ka 5be3 U-Series on capping tufas
and carbonates bracketing deposits
Smith et al., 2004a,2004b
Kleindienst et al., 2009
Contrebandiers Morocco 107 4e96 4 ka 5d-5b Single-grain OSL Jacobs et al., 2011
E.M.L. Scerri / Quaternary International xxx (2012) 1e204
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
3.1. The Aterian as an adaptive response
The arguments supporting the Aterian as a desert adaptation
centre on the fact that many Aterian sites are apparently in close
proximity to ancient sourcesof water such asartesian springs,lakes or
rivers(e.g. Garcea, 2012), thattanged tools areprojectile points(Clark,
1989), perhaps suggesting an adaptation to drier and therefore more
open environments, and the fact that Aterian sites are sometimes
associated with glacialperiods (Richter et al.,2010;Blome et al., 2012;
Hawkins, 2012). Whilst evidence may yet be found for specific‘dry-
environment’adaptations in the Aterian, suggestions premisedon the
above arguments warrant caution, particularly where the resolution
of the record does not permit the detailed observation of adaptive
changes through time. Whilst it is true that many Aterian sites are
found it close proximity to water, the suggestion that Aterian groups
were tethered to water sources implies theopposite to adaptation (cf.
Cancellieri and di Lernia, in this volume). The correlation of sites with
fluvial and lacustrine features is hardly distinctive of the Aterian,
being found in many areas (see e.g. Groucutt and Petraglia, 2012 on
Arabia). The apparent correlation of archaeological sites and fluvial
and lacustrine networks is common even in climates associated with
plentiful supplies. It is debatable, on these terms, whether proximity
to water could be considered ‘adaptive’behaviour without evidence
for specialist activities. Adaptation in a substantive sense in ‘desert’
conditionsmight, for example, involve features such as water storage.
However, to date, there is no conclusive evidence of any Pleistocene
populations living on conditions analogous to modern deserts
(Groucutt and Blinkhorn, in this volume). It is finally worth noting
that modern roads and settlements are often associated with rivers
and lakes, so there may also be an element of survey bias (see Fig. 1).
The majority of dated Aterian sites are associated with the
climatic amelioration of MIS 5 (Brooks et al., 2005;Osborne et al.,
2008;Drake et al., 2011;Blome et al., 2012;Drake et al., in this
volume). If non-Aterian sites were furthermore concurrent with
Aterian sites throughout the North African MSA, as is suggested by
limited dated sites (e.g. Mercier et al., 1999;Van Peer et al., 2003;
Ramos et al., 2008), it would imply that both Aterian and non-
Aterian MSA groups may have been subjected to similar selective
environments. Rots et al. (2011) have argued that the whole of the
North African MSA displays an increased concern with technolog-
ical reliability through the hafting of tools. Hawkins (2012) has
made the same argument specifically for the Aterian. The evidence
therefore suggests that concurrent Aterian and non-Aterian
assemblages reflect comparable behavioural changes.
The evidence for tanged tools as projectile weapons is also
tenuous. The ubiquitousness of tanged tools in Aterian assemblages
first led Caton-Thompson to suggest that a ‘new and formidable
mechanical force was let loose upon the African world’(Caton-
Thompson, 1946, 88). Shea (2006), however, suggested that Aterian
points were unlikely to have been projectile weapons but may
instead represent thrusting spear tips. Microwear studies have since
indicated that tanged tools may have been used on soft hides,
possiblyas scrapers (Bouzouggar and Barton, 2012). Iovita (2011) also
suggests that patterns of tanged tool resharpening/edge rejuvena-
tion may reflect their use as tanged scrapers or knives. Massussi and
Lemorini (2007) haveargued that the tang itself was sometimes used
for scraping functions, although such a function is perhaps likely to
represent recycling activities after the retooling of hafts. It would be
interesting tosee if purported use-wear wear was consistently found
on the tanged portions of significantly reduced tools. Unfortunately
no study to date has combined the analysis of use wear and retouch
intensity with the effect of retooling hafts on Aterian assemblages.
With its function as a hafted spear tip in doubt, tanging itself has
been perceived as a hafting-related adaptation for use with bind-
ings, indicating a lack of resin-bearing trees. Although the ‘wet
conditions’described in palaeoclimatic literature (see above) are
somewhat confusingly associated with ‘dry adapted technology’in
the archaeological literature, it should be made explicit that the
time-averaged conditions associated with the Aterian ranged from
savannah to semi arid environments. According to the Köppen
climate classification, these environments are qualitatively distinct
from deserts in terms of both amounts of precipitation and vege-
tation type. On these terms, the co-presence of basally thinned and
shouldered tools with tanged tools, together with the evidence
demonstrating the increased use of resin for hafting in southern
Egypt in the Upper Pleistocene (Rots et al., 2011) undermines the
argument for tangs as specifically ‘dry-adapted’.
It is rather perhaps the association between the Aterian and
climatic oscillation (cf. Hawkins, 2012) that is significant, especially
since the Aterian appears to be a flexible technology with an
emphasis on tool reliability (Bouzouggar and Barton, 2012). With
diverse forms of hafting and manufacturing techniques, Aterian
groups may instead be characterised as flexible generalists in an
uncertain landscape. If this is the case, then it is necessary to
understand the organisation of technology within the Aterian in
terms of whole assemblages.
3.2. Beyond tanged tools: the Aterian in context
Comparability of MSA assemblages is unfortunately impacted by
the effect of incompatible regional nomenclature and different
methodologies. The North African MSA is variously described as an
‘undifferentiated’or ‘generalised’backdrop to the Aterian (cf.
Willoughby, 2007), or inappropriately categorised using European
terminology. The MSA in the Maghreb is for example described
variously as ‘Classic Mousterian’(e.g. Betrouni, 1997,2001;Ramos
et al., 2008), Mousterian of Acheulean tradition (MAT) (Aumassip,
2004), ‘Denticulate Mousterian’and ‘Ferassietype’(Aumassip, 2004).
In the Sahara assemblages with Levallois debitage (and some-
times without) but few or no retouched tools are typically associ-
ated with the MSA in the place of a ‘Mousterian’. At the Libyan site
of Uan Afuda, undiagnostic lithic material is attributed to the MSA
on the basis of some ‘pseudo-Levallois’technology and OSL dates
between 90 and 69 ka (di Lernia, 1999). Similar to the Uan Tabu
assemblage is the small collection from Wadi Adroh, also in the
southwestern Fezzan region of Libya (Van Peer, 2001). Following
Tillet (1985),Van Peer (2001) has theorised that these assemblages
represent a non-Aterian MSA with no connection to ‘Mousterian’
industries described in the Maghreb coastal and inland regions.
Aumassip (1986) has, however, referred to a Mousterian of
Acheulean Tradition in the Sahara as well as the Maghreb, and
references to a ‘Denticulate Mousterian’have been made at Tarf
H’Mer in Mauritania (Pasty, 1999).
