You are on page 1of 20

Quaternary International 247 (2012) 162e181

Contents lists available at ScienceDirect

Quaternary International
journal homepage: www.elsevier.com/locate/quaint

Time and space in the formation of lithic assemblages: The example of Abric
Roman Level J
Manuel Vaquero a, *, Mara Gema Chacn a, b, Mara Dolores Garca-Antn a, Bruno Gmez de Soler a,
Kenneth Martnez a, Felipe Cuartero c
a
b
c

Institut Catal de Paleoecologia Humana i Evoluci Social (IPHES), Universitat Rovira i Virgili (URV), Plaa Imperial Tarraco 1, 43005 Tarragona, Spain
UMR7194 e Dpartement de Prhistoire, Musum national dHistoire naturelle, 1, rue Ren Panhard, 75013 Paris, France
Departamento de Prehistoria y Arqueologa, Universidad Autnoma de Madrid, Ciudad Universitaria de Cantoblanco, 28049 Madrid, Spain

a r t i c l e i n f o

a b s t r a c t

Article history:
Available online 23 December 2010

Behavioral strategies are a primary focus in the study of Middle Paleolithic assemblages. Since the
emergence of the processual paradigma, this research has been partly based on the use of interpretive
frameworks derived from ethnoarcheological sources. However, this approach is awed by the lack of
correspondence between the time scale of the ethnographic information and the time scale of the
archeological record. This paper presents the lithic assemblage from level J (ca. 50 ka BP), one of the
Middle Paleolithic layers excavated in the Abric Roman (Capellades, Spain). The study of this assemblage
has been carried out from a spatio-temporal perspective, trying to discern two different time scales
involved in the formation of the archeological record: the geological time scale of the assemblage-as-awhole and the ethnographic time scale of the individual events. The results suggest that several domains
of lithic variability, like raw material provisioning, artifact transport and spatial patterns, are timedependent and should be approached taking into account the temporal depth of the archeological
assemblages.
2010 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction: time perspectivism and Neanderthal


behavior
Behavioral patterns are one of the main concerns in the study of
Middle Paleolithic, as the behavioral capacities of Neanderthals are
a primary issue in clarifying the scope of the differences between
them and modern humans. In addition, accessing behavior is the
only surere way of approaching the variability of Middle Paleolithic
lithic assemblages, which seems closely tied to economic strategies
and daily activities. Nevertheless, any approach to Neanderthal
behavior must take into account some methodological questions
associated with the interpretation of the archeological record. The
behavioral perspective in Paleolithic archeology has been linked to
a generalization of ethnoarcheological models, particularly in studies
devoted to the most systemic levels of behavior, like settlement
strategies or intra-site spatial patterns. However, the use of lithic

* Corresponding author. Fax: 34 977 55 95 97.


E-mail addresses: manuel.vaquero@urv.cat (M. Vaquero), gchacon@prehistoria.
urv.cat, gema.chacon@mnhn.fr (M.G. Chacn), lgarcia@prehistoria.urv.cat (M.D.
Garca-Antn), bruno@prehistoria.urv.cat (B. Gmez de Soler), kenneth@
prehistoria.urv.cat (K. Martnez), felipecuartero@yahoo.es (F. Cuartero).
1040-6182/$ e see front matter 2010 Elsevier Ltd and INQUA. All rights reserved.
doi:10.1016/j.quaint.2010.12.015