Terminology varies again in southern Mauritania and northern
Senegal where references are made to an ‘evolved Palaeolithic’
described as containing tanged tools, bifacial foliate points and
blades (Camara and Duboscq, 1997). The MSA/Mousterian is further
described as ‘Mousteroid’. It is unclear why the former is not
designated as Aterian, however it is theorised (cf. Tillet, 1997)to
represent a ‘Mousteroid’acculturation to an intrusive Aterian and
has been labelled ‘Tiémassassian’.
In northeast Africa, both the Nilotic region and the high desert
(Olszewski et al., 2010) are characterised by the presence of the
Nubian Complex, a technocomplex which has been defined in terms
of specific Nubian Levallois reduction methods (Guichard and
Guichard, 1965). It is hypothesised that the Nubian Complex devel-
oped from a local Nilotic Lupemban (Marks, 1968a;Van Peer and
Vermeersch, 2007;Rose et al., 2011), a technocomplex usually
associated with sub-Saharan Africa (cf. Barham, 2001;Cornelissen,
2002). The Nubian Complex is also hypothesised to have had a late
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 5
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
persistence: apparently post-MIS 5 industries such as the Khormu-
san (Marks, 1968b) may represent a late derivation of the Nubian
Complex (Van Peer,1998). The Nubian Levallois reduction method is
distinguished from classic and bidirectional Levallois point produc-
tion by the creation of a distinctively steep median-distal ridge. The
ridge is prepared either through distal divergent and/or lateral
removals and thisvariability has been related by some researchers to
chronological context (Rose et al., 2011;Usik et al., in this volume). It
is the increasing use of these methods over time, alongside the
reduction of size and frequency of Lupemban bifacial foliates thatled
Van Peer and Vermeersch (2007, 189) to describe the Nubian
Complex as ‘a changed Lupemban technology’. Of interest is the
recently demonstrated presence of Nubian Complex assemblages as
far east as Oman (Rose et al., 2011;Usik et al., in this volume).
The degree to which these various MSA assemblages reflect
related trajectories of technological development is unclear: there
is little chronological control and there are no quantitative
comparative studies. On typological grounds it is at least clear that
diagnostic assemblages such as the various ‘Mousterian’groups,
Aterian and the Nubian Complex share specific types with each
other. Without dedicated study on the attribute level, it is uncertain
whether similarities and differences reflect culture diversity, task
specialisation, ecological adaptation and/or isolation by distance or
merely inappropriate and incompatible analytical procedures.
3.3. Attribute sharing in the North African MSA
It is clear from the evidence that multiple trajectories of behav-
ioural change are represented in the North African MSA. Some may
be unique, autochthonous and isolated developments. For example,
the Khargan in northeast Africa (Caton-Thompson,1952;Hester and
Hobler, 1969), a flake industry in which truncations are frequently
used to form tools, and the sequence of Haua Fteah in Cyrenaica
(McBurney, 1967) seem to be local developments. Although the
Aterian has been tentatively described at Haua Fteah, like the Nile
Valley, the Aterian’s presence at this site is questionable (Débenath
et al., 1986;Hawkins, 2001;Moyer, 2003;Reynolds, in this volume)
and the lithic material is argued to resemble Levantine industries
more than African ones (cf. Moyer, 2003). Both the Nile Valley and
Haua Fteah document a long human presence, which has been
speculated to have prevented Aterian incursion into that territory
(Kleindienst, 2000 contra Garcea, 2012 who makes purely adaptive
arguments). If this is true, it is the most compelling evidence of
population structure in the North African MSA.
Other trajectories of behavioural change as represented in lithic
technology may represent the fragmentation, dispersal and diversi-
fication of related populations. The Nubian Complex, shares a number
of featureswith the Aterian.Like the Nubian Complex, the Aterian has
also been associated with the Lupemban technocomplex of sub-
Saharan Africa (Clark, 1993;Kleindienst, 2001;Garcea, 2004). Both
Kharga Oasis and Adrar Bous (Clark et al., 2008)havebeenassociated
with bifacial foliates strongly reminiscent of the Lupemban (Fig. 3).
Equatorial origins for the Aterian have in fact been postulated since
Caton-Thompson (1946). Nubian Levallois methods are also found in
Aterian contexts in Libya (Cremaschi et al., 1998), Algeria (Van Peer,
1986) and as far west as Mauritania (Pasty, 1997). The early Nubian
Complex, like the Aterian, also features a bifacial component and
basal thinning (Van Peer and Vermeersch, 2000), although these
features are found in many African MSA contexts (McBrearty and
Brooks, 2000). On the basis of all these similarities, Van Peer (2001)
has proposed that Nubian Levallois techniques present in the
Aterian indicate a dispersal and subsequent adaptation of intrusive
‘Nubian’groups into the Sahara during the last interglacial.
Small centripetal Levallois cores, often perceived to be a feature
of the northwestern Aterian (Bouzouggar, 1997;Bouzouggar and
Barton, 2012) are also seen in Aterian sites further east (e.g. Wadi
Gan in Libya, Site 8708, Site 8751, Western Desert of Egypt, author’s
observations esee Fig. 3) as well as in many other MSA settings,
such as Porc Epic, Ethiopia (Groucutt, pers. comm.). Several studies
(e.g. Betrouni, 1997;Aumassip, 2004) point to non-Aterian MSA
assemblages in North Africa with ‘small’Levallois debitage (e.g. Sidi
Saïd, Oued el Akarit, general surface sites from the southern
Sahara). This tendency for diminutive proportions extends to small
Nubian Levallois cores (Fig. 4) in Aterian contexts both in the
Egyptian western desert (author’s observation), as well as south-
western Algeria (Alimen and Chavaillon, 1956). Other Aterian sites
conversely feature much larger artefacts (an unusually large tanged
tool example given by Lhote in 1944) as well as non-diminutive
cores (e.g. Kharga Oasis). It is not clear how far raw material is
influencing this variation.
Clearly, a new framework is required for the whole of the North
African MSA in order to contextualise individual sites and assem-
blages. The provision of such a framework is an ambitious task in
terms of both scale and the identification of appropriate heuristics.
Since culture is multiscalar (Foley and Mirazon Lahr, 2011), deter-
mining the nature of a prehistoric industry is dependent upon the
research question and its scale of inquiry. At the macro scale,
culture is arguably the result of the evolutionary dynamics of
populations manifested as social boundaries (Wobst, 1974;Henrich
and Boyd, 1998;Richerson and Boyd, 2004;Le Galliard et al., 2005).
On these terms, and on the basis of the above review, a regional
scale study oriented around the concept of social and ecological
boundaries offers coherency and a key unifying heuristic. This
paper presents an exploratory pilot study of such an approach.