assemblages to reconstruct behavioral patterns confronts an initial


problem: the lack of correspondence between the geological time
used to dene the lithic assemblages and the ethnographic time of
the events that produced the artifacts. This was the central argument
in the Pompeii premise debate (Binford, 1981, 1986; Schiffer, 1985).
In general, assemblages are dened according to a geological
time scale. All the remains found in the same stratigraphical unit
are included in the same assemblage. The slow sedimentation rates
dominant in most deposits, together with the reduction of the
sedimentary volumes caused by some post-depositional processes
(Brochier, 1999), make unlikely the recovery of occupation oors,
especially in caves and rockshelters. Practically all archeological
assemblages are palimpsests, the formation of which can span
hundreds or even thousands of years and to which many natural
and cultural processes may have contributed (Bailey, 2007). The
temporal depth of these palimpsests depends on the deposition
rate and in some deposits that formed rapidly the geological time
may be close to the ethnographic time. However, the succession of
different events has even been documented in archeological
assemblages characterized by their high temporal resolution (Julien
et al., 1992; Ketterer et al., 2004). One cannot help but wonder to
what point the use of ethnographic models, dened by very
different time scales, are suitable for explaining these assemblages

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

and what misconceptions might be introduced in interpretation by


differences concerning formation time. The disjunction between
ethnographic models and the low temporal resolution of many
archeological assemblages can make ethnographically derived
interpretations problematic (Smith, 1992; Stern, 1993; Lake, 1996;
Murray, 2002; Holdaway and Wandsnider, 2008).
The archeological record is the outcome of processes operating
at different time scales (Bailey, 1981, 1983). There are reasons to
believe that the shortest time scale e the event e is the best suited
to make behavioral inferences (Brooks, 1982). The deposition and
characteristics of material remains depend on decisions made by
individuals at specic times and places with the aim of solving
specic needs. Stratigraphically dened assemblages are simply the
sum of an unknown number of such decisions. An ideal explanation
of an archeological assemblage would be one that accounts for each
of the activity events that contributed to its formation.
Formation length may be an important factor in assemblage
variability. If an assemblage was formed over a long period, it would
be more likely that different activities would be carried out, including
some uncommon ones. Assemblage variability would therefore
increase as the formation period of that assemblage increased (Shott,
2008). Many times the whole assemblage is explained as the product
of the same behavior, as it is assumed that the same constraints
conditioned all the events represented in the assemblage. This is
an unwarranted assumption as there may have been signicant
differences concerning the contexts, circumstances, needs, and
constraints affecting those events. This problem becomes evident
when analyzing how assemblage variability can be interpreted. The
behavioral variability attested to by an archeological assemblage can
be considered as an expression of the different options available for
humans during the period in which the assemblage was formed. In
this sense, the variability of assemblages can be correlated to the
variability of human behavior at a point in time. However, this same
assemblage variability can be alternatively interpreted as the
temporal succession of different behaviors during the assemblage
formation period. In this case, there would be no correspondence
between assemblage variability and behavioral variability, since the
former would be the result of a pooling together of different behavioral moments. Nevertheless, it should be recognized that an
archeological assemblage may be affected by factors of variability
operating at different time scales. The lack of correspondence
between assemblage and behavioral variability may be less acute in
domains of behavioral variability depending on long-term processes.
The study of level J lithic assemblage will try to establish whether
an approach based on the identication of single events can yield
a different light on Neanderthal behavior. Special attention will be
paid to temporal dynamics that formed the lithic assemblage and
their consequences. The Abric Roman is a suitable site to apply this
study, due to its sedimentary context e archeological levels
embedded between travertine layers  and the eldwork methodology based on the excavation of a large surface and the threedimensional record of the archeological remains. In addition to
a conventional assemblage-as-a-whole approach, an attempt is
made to identify specic activity events and to reconstruct their
timing during the formation of level J. This methodology has been
already tested in another level of the Abric Roman (Vaquero et al.,
2004; Vaquero, 2005, 2008).
2. Materials and methods
This study considers two levels of analysis, which correspond to
different time scales. The rst is the assemblage level, in which all
the lithic remains are considered as a whole. This is the highest
temporal resolution that can be obtained through geological
criteria, so assemblages constructed using such criteria should be