4. Methodology
Research questions are structured into the pilot study as two
primary concerns:
Fig. 3. Foliate from Kharga Oasis (right eafter Caton-Thompson,1946) with Lupemban
foliate (left) from Angola (after Clark, 1966).
E.M.L. Scerri / Quaternary International xxx (2012) 1e206
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
1. To test whether the Aterian was a coherent climate-adapted
trajectory within the North African MSA as suggested in the
literature: as the simpler explanation, it is hypothesised that
the Aterian is not a discrete chronostratigraphic unit and that
tanged tools cannot serve as a main criterion for the definition
of an ‘Aterian’technocomplex (e.g. Richter et al., 2010). This
will be tested quantitatively through a multivariate interre-
gional comparison of a spatially and temporally representative
sample of North African MSA assemblages.
2. To establish whether Aterian and wider MSA variability was
primarily driven by environmentally or socially constituted
adaptations: as the simpler explanation, it is hypothesised that
distributions of shared and non-shared attributes reflect an
ecologically structured isolation-by-distance, however struc-
ture driven by social boundaries cannot be discounted. The
structure of attribute sharing between assemblages in the
sample will therefore be quantified against model expectations.
This study bases data analysis on a model concerning group
interaction dynamics. At a high level this is a cultural ecology model
seeking to understand the conditions of social group boundary
formation (e.g. Foley, 2004;Foley and Mirazon Lahr, 2011),
a behaviour argued to have evolved at least with the earliest
modern humans (Gamble, 1998;Wiessner, 1998;Garrigan et al.,
2007;Atkinson et al., 2009). At the level of Middle Range Theory
(sensu Binford, 1977), the model recognises that gross measures of
similarity/difference between assemblage groups provide little
reflection of the intricacies of social boundaries, as demonstrated
by a significant body of ethnographic work and archaeological
theory (e.g. Hodder, 1979;Wiessner, 1982,1983,1984,19 97;
Tostevin, 2006,2007). Instead, ethnological and theoretical
approaches have suggested that the presence of structured social
differentiation can be inferred from the retention of material
culture distinctions, whilst others appear to cross divides (e.g.
Wiessner, 1983;Tostevin, 2006,2007;Vanhaeren and d’Errico,
2006). Whilst the specificities of which material culture styles
represent group identity are impossible to predict, the vectors of
such stylistic information have been repeatedly shown to occur on
artefact classes that are durable, whose attributes are most visible
from a distance and likely to travel along the edges of foraging
ranges where strangers interact the most (Hodder, 1979;Wiessner,
1982,1983,1984,1997). Points in particular have been identified as
vectors of such style (e.g. Wiessner, 1982,1983).
In the light of this theoretical and ethnographic work, the
methodology of this study examines the levels of similarity/
difference between comparable heuristic stages of the chaîne
opératoire on comparable artefact classes. Blanks (i.e. unretouched
flakes and blades organised by weight class), heavy tools (e.g. large
retouched lithics) and lightweight, standardised, and often hafted
tool classes are analysed separately. The latter category is divided
Fig. 4. Examples of small centripetally prepared and Nubian Levallois cores from Aterian and non-Aterian contexts in the North African MSA (photographs by the author). A: Nubian
Complex Site 1033 (Marks, 1968), Sudan; B: MSA site 8751 (Hester and Hobler, 1969), Egypt; C: Aterian site Wadi Gan (McBurney and Hey, 1955), Libya; D: Aterian Site 8708 (Hester
and Hobler, 1969), Egypt; E: Aterian Site 8735 (Hester and Hobler, 1969), Egypt. Object C (670) reproduced with kind permission from the Museum of Archaeology and Anthro-
pology, Cambridge. All other objects are from the Wendorf Collection and reproduced with kind permission from the British Museum.
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 7
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
into lightweight undifferentiated retouched tools, tools with
observable hafting modifications (e.g. tangs, basal thinning,
shouldering) and points (bifacial foliates, Levallois points and
retouched convergent flakes). Each artefact was then examined
through an attribute analysis. This procedure broadly follows
Tostevin (2006, 2007, 2009) who suggests independent domains of
action within the chaîne opératoire. As the attributes reflecting the
chaîne opératoire are interdependent within domains, but inde-
pendent between them, this differentiated comparison between
assemblages is rendered more sensitive to the different sources of
variability affecting the manufacture of stone tools and more robust
against the problems of chaîne opératoire-based assumptions (e.g.
Bar-Yosef and Van Peer, 2009;Holdaway and Douglass, 2011). These
domains are illustrated schematically in Fig. 5. Table 3 gives an
example of the relationship between the attribute states and the
domains.
This methodology supports a set of model expectations
regarding group interaction dynamics because they have been
specifically designed to interpret different sources of variability,
from raw material constraints to social interaction. These model
expectations are illustrated in Fig. 6 and Table 4.
4.1. Analysis
Six sites from across North Africa were accordingly selected for
this analysis: Sai Island in northern Sudan (cf. Van Peer and
Vermeersch, 2007) and Bir Tarfawi 14 in the Western Desert of
Fig. 5. Schematic illustration of the different domains of analysis used in the study (illustrations by the author).
Table 3
Sample domain with associated independent action clusters with sample attributes
and specific attribute states that could be recorded. Exhaustive list is included in the
Supplementary material.
Domain name Independent
cluster
Attribute Example
attribute state
Core modification Core orientation Direction of removal Longitudinal
Number of removals
(e.g. on debitage
surface)
6
Core management Platform treatment Faceted
Surface modification Débordant
removal
Table 4
Model expectations for data pattern groupings and boundary structures with four
examples modelled in Fig. 6.
Label Between-assemblage patterns Boundary structure
a Blank production similar, non-hafted
toolkit morphology similar, point
morphologies different
Present and non-permeable
b Blank production similar, non-hafted
toolkit morphology similar, point
morphologies similar
Present and permeable
c Blanks similar, point morphologies
similar, other morphologies
different: random patterns
Not present/applicable,
difference by distance/
isolation
d All patterns different May be impermeable
boundary or isolation,
requires wider ecological
information
E.M.L. Scerri / Quaternary International xxx (2012) 1e208
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Egypt (cf. Wendorf and Schild, 1980;Wendorf et al., 1993) for the
Nubian Complex; Haua Fteah in Cyrenaica, Libya (cf. McBurney,
1967) for MSA/independent sequences; Kharga Oasis in the
Western Desert of Egypt (cf. Caton-Thompson, 1952), Wadi Gan in
Libya (cf. McBurney and Hey, 1955), and Grotte des Contrebandiers
in Morocco (cf. Roche and Texier, 1976) for the Aterian. The sampled
layers and sample sizes are illustrated in the Table 5 and site
locations in Fig. 7.
Samples consisted of cores, tools and blanks (i.e. unretouched
flakes). These were analysed using correspondence analysis (CA)
on categorical variables and principle components analysis (PCA)
for scale variables. These methods represent powerful
datamining techniques capable of exploring the interactions and
relationships between variables in cross-tabulational data by
transforming each variable in the table into a dimension of var-
iance in terms of its accompanying variables. This transformation
is defined in such a way that the first dimension created has
thelargestpossiblevariance(i.e.accountingforthemajorityof
the variability in the data), successively decreasing with each
succeeding dimension under the constraint that it be orthogonal
to (i.e. uncorrelated with) the preceding dimensions. As the
dimensions represent uncorrelated drivers for variability
these techniques have the advantage of being able to isolate
different sources such as raw material constraints, etc. Where
the variability represented factors that could not easily
be explained as purely functional, they were subjected to
ANOVAS using Tukey’s multiple comparison test against site
names, thus contextualising the variability spatially for statistical
significance at the 0.05% level. Multiple Regression was also used
to further understand the relationship between variables where
appropriate.
The data was prepared for analysis in several ways reflecting the
fact that these multivariate techniques require data to have normal
distribution. Scale variables were first normalised using Box-Cox
transformations and then standardised for analysis. Where
required, scale variables were transformed to categories for corre-
spondence analysis using a using a clustering algorithm which
ensured categories are evenly divided and weighted. All variables
with low numbers (<5%) were added to the correspondence anal-
ysis as supplementary variables so as not to be overrepresented in
the descriptive dimensions. It should be noted that some missing
variables result in a few artefacts being omitted from this analyses.
The counts in the analysis and the samples as presented previously
do not always tally exactly.
4.2. Caveats
A number of caveats should be made explicit in this study. As
a pilot study, sample sizes are small and resolution is low.
Assumptions have been made about the MIS 5 contexts of the
assemblages, which even as a temporal bracket is itself wide and
known to contain climatic variation (e.g. Drake et al., 2011). Within
the model, problems are presented by personal identity style bias,
chronological issues and assumptions about tool function. Never-
theless it is argued the above model is robust against all these
issues.
By taking a differentiated approach to lithic analysis together
with appropriate statistical methods, it is argued that meaningful
patterns can be obtained even from small datasets. Although
resolution is low, the research questions posed in this pilot study
match the resolution of the North African scale in that they search
for patterns in the data indicative of time-averaged behaviours that
transcend generations. As time-averaging mutes temporal and
spatial variation, it is argued that any observed patterns are
therefore rendered more meaningful. The assumption regarding
the contemporaneity of sites is also addressed in this way: this
research in concerned with spatial, rather than temporal variation
and seeks to understand the degrees of attribute sharing rather
than the existence of specific adaptive changes over time. Function/
style concerns are built into the approach by testing against factors
such as distance or raw material constraints to explain variability
before inferring cultural factors.
It should be made clear that some samples come from older
excavations. The majority of these assemblages exhibit a degree of
fresh chipping due to storage conditions which in some instances
could be mistaken for retouch. In view of this conservative esti-
mates for the presence of retouch were taken.
Fig. 6. 3D model illustrating the relationship of different lithic data patterns to
boundary recognition. The model expectations for boundary formation for the four
points a, b, c and d are listed in the Table 4.
Table 5
Sites and sample sizes used in this study.
Site name Location Cultural
affiliation
Layers MIS
stage
Sample
size
Sai Island Sudan Nubian
Complex
Intercalated
Gravels
5 Blanks: 9
Cores: 1
Tools: 71
Total: 81
Bir Tarfawi Egypt Nubian
Complex
BT-14 C-
excavated
5 Blanks: 166
Cores: 5
Tools: 43
Total: 217
Haua Fteah Libya MSA 34/35
31/32
5 Blanks: 133
Cores: 15
Tools: 124
Total: 272
Kharga Oasis Egypt Aterian KO6E 5-3 Blanks: 14
Cores: 78
Tools: 130
Total: 222
Wadi Gan Libya Aterian Spit 3
Spit 2
Above
Alluvium
Base of
pink silt
5 Blanks: 67
Cores: 60
Tools: 75
Total: 202
Contrebandiers Morocco Aterian Layer 11
Layer 10
Layer 9
5 Blanks: 79
Cores: 49
Tools: 256
Total: 384
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 9
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
5. Results
5.1. Cores and blanks
The results of the core analysis domain indicate that different
sources of variability affect the processes of blank manufacture. PCA
(Fig. 8) and CA for cores both reflected differences between all sites.
In the case of the CA the first component was primarily driven by
raw material quality and type (18%), suggesting that the greatest
portion of variability is driven by functional/environmental factors.
Current sample sizes for different Levallois types are currently too
small to permit evaluation solely of components considered to be
more ‘stylistic’(e.g. Nubian Levallois reduction methods), however
this consideration will be part of future analyses when sample sizes
are greater.
Despite these differences in the reduction of cores of different
raw materials, the patterning of blank attributes is very similar
between Aterian and Nubian Complex sites. As sample sizes for
blanks were larger, PCA and CA dimensions representing blank
analysis domains (e.g. both the relative proportions and dimen-
sions of blanks and flaking techniques) were subjected to ANOVA.
All tests showed no statistically significant differences between
Aterian and Nubian Complex sites (example displayed as boxplot in
Fig. 9). A further ANOVA examining the way surface convexities
Fig. 7. Location of sites used in the study (map modified from Ancient World Mapping Center). 1: Contrebandiers (Morocco), 2: Wadi Gan (Libya), 3: Haua Fteah (Libya), 4: Kharga
Oasis (Egypt), 5: Bir Tarfawi 14 (Egypt), 6: Sai Island (Sudan).
Fig. 8. Core Principal Component Analysis (PCA) score plot for the first principal component axis against the second. This PCA examined core dimensions with flaking direction and
intensity.
E.M.L. Scerri / Quaternary International xxx (2012) 1e2010
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
were used to create different lateral edge shapes, cross sections and
profiles also showed no significant differences between these
groups (Fig. 10). The analysis of blanks and cores overall illustrates
a concern with producing blanks of comparable size and shape
within Aterian and Nubian contexts but not outside it: Haua Fteah
retains a statistically significant difference in both blank size and
manufacture.
5.2. Tools
Both CA and PCA were carried out on the relevant domains of
analysis. For both analyses, lightweight and heavy tools follow the
same pattern as the blanks in terms of both the flaking techniques
and in terms of their selection from the pool of blanks (see Figs. 11
and 12). Although all sites have an equifinality of tool type
frequency and retouch type, Haua Fteah displays significantly
different patterns for attributes relating to both form and flaking
suggesting different traditions were in place to meet similar func-
tional goals. Conversely, Aterian and Nubian Complex sites cluster
together spatially, differences increasing with distance: there are
clear differences in the selection of blanks for retouch between sites
at the farthest ends of North Africa (Contrebandiers vs. Kharga
Oasis), again likely to reflect adaptation within a flintknapping
tradition that is remarkably homogenous.