163

considered as palimpsests formed by an unknown number of


technical events. This assemblage level is focused on a technological perspective based on the chane opratoire concept and a functional study. The approach to this level will focus on attribute and
use-wear analyses. Methods for the attribute analysis of cores,
akes and retouched artifacts have been largely described in other
works (Vaquero, 1997; Carbonell, 2002). Use-wear analysis has
been carried out using a scanning electron microscope (SEM),
coupled with a morphological analysis of tool edges, following the
procedure outlined in Martnez (2005).
The second level of analysis might be called the event level, and
focuses on identifying the maximum number of technical episodes.
It consists of the highest temporal resolution that can be achieved
and it is the best approach to ethnographic time. Retting and
identication of Raw Material Units (RMUs) are the basic empirical
procedures in this level. An RMU incorporates the artifacts
produced during the reduction of a single nodule (Roebroeks, 1988)
and is dened according to macroscopic characteristics like color,
grain size, texture, inclusions and type of cortex. This procedure,
also known as minimum analytical nodule analysis (Hall, 2004;
Larson, 2004; Odell, 2004), is especially useful in assemblages
with lithics of variable appearance. RMUs have been characterized
taking into account two main features: how they were introduced
into the site and what kind of intentional modication was carried
out on them inside the rockshelter. Retting, nodule analysis, and
spatial distribution, together with archeostratigraphy, form the
basis for interpreting the assemblage in temporal terms.
The spatial distribution of RMUs and retting groups is important in testing hypotheses on cultural and natural formation
processes, post-depositional disturbance, occupation strategies and
temporal patterns. Rets are especially informative regarding the
temporal relationships among different activity areas. The spatial
distribution of RMUs indicates the location of knapping activities.
The scattering of RMUs is also important in studying the temporal
dynamics in the formation of the assemblage. The dispersion of the
artifacts resulting from a knapping episode depends on its temporal
location in the sequence of technical events (Stevenson, 1985,
1991). Earlier episodes tend to be more widely scattered, as they
would have been more affected by intentional and unintentional
dispersion factors. As the knapping events approach the latest
occupation phases, their scatters are less subject to these dispersion
processes and they therefore tend to be more clustered.
3. The Abric Roman and the raw material sources
The Abric Roman is located in the NE of the Iberian Peninsula,
50 km west of Barcelona (Fig. 1). It is a wide rockshelter (Abric) in
a travertine cliff called Cinglera del Capell, located in a karst landscape in the town of Capelladesat the west bank of the Anoia river.
The Abric Roman (41 320 N, 1 41 030"E) has an elevation of 265 m
above sea level. The stratigraphy is made up of 20 m of well-stratied
travertine sediments dated by U-Series between 40 and 70 ka
(Bischoff et al., 1988; Vaquero et al., 2001b). Level J is one of the
richest archeological levels, and has yielded almost 7.000 lithic
artifacts. There are two main archeostratigraphic units, sublevels Ja
and Jb, although they have been distinguished only in the middle of
the site. U-series dating has provided dates around 49 ka BP for the
overlaying tufa (49.3  1.6 and 49.2  2.9 ka BP), and around 50 ka BP
(50.0  1.6 and 50.8  0.8 ka BP)for the underlying tufa (Bischoff
et al., 1988). In addition, a charcoal sample from the archeological
level has been dated at 47.1  2.1 14C ka BP (NZA-2316). Level J shows
a spatial distribution of lithics less clustered than other Roman
levels characterized by well-dened discrete accumulations
(Vaquero and Past, 2001). Nevertheless, as in the rest of the levels,
lithics tend to be associated to hearths. The highest concentrations

164

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 1. Geographical location, main geochronological units and incidents of the Capellades region, situated in the NE of the Iberian Peninsula. Legend (IGME, 1975): 1 Plutonic
intrusions. 2 Paleozoic. 3 Mesozoic. 4 Cenozoic. 5 Quaternary travertines. 6 Quaternary. 7 Anticline. 8 Syncline. 9 Inverse (thrust) fault. 10 Inverse (thrust) fault. 11 Fault. 9 or 10
Normal fault.