5.3. Tanged tools
Since the Aterian is understood as a technocomplex defined
by tangs, an analysis of these tool types was carried out sepa-
rately. Non-Aterian samples were not included as by this defini-
tion they do not include tanged tools. In view of this problem,
a further analysis is attached to this section consisting of the
quantification of other forms of basal modification across sites
which are also common in the Aterian: shouldering and basal
thinning.
Primarily this analysis investigated whether the tanged element
was manufactured the same way across all Aterian sites. Specifically
it aimed to investigate the relationship between inactive and active
tool edges, particularly in terms of whether these portions of the
tools were manufactured similarly. PCAs and subsequent ANOVAs
for tang dimensions (width, thickness, length, ventral and dorsal
retouch, shoulder to corner width) indicate a remarkable degree of
similarity across North Africa. The PCA investigating the proportion
of tang length/width ratios to tool length/width ratios however
uncovered a variability that is likely to reflect resharpening. Iovita
(2011) has already demonstrated the significant impact of
resharpening on tanged tool form and this seems to be upheld by
this investigation. Whilst tang length is a good predictor of tool
length for mechanical reasons (the tang usually being a third of the
Fig. 10. ANOVA for Correspondence Analysis axis 1 (12%) reflecting differences in the use of platform angles, controlling thickness and blade edge length in blanks. Kharga Oasis
could not be included because of the low blank sample from this site.
Fig. 9. Boxplot of means of the first principal component axis against the second axis for blank analysis. This analysis explored flake size and morphology.
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 11
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
tool length), the additional noise in the regression is accounted for
by retouch intensity (see Figs. 13 and 14 ).
Active tool edges (largely defining typological categories) and
their manufacture were investigated through a CA quantifying the
variability caused by retouch types, tool types, active edge lengths
and variables for overall tool form and manufacture. Although there
was some geographically driven variability between the most
distant sites, tanged elements and their manufacture display
a remarkable degree of homogeneity. Variability was primarily
driven (19%) by the presence of the Levallois technique and varying
frequencies of tanged points to tanged scrapers. The resulting
ANOVA (Fig. 15) displays a statistically significant difference
matching the blank analysis results.
The final analysis considered the relationship between tanged
tools and other forms of hafting modifications. This analysis
showed that although tanged tools themselves are remarkably
homogenous, their relationships with other forms of basal modi-
fication are more complex. This is important because Aterian sites
are primarily recognised on the basis of the presence of tanged
tools. Cultural integrity of this sort is maintained when the pres-
ence of tanged tools drives the variability of sites, however when it
does not, this cultural integrity collapses: basal thinning is more
widespread in the northeast at both Aterian and Nubian Complex
sites and appears to be correlated with shouldering. Although
shouldering has been considered as an embryonic form of tanging,
this research suggests that, like basal thinning, it is instead related
to cleft hafting, a conceptually and mechanically different form of
hafting to socketing. Basally thinned and shouldered tools
appeared to be correlated with lower levels of active edge retouch
intensity compared to tanged tools, however a larger sample size is
needed to confirm this, as each site uses different raw materials
which may affect results. Perhaps more interestingly the variability
driven by the presence of these alternative forms of basal modifi-
cation is prevalent only in northeast Africa (see Fig. 16), a feature
already noted by Kleindienst (2001).
The results from this study suggest that tanged tools serve
a functional requirement explaining their remarkable homogeneity
across significant geographic distances. Their levels of retouch
intensity and resharpening also reinforce this: such continual use
significantly alters cutting edge shape, negating a continual role in
social signalling. Differences are noted at Kharga Oasis, however, in
terms of tool frequencies and levels of resharpening/edge rejuve-
nation, however for the reasons cited above these are not likely to
transcend functionality/adaptiveness.
Fig. 12. ANOVA for Correspondence Analysis axis 1 (13%) driven by flaking techniques of heavy tool classes. Aterian and Nubian Complex sites cluster together with the two most
different means pertaining to sites at opposite ends of North Africa.
Fig. 11. Heavy Tool Principal Components Analysis (PCA) score plot for the first principal component axis against the second. This PCA examined artefact size and retouch.
E.M.L. Scerri / Quaternary International xxx (2012) 1e2012
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Fig. 14. Boxplot and ANOVA of tanged tools whose lengths and shapes are impacted by resharpening. Results show that levels of resharpening and length of tangs tools are similar
only between Wadi Gan and Contrebandiers.
Fig. 13. Probability Plot determining the significance of regression residuals predicting tool length. Results show this can be accurately predicted by tang length and levels of
resharpening.
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 13
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
5.4. Points
Bifacial foliates sample sizes are too small to permit tests of
significance. Nevertheless, some consideration of foliates is given at
the end of this section. Haua Fteah could not be included in this
analysis because the required sample size for analysis was >20.
Both PCA (Fig. 17) and CA demonstrated no differences between
Contrebandiers and Wadi Gan. Conversely, Kharga Oasis clustered
with Sai Island and BT-14. The ANOVA carried out for PCA 1, driven
by retouch intensity and blade edge lengths, showed no significant
differences between Contrebandiers and Wadi Gan, following
a pattern of similarity observed in all the previous analyses (Fig. 18).
The same pattern was observed for the CA which considered
attribute values for shape. In the northeast, points showed no
significant differences between Aterian and Nubian Complex sites
both in terms of their manufacture, overall size and shape. These
results suggest that although the pool of knowledge for knapping
was broadly similar for Nubian Complex and Aterian sites, clear
differences exist between the northeast and the northwest: the
northeastern Aterian appears to be more similar to other non-
Aterian northeastern sites than other Aterian sites in the northwest.
6. Discussion
This analysis has yielded some significant results, despite con-
sisting of a relatively small sample size. The most obvious inference
is the consistent observation of difference seen in the Haua Fteah
sample compared to other assemblages. Although only layers and
spits formerly associated with the Aterian are considered here,
Haua Fteah may in fact be a technological or cultural isolate,
a hypothesis supported by its geography, climate and observations
of its sequences. Whilst new discoveries may change this view, it is
speculated here that Haua Fteah maintains distinctions either due
to geographical or cultural isolation.
At the level of core reduction, it is clear that differences between
both Aterian and Nubian Complex sites are marked, reflecting
different environments and different raw material characteristics.
Although these groups reduced cores differently, adapting them-
selves to the fracture mechanics of the various raw materials, the
resulting blanks are not significantly different in terms of overall
size and proportion. Whilst Haua Fteah remains consistently
different on these terms, blank production largelysee no significant
differences between Aterian and Nubian Complex sites. It is difficult
at this stage to claim that the lack of differences is evidence for
a shared tradition. Whilst these groups are speculated to have sub-
Saharan African roots this is not necessarily related to a single
source. It is the consistent differences shown by Haua Fteah that are
more noteworthy at this point. The clear and statistical differences
strongly suggest a completely separate population.
Heavy and lightweight tool classes show a different pattern.
Where blanks show no significant differences between Aterian and
Nubian Complex sites, at the level of tool analyses, a consistent
differentiation is observed. Wadi Gan and Contrebandiers show no
significant differences between each other, but are in turn signifi-
cantly different from both Kharga Oasis and the other northeastern
sites. This latter group of sites themselves cluster together in the
analysis: Kharga Oasis, Sai Island and BT-14 show no significant
differences between them, revealing a persistent differentiation
between northeastern sites and Wadi Gan/Contrebandiers. As Wadi
Gan is almost equidistantly situated between Contrebandiers and
Kharga Oasis, it is suggested that geographical distance alone is
insufficient to explain this consistent and statistically significant
clustering. This pattern is almost consistent throughout the
domains of analysis of this tool class. The theme continues for
tanged tools. Although the tanged element shows a remarkable
homogeneity of manufacture between all Aterian sites, there is
variability in size, levels of resharpening of the active tool edges,
manufacture of tool portion and relationship with other forms of
basal modification. It is significant that all this variability consis-
tently patterns itself in terms of differences between Kharga Oasis
and the Wadi Gan/Contrebandiers group. It is also significant that
the composition and relationships between other forms of basal
modification are not distinguishable between Kharga Oasis and the
Nubian Complex sites. The similarities between Wadi Gan and
Contrebandiers mean that distance alone may not be a valid
enough parameter to explain these consistent patterns of
difference.
These divisions are most consistently evident for point classes.
For every analysis conducted on points, significant differences were
apparent between the Wadi Gan/Contrebandiers group and the
Fig. 16. ANOVA showing geographic distribution driven by the variability of basally thinned and shouldered tools. BT-14 was removed because of the small sample size.
Fig. 15. ANOVA showing the distribution of variability of active edges on tanged tools.
E.M.L. Scerri / Quaternary International xxx (2012) 1e2014
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Kharga Oasis/BT-14/Sai Island group. Although the foliate point
samples are too small to be considered statistically valid, it is
interesting that they pattern themselves along identical lines.
It is clear that these data display specific discontinuous patterns
that require interpretation. Consequently, they were compared
with the model expectations elaborated in Fig. 6 and Table 4. The
relationship between Haua Fteah and the other samples approxi-
mates expectation d. which suggests that differences are attribut-
able to geographic isolation or impermeable social boundaries. On
the other hand the relationship between the Contrebandiers/Wadi
Gan group with the northeastern Aterian and Nubian Complex
group is also similar: they suggest that these two groups were
either geographically isolated or that they had impermeable social
boundaries between them. As a group, Contrebandiers and Wadi
Gan, conform to expectation b. which, despite the distances,
suggests the presence of permeable boundaries between groups.
These results could also indicate the extent of the foraging rangesof
the same groups. Kharga Oasis, BT-14 and Sai Island also conform
to this expectation. The model therefore suggests that some degree
of population structure is present in different parts of the North
African MSA. It is suggested that behavioural adaptation/
geographic isolation is not enough to account for the patterning
and structure of similarity/difference.
Results must however be considered in terms of the ecological
patterning North Africa in MIS 5. Palaeoenvironmental information
for MIS 5 suggests a series of linked lakes, rivers and inland deltas
(Drake et al., 2011) which may have structured mobility and
consequently group interaction (see Fig. 19).
Fig. 18. Boxplot of the first principal component axis, reflecting retouch intensity and blade edge lengths.
Fig. 17. Score Plot of Principal Components Analysis (PCA) for Points showing northeastern sites clustering together and Wadi Gan and Contrebandiers clustering together. Haua
Fteah is notably different.
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 15
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
The Atlantic coast of Morocco is typically regarded as a refuge
area. Wadi Gan falls on the edges of the minimum geographic area
known to have experienced humid conditions in MIS 5(Fig. 19),
but is largely separated from Haua Fteah via the Gulf of Sirte,
a known biogeographic barrier (cf. Reynolds, in this volume). To
the east is the western desert containing lakes and springs such as
at Kharga Oasis and Bir Tarfawi, as well as the Nile. At present,
Fig. 19 indicates that this river network increases in intricacy and
is better connected to the rest of North Africa’s MIS 5 riverine and
lacustrine corridors to the far south of Kharga Oasis and Bir
Tarfawi.
A number of tentative inferences may be drawn at this point.
The results have shown that Kharga Oasis is more similar to Nubian
Complex sites in the same region than it is to other Aterian sites. Of
particular interest, given the fact that Wadi Gan is equidistant
between Kharga Oasis and Contrebandiers is its position inside the
catchment area shown in Fig. 19. It seems reasonable that there is
a degree of ecological separation between the Nubian and Aterian
sites in the northeast and the rest of North Africa. In effect, Wadi
Gan and Contrebandiers may be an example of more coherent units
within the general technocomplex structure, an expectation of
Clarke’s(1968)model. The results of the analysis of the north-
eastern assemblages are more difficult to understand. Adaptation
to a similar ecological milieu may explain the similarities but
a number of provocative suggestions also emerge: is Kharga Oasis
an Aterian site or an example of technological convergence? Does
the clustering of the sites in the analysis suggest population
structure? This study has demonstrated that tanged tools in
themselves say little about the similarity/difference patterns
between sites: had this research been based solely on such types,
the subtle similarity/difference patterns observed here would
almost certainly not be evident.
7. Conclusions
This study has been structured in terms of two primary
concerns. These concerns are now revisited in the light of the above
results.
i. The Aterian is not a discrete chronostratigraphic unit and
tanged tools cannot serve as a main criterion for the defini-
tion of an ‘Aterian’technocomplex.
This research strongly suggests that tangs as hafting elements
cannot serve as the only criterion for the definition of an
‘Aterian’technocomplex, however it would be premature to
reject chronostratigraphic significance, particularly in western
and central North Africa. As a flexible technology, Aterian
designation should instead be based on quantified comparisons
of features such as common reduction strategies. In taking this
approach, this study suggests that tanged tool assemblages show
significant differences across North Africa and may reflect pop-
ulations with different histories and origins. Tanged tools are,
after all, found in many different contexts (e.g. Font Robert
points in western Europe, see also Scerri, 2012)anditcould
hardly be assumed that they represent a single related tradition.