correspond to the area where most hearths are located, between the
shelters wall and the outer line of blocks (Fig. 2).
The Capellades region shows a high diversity of lithic raw
materials. The Anoia valley connects three structural units that
have lithic resources: the Ebro basin (1), the Prelittoral Range (2),
and the Prelittoral Depression (3). The Ebro basin is the unit with
more chert-bearing formations and corresponds to a Paleogene
sedimentary basin with Eocene deposits (Sol Sabars, 1958e1964).
The Prelittoral Range provides mainly metamorphic and igneous
rocks and is divided by the Anoia into two different lithological
areas: the Paleozoic materials to the east and the Triassic materials
to the west. The Prelittoral Depression, with chert and limestone in
secondary position, was formed by the sinking of a large block
during the last movements in the Alpine orogeny and is lled by
Triassic materials from the Prelittoral Range, marine Paleogene
materials from the Ebro basin and uvial Quaternary sediments.
The Roman lithic assemblages indicate that two main zones can
be distinguished in lithic provisioning. Quartz and limestone can be
found within a radius of 5 km of the site and they are plentiful in the
surroundings. In contrast, chert nodules are very scarce in this zone.
The quartz sills cross the Paleozoic slate formations surrounding the
site (Garca Rodrigo, 1957). Quartz nodules can be located in primary
and subprimary positions and exhibit angular forms and medium

sizes (10e50 cm). They also appear in secondary position in the river
terraces and other colluvial formations. The limestone and sandstone
used in level J varies greatly. Cortical surfaces indicate that the
limestone blanks found in the site have a secondary origin. Some
limestone cobbles show conchoidal fracture and are suitable for
knapping, while others exhibit a very poor quality. Among the
several types of limestone, only that from the Orp formation have
been identied (Ort, 1990). It presents orange colorless, conchoidal
fracture and micritic texture and content alveolines.
The second zone, ranging between 5 and 25 km from the site,
presents several primary and secondary chert sources (Fig. 3). Chert
provisioning came principally from the Ebro basin formations,
including the Valldeperes formation (20 km from the site), the St.
Mart de Tous formation (15 km), and the Montmaneu formation
(25 km). These types of chert show the following properties:
 Valldeperes chert (VLD). It presents a calcareous-marly cortex
with 1e2 mm of thickness. Its color goes from white to black in
a range of gray-blue and has a translucent appearance. The
texture is soft to saccharoid, with a conchoidal fracture good for
knapping. Quartz and gypsum druses have been found, as well
as vegetable bers and gypsum ghosts. This chert type exhibits
a tendency to white patina.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

165

Fig. 2. Spatial distribution of lithics in sublevel Ja. Drawing by P. Saudo.

 St. Mart de Tous chert (SMT). The cortex is calcareous, 1e3 mm


thick with rough surface. The chert has a translucent appearance and its color varies from light to dark gray in a range of
gray-blue. The texture is soft to rough to the touch, with
a conchoidal fracture, which gives it a variable aptitude for
knapping. This chert had laminated sedimentary structure, and
presents a weak white patina.
 Montmaneu chert (PAN). It appears in lenses of decimetric size.
The cortex is calcareous and less than 1 mm thick. The color is
blackish-green with opaque appearance. When oxidations are
present, its color varies toward light brown. The texture is soft
to the touch, and presents a conchoidal fracture with very good
knapping properties. The sedimentary structure is laminated
and contains several alochems, ooids, pellets and some bioclasts. This chert can present some grayish-brown patina.
The nearest raw materials were not the most exploited. Quartz
and limestone are overwhelmingly dominant in the uvial and
colluvial deposits close to the site. They are clearly the more abundant materials within a 5 km radius, while chert nodules are
extremely rare. Nevertheless, chert was preferentially selected for
knapping sequences. This might explain the economizing behavior
inferred from core reduction sequences. As usual at the Abric
Roman, the most common material in level J is chert (75% of the
artifacts), followed by limestone and quartz (about 10% each one),
and other materials (porphyry, quartzite) with less than 3%. The SMT
and VLD are the most represented cherts. Cortical surfaces indicate
that both primary and secondary chert outcrops were exploited, but
cobbles from alluvial formations were dominant. These cobbles
exhibit a high variability in size and shape, depending of the proximity to the primary outcrops.

4. The assemblage-as-a-whole: technology, typology, and usewear


The technological analysis of level J has been widely presented
in another work (Vaquero et al., in press) and will be only briey
summarized here. Core and ake attribute analysis indicates that
exibility and expediency were basic features of technical behavior.
Core reduction strategies were directed at maximizing the protability of the blanks in the simplest possible way. Knapping strategies were based on two fundamental criteria:
1) The main goal of core reduction sequences was to produce the
highest number of blanks per core, with a little concern on the
form and size of the products. Many reduction sequences were
directed toward the production of small and very small akes.
2) There was a constant adaptation to the shape and size of the
exploited blanks, taking advantage of their natural morphology,
as can be seen for example in the case of core-on-akes, which
are well represented.
This expedient technology produces a high variability in core
shape. Core structure is characterized by the dominance of bifacial
knapping e cores present two opposed aking surfaces separated by
an intersection plane e and the morphology of many cores corresponds more or less to the discoidal method (Fig. 4). However, other
cores exhibit hierarchized structures, similar to those of Levallois
cores. There are also unipolar cores that indicate a volumetric
exploitation, and some cores show the detachment of elongated
blanks. This expedient strategy is used on cores of both int and
limestone, regardless of the phase of the knapping process. This
excludes the possibility of distinguishing between reduction stages

166

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 3. Geological map showing the location of the primary chert-bearing formations.

producing desired end-products (predetermined) and reduction


stages yielding wasted akes (predetermining).
The different core morphologies should not be interpreted as
different reduction methods, but as the consequence of applying
these expedient criteria to a wide range of blank forms. These
expedient contexts tend to produce a higher variability in core
morphology as opposed to contexts characterized by more elaborated strategies (as the Levallois method), which tend to produce
more standardized cores. In this sense, the discoidal and Levallois
cannot be interpreted as equivalent methods. Levallois may be
considered as a true method, since it determines a specic core
structure. Discoidal cores would be the result of applying the recurrence principle in an expedient context. There was not a mental
template guiding the core structure. The core morphology was the
result of adapting the recurrence principle to changing circumstances. Sometimes this led to the appearance of typical discoidal
cores, but not other times.
Knapping processes are characterized by spatial and temporal
fragmentation. This is observed basically in chert cores, as there are
few cases in which the whole exploitation was carried out in the
site. Cores habitually arrived at an advanced stage of reduction and
were nished off in the shelter. Cores are economized, knapped
through short sequences when tools are needed. Flakes show
similar technical features and microwear does not show differences
among them, which excludes the existence of a specialized toolkit.
Neither is there differentiation between akes and tools, because
all akes with an acute edge were potential tools. Cores were
mobile objects transported around as reserve of tools, they always
were knapped following the same criteria, and the produced akes
show the same technical morphologies. This suggests that activities
and tools were the same in the shelter and outside.
Retouched artifacts are scarce (less than 3% of the assemblage)
and, as usual in the Abric Roman, denticulates and notches are