The scale of North Africa and the small population sizes
(Atkinson et al., 2009) suggest that technological convergence is
the most parsimonious explanation for the widespread presence
of tanged artefacts in the North African MSA. The significant
increase in hafting observed in the North African MSA (cf. Rots
et al., 2011) may have led to similar inventions in different
parts of the North African MSA thanks to similar environments,
resources and selective pressures. Consistent similarities within
geographical clusters, however, suggest that it may be possible
Fig. 19. Map of MIS 5 palaeohydrology of North Africa (modified from Drake et al., 2011) with sampled environmental sites marked (grey squares). The red squares demonstrate the
different archaeological site groupings, as extrapolated from this study, in their ecological context. Comparison with Fig. 1 shows that these squares may be expanded to include
many more Aterian sites with further analysis. The red sites are Aterian, the green sites are Nubian Complex. Haua Fteah is represented by the blue circle. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
E.M.L. Scerri / Quaternary International xxx (2012) 1e2016
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age, Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
to identify technological groupings (e.g. an ‘Aterian’in the
northwest and a ‘Nubian Complex’in the northeast) on the basis
of other features (Table 6). More data is required to test this
hypothesis.
ii. The distribution of shared and non-shared attributes reflect
an ecologically structured isolation by distance.
This research suggests a strong element of ecologically struc-
tured isolation-by-distance. The Haua Fteah samples reflect
a different population. Aterian and Nubian Complex sites in the
northeast may similarly reflect different populations to those
present in areas well connected by lacustrine and fluvial networks
(Fig. 19), perhaps represented by Wadi Gan and Contrebandiers.
The distances between the latter two sites, however, and the lack
of significant difference between them at all levels of analysis
suggests that geographical distance alone cannot explain all
the similarity/difference patterns observed in the study. Instead
site clusters may reflect reflecting seasonal aggregation and
fragmentation of different populations. It is suggested that
these groups (with the exception of Haua Fteah) are loosely
united by common sub-Saharan origins. On these terms, it is likely
that the formation of social boundaries drove the behavioural
changes associated with the appearance of pigment use, personal
ornamentation and bone industries witnessed in the North
African MSA.
On a final note, this pilot study strongly indicates that
a comprehensive investigation of the North African MSA has the
potential to reveal significant information about the structure of
populations in the North African MSA. The wider implications of
this study concern the evolutionary trajectories of populations
which are crucial for understanding dispersal within and out of
Africa. Current debates frame such dispersals in terms of pioneers
or of groups expanding their foraging ranges. This study suggests
a third perspective in terms of the possibility of structured groups
both to expand and ‘push’less structured groups into new terri-
tories. In both cases such ‘pushes’may have extended out of
Africa.
Acknowledgements
The research presented here was supported by The University of
Southampton Faculty of Humanities. I am indebted to many people
who supported and assisted me in developing this research: in
particular, Clive Gamble, John McNabb, Gilbert Tostevin, Harold
Dibble, Mohammed el Hajraoui, Graeme Barker, Philip Van Peer,
Marc Naura and Kathleen Nicoll. I am grateful to Rob Foley, Nick
Barton, James Blinkhorn and Huw Groucutt for helpful comments
on drafts of this paper. I would also like to thank Huw Groucutt and
Jimbob Blinkhorn for inviting me to participate in the Middle
Palaeolithic in the Desert Conference and to contribute to these
proceedings.
Appendix A. Supplementary material
Supplementary data related to this article can be found online at
http://dx.doi.org/10.1016/j.quaint.2012.09.008.
References
Alimen, H., Chavaillon, N., 1956. La technique levalloisienne au Sahara nord-occi-
dental. Bulletin de la Société préhistorique de France 53, 344e351.
Antoine, M., 1932. Notes de préhistoire marocaine V: station moustérienne
a quartzite du plateau de la carrière Martin a El Hank. Bulletin Société Pré-
histoire du Maroc 6, 23e46.
Antoine, M., 1934. Notes de préhistoire marocaine, VIII. eUn gisement atérien en
place dans les alluvions de l’Oued Goréa près de Casablanca. Bulletin Société
Préhistoire du Maroc 8, 7e34.
Table 6
Major central tendencies defining the differences between Wadi Gan/Contrebandiers and Kharga Oasis/Sai Island/BT-14.
Wadi Gan & Contrebandiers Kharga Oasis, Sai Island, BT-14
Cores Small cores of different materials
Varied irregular core types, possibly reflecting
more opportunistic strategies
Flakes removed on broad-face
Higher numbers of scars, high flaking intensity
Cores managed circumferentially
Medium to large sized cores of different materials
Variation in Levallois core types, use of Nubian methods
Flakes removed on longitudinal face
Cores less intensively flaked
Greater use of débordant removals to maintain convexity.
Blanks Production of small flakes
Many triangular and ovoid forms
Many plain platforms, some cortical
Poor raw material used for blanks compared to
prepared flakes and blanks chosen for retouch
Unidirectional and bidirectional flaking
Production of medium sized flakes
Laminar flakes
Many finely faceted platforms
No difference in raw material between blank and retouched
tool production
Subcentripetal and centripetal flaking
Tools Few heavy tools
All higher percentage of tools have faceted
platforms compared to blanks
Many denticulates and notches
Centripetal flaking
Heavy tool classes present
Heavy tools have plain platforms
Medium and lightweight tools have faceted platforms
Many denticulates and notches
Unidirectional or bidirectional flaking for heavier tools,
unidirectional convergent and bidirectional on lighter classes
of retouched tools
Hafted tools Tanging as primary hafting modification
Tanged tools retouched
Tanged tools heavily resharpened/‘scraper’forms
Ovoid bifacial foliates
Basal thinning as primary hafting modification
Tanged tools unretouched
Tanged tools on Levallois flakes and laminar forms
Basal thinning and shouldering correlated to low levels
of retouch
Long thin bifacial foliates
Points Small points and few
Classic and convergent Levallois
Medium sized sometimes elongated points
Nubian Point methods
Greater control of platform dimensions to length ratios
Thinned tip points
E.M.L. Scerri / Quaternary International xxx (2012) 1e20 17
Please cite this article in press as: Scerri, E.M.L., The Aterian and its place in the North African Middle Stone Age,Quaternary International (2012),
http://dx.doi.org/10.1016/j.quaint.2012.09.008
Antoine, M., 1939. Notes de prehistoire marocaine. XIV eUn cône de résurgence du
Paléolithique moyen a Tit-Mellil, près Casablanca. Bulletin Société Préhistoire
du Maroc 12, 1e95.
Aouadi-Abdeljaouad, N., Belhouchet, L., 2008. Recent prehistoric field research in
central Tunisia: prehistoric occupations in the Meknassy basin. African
Archaeological Review 25, 75e85.
Atkinson, Q., Gray, R.D, Drummond, A.J., 2009. Bayesian coalescent inference of
major human mitochondrial DNA haplogroup expansions in Africa. Proceedings
of the Royal Society B: Biological Sciences 22, 276 (1655), 367e373. http://dx.
doi.org/10.1098/rspb.2008.0785.
Aumassip, G., 1986. Le Bas-Sahara dans la Prehistoire. Editions du CNRS, Paris.