dominant (84.6% of retouched tools) (Fig. 5). Other tool types, and
notably sidescrapers, are practically absent. Large and thick
supports were preferentially selected, and the denticulate edge is
commonly opposed to an abrupt side. Retouched tools were manufactured on ordinary akes, especially blanks from the beginning
of the reduction sequence. The retouch is mainly located in only
one side, the most potentially suitable and longer edge. Retouched
edges do not show evidence of intense resharpening as retouch is
not invasive or stepped. In addition, the microwear analysis shows
a low degree of tool using. Most tools were already retouched when
transported into the shelter. Small akes from cores exploited in
the site were not usually selected for retouching, although retting
shows that some retouched tools were manufactured inside shelter.
Both denticulates produced in and outside the shelter show the
same technical features, which suggests that they were used for the
same activities or that denticulates were suitable for all the range of
activities. Therefore, human groups had a versatile technology
without specialized tools.
Microwear analysis included akes and retouched tools. All the
retouched tools were analyzed and akes were selected from among
the diversity of artifacts common at the site, with preference given to
large akes and excluding retted objects because their microwear
alteration. The microwear analysis showed a reduced percentage of
identication, with a low degree of development of use traces, except
for actions on hide and wood. Cutting actions on animal tissue are
the most common. Retouched tools were used mainly on hard
animal matter, during butchery activities in which bone and other
hard materials of carcasses are rubbed. Flakes were mostly used in
butchery, showing cutting actions on soft animal tissue and, in
a lesser extent, harder animal material (bone and hide). It seems that
retouched tools were used in deeshing and dismembering, while
akes were used in skinning and cutting meat. Only one small
dbordant ake was used in a whittling action on wood. Otherwise,

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 4. Cores from sublevel Ja. Drawing by S. Alonso.

167

168

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 5. Retouched artifacts from level J. Drawing by S. Alonso.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

retouched tools were used in tanning hide in transversal negative


actions. The abrupt angle of retouched edges was used in scraping
actions for cleaning the dermis tissues, which could be related to the
rst phase of tanning process.
Except for one retouched tool that was used for cutting meat and
scraping fresh hide, denticulates were used in only one action with
only one edge, always the retouched one. Therefore, retouching
seems to be related only with tool using. Retouched tools were not
reused and maintained for further activities in or outside shelter, as
we have not identied microwears detached by later series of
retouches. Among the akes with use-wear traces, those presenting
asymmetrical proles are particularly well represented. They are
characterized by an abrupt side opposed to the used edge (dbordant
and naturally backed akes), which allows for comfortable
handling. These data show the versatility of these geometric models
(Beyries and Boda, 1983; Lemorini, 2000). Microwear analysis
suggests a functional duality between two kinds of complementary
tools; acute angles of akes were used in cutting actions on soft
animal tissue, while abrupt retouched edges were used in harder
and longer actions that needed stronger edges.
The organization of lithic production and use follows a clear and
well established technical model, based on the use of expedient
recurrent knapping methods and the manufacture of denticulates
and notches. These technical models are used in any time and
condition. Likewise, tools are used following the same technical
criteria and modes of use. In spite of this, when technical models of
production and use are executed, the specic needs and conditions
of each episode produce a wide range of variability due to the
exibility and versatility of these technical models.
5. The event level: spatial and temporal patterns
According to rets and the macroscopic characteristics of raw
materials, more than 500 Raw Material Units were identied, each
one corresponding to a singular technical event. Moreover, 262
retting groups, totaling 719 artifacts (10.4% of the lithic assemblage),
could be realized. These data form the basis for a temporal approach