Aumassip, G., 2004. Prehistoire du Sahara et de ses abords. Maisonneuve & Larose,
Paris.
Balout, 1955. Préhistoire de l’Afrique du Nord: Essai de Chronologie. Arts et Metiers
Graphique, Paris.
Balter, M., 2011. Was North Africa the launch pad for modern human migrations?
Science 331, 20e23.
Bar-Yosef, O., Van Peer, P., 2009. The Chaine Operatoire approach in Middle Palae-
olithic archaeology. Current Anthropology 50, 103e131.
Barham, L., 2001. Central Africa and the emergence of regional identity in the
Middle Pleistocene. In: Barham, L., Robson-Brown, K. (Eds.), Human Roots:
Africa and Asia in the Middle Pleistocene. Western Academic and Specialist
Press, Bristol, pp. 65e80.
Barton, R.N., Bouzouggar, A., Collcutt, S.N., Schwenninger, J.L., Clark-Balzan, L., 2009.
OSL dating of the Aterian levels at Dar es-Soltan 1 (Rabat, Morocco) and
implications for the dispersal of modern Homo sapiens. Quaternary Science
Reviews, 1e18.
Betrouni, 1997. Le paléokarst de Sidi Saïd (Tipasa, algérie) et la question du Paléo-
lithique supérieur maghrébin. In: Fulolla, J.M., Soler, N. (Eds.), El Mon Medi-
terrani Despres el Pleniglacial (18.000e12.000 BP): Serie Monografica, vol. 17.
Centre d’Invesigacions Arqueologiques, Girone, pp. 57e68.
Betrouni, 2001. Le paleokarst de Sidi Said. Aspects chrono-culturels. In: C. N. R. PH.
(Ed.), L’Homme Maghrebin et son environment depuis 100,000 ans. Centre
national des recherches prehistoriques, anthropologiques et historiques, Alger,
pp. 101e11 2.
Binford, L.R., 1977. General Introduction. In: Binford, L. (Ed.), For Theory Building in
Archaeology: Essays on Faunal Remains, Aquatic Resources, Spatial Analysis and
Systematic Modeling. Academic Press, New York, pp. 1e10.
Bisson, M., 2000. Nineteenth century tools for twenty-first century archaeology?
Why the Middle Palaeolithic typology of Francois Bordes must be replaced.
Journal of Archaeological Method and Theory 7, 1e48.
Blome, M.W., Cohen, A.S., Tryon, C.A., Brooks, A.S., Russell, J., 2012. The environ-
mental context for the origins of modern human diversity: a synthesis of
regional variability in African climate 150,000e30,000 years ago. Journal of
Human Evolution 62, 563e592.
Bouzouggar, A., 1997. Economie des matières premiers et du debitage dans la
séquence atérienne de la grotte d’el Mnastra I (Ancienne grotte des Con-
trebandiers eMaroc). Préhistoire Anthropologie Méditerranéennes 6, 35e51.
Bouzouggar, A., Barton, R.N.E., 2012. The identity and timing of the Aterian in
Morocco. In: Hublin, J.-J., McPherron, S. (Eds.), Modern Origins: a North African
Perspective. Springer, pp. 93e105.
Bouzouggar, A., Kozlowski, J.K., Otte, M., 2002. Study of the Aterian lithic assem-
blages from El Aliya Cave in Tangier (Morocco). L’Anthropologie 106, 207e248.
Bouzouggar, A., Barton, R.N.E., De Araujo Igreja, 2007. A brief overview of recent
research into the Aterian and Upper Palaeolithic of Northern and Eastern
Morocco. In: Barich, B. (Ed.), Tra il Sahara e il Mediterraneo: il Jebel Gharbi
(Libia) e l’Archeologia del Maghreb. Edizioni Quasar. Scienze dell’Antichità,
Storia Archeologia Antopologia, vol. 12, pp. 473e488.
Breuil, H., 1912. Le Subdivisions du Paléolithique supérieur et leur signification. In:
CongrèsInternationald’Anthropologie et d’Archéologie Préhistorique, pp.165e238.
Breuil, H., 1931. L’Afrique préhistorique. In: Frobenius, L., Breuil, H. (Eds.), Afrique.
Cahiers d’Art, Paris, pp. 449e482.
Brooks, N., Chiapello, I., Di Lernia, S., Drake, N., Legrand, M., Moulin, C., Prespero, J.,
2005. The climate-environment society nexus in the Sahara from prehistoric
times to the present day. The Journal of North African Studies 10, 253e292.
Camara, A., Duboscq, B., 1997. Bassin sénégalo-mauritanien. In: Tillet, Th. (Ed.),
Sahara: Paléomilieux Et Peuplement Préhistorique Au Pléistocène Supérieur.
L’Harmattan, pp. 67e99.
Cancellieri, M., di Lernia, S. Middle Stone Age human occupation and dispersals:
new data from the Messak plateau (SW Libya, central Sahara). Quaternary
International, in this volume.
Carriere, G., 1886. Quelques Stations Préhistoriques de la Province d’Oran. Bulletin
de la Société de Géographie et d’Archéologie de la Province d’Oran IV, 136e154.
Caton-Thompson, G., 1946. The Aterian industry: its place and significance in the
Palaeolithic World. The Journal of the Royal Anthropological Institute of Great
Britain and Ireland 76, 87e130.
Caton-Thompson, G., 1952. The Kharga Oasis in Prehistory. Athlone Press, London.
Clark, J.D., 1966. The Distribution of Prehistoric Cultures in Angola. Museo do Dundo
Publicacoes Culturais, Lisbon, p. 73.
Clark, J.D., 1980. Human populations and cultural adaptations in the Sahara and Nile
during prehistoric times. In: Williams, M.A.J., Faure, H. (Eds.), The Nile and the
Sahara. Balkema, Rotterdam, pp. 527e582.
Clark, J.D., 1982. The cultures of the Middle Palaeolithic/Middle Stone Age. In:
Clark, J.D. (Ed.), The Cambridge History of Africa. From Earliest Times to c. 500
BC, vol. I. Cambridge University Press, pp. 248e341.
Clark, J.D., 1989. The origin and spread of modern humans: a broad perspective on
the African evidence. In: Mellars, P.A., Stringer, C. (Eds.), The Human Revolution:
Behavioural and Biological Perspectives on the Origins of Modern Humans.
Edinburgh University Press, Edinburgh, pp. 565e588.
Clark, J.D., 1993. The Aterian of the Central Sahara. In: Krzyzaniak, L.,
Kobusiewicz, M., Alexander, J. (Eds.), Environmental and Cultural Change in
North Africa through the 2nd Millennium BC. Poznan Archaeological Museum,
Poznan, pp. 129e143.
Clark, J.D., Schultz, D.U., Kroll, E.M., Freedman, E.E., Galloway, A., Batkin, J.,
Kurashina, H., 2008. The