169

to the formation of the lithic assemblage. Focus is on two interrelated


issues. First, provisioning strategies are considered, distinguishing the
depositional contexts associated with the different ways of transporting lithic resources into the site. Second, the temporal nature of
the lithic assemblage is highlighted, providing some examples from
retting and spatial data.
Raw materials from local sources are dominant in lithic assemblages, especially during the Lower and Middle Paleolithic (Geneste,
1989; Turq, 1992; Dibble et al., 1995; Fblot-Augustins, 1997),
although there are some Mousterian assemblages in which exogenous materials are the most abundant (Ros, 2005). The percentage
of remains tends to decrease as their origin becomes more distant,
especially when these sources are located more than 10e20 km from
the site. Moreover, the mode in which lithic resources are introduced
into the sites also varies according to distance. Local raw materials
tend to be transported as bulk resources that are processed at the
site. As distance to the lithic sources increases, resources tend to be
transported in more elaborate forms. This pattern may be modied
in contexts characterized by a logistical provisioning, in which bulk
resources are transported into the sites from a long distance (Henry,
1992), but this provisioning strategy does not seem to be common in
the Middle Paleolithic.
The lithic resources of level J were introduced into the site in
different forms: 1) entire or almost entire nodules, 2) angular
fragments, 3) partially reduced cores, 4) single blanks (akes and
retouches artifacts), 5) sets of blanks from the same reduction
sequence. In general, there is a relationship between the introducing ways and the origin of raw materials. Strictly local materials
(limestone and quartz) tend to be introduced as entire nodules.
Exogenous materials from more than 30 km are normally introduced as single blanks. However, many exceptions to this general
rule have been observed, especially as far as the introduction of
local materials is concerned. Transport of local materials as single
artifacts, such as some limestone dbordant akes, has been well
attested. The selective mode of transport is dominant for more
distant materials, but some bulk procurement events have also
been documented. The average behavior ts the predictions of the

Fig. 6. Transported artifacts from level J. Dbordant akes. Photo by G. Campeny.

170

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

minimizing-weight model, but a higher variability can be found at


the event level, which suggests that all the technical episodes
carried out at level J were not affected by the same constraints or
economic considerations. In addition, this variability is particularly
evident in chert provisioning, which shows nearly all the types of
introduction.
Regardless the type and the origin of raw materials, the distribution of lithics by RMU indicates that two classes of artifacts can
be distinguished according to their provisioning strategies. These
classes of artifacts form two different assemblages:
a) The artifacts produced during the knapping sequences carried
out in the rockshelter or derived from the in situ breakage of
nodules. Resources were brought into the site as entire or
almost entire nodules or partially reduced cores.

b) The artifacts produced outside the shelter and introduced into


the site as single items, essentially akes and retouched artifacts.
This second provisioning strategy is the most common. Of all the
introductions identied, 284 (50.2%) corresponded to isolated
akes or retouched artifacts produced outside. The RMUs introduced as entire or almost entire nodules are less common.
However, these uncommon events form the bulk of the assemblage,
which creates a contradiction between the frequency of provisioning strategies and their visibility in terms of the amount of
remains that they represent in the lithic assemblage. The more
common strategy (introducing isolated blanks) produces a relatively small assemblage, while the lesser used strategy (introducing
entire or nearly entire nodules) provides most of the lithics remains
found in the level.

Fig. 7. Retting of reduction sequences on chert. The small dimensions of the exploited blanks allow the production of only some very small removals. Photo by G. Campeny.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

These single blanks make up the transported toolkit and correspond to the strategy of provisioning individuals dened by Kuhn
(1995) or the personal gear described by Binford (1977). This
toolkit was formed basically by akes and retouched artifacts and
most tools correspond to this provisioning strategy. Some of the
cores introduced in a more or less advanced reduction stage would
probably have been included in this transported gear. There are
clear size differences between the transported assemblage and the
lithics from the reduction sequences carried out in the rockshelter.
Among the transported artifacts, medium, large and very large
items are dominant, while in situ production was principally aimed
at producing very small and small akes. For the large and very large
categories, the number of transported blanks exceeds that of lithics
from in situ knapping events. Large artifacts are especially suitable
for transport, since they allow an extended period of use. In addition, it seems that the transported toolkit was selected according to
some technical attributes, like the presence of an abrupt side
opposed to the edge. This feature is characteristic of dbordant
akes, which were preferentially selected for transport, but this
selection also includes any blank showing an abrupt side (naturally
backed akes) (Fig. 6). These criteria, as well as the preferential
selection of large blanks, led to a high percentage of cortical dorsal
surfaces in this transported assemblage. In addition, most of the
retouched artifacts belong to the transported toolkit. Few retouched
tools have been clearly linked to core reduction sequences carried
out in the site.
However, there are no clear-cut technical differences between the
reduction sequences from which the transported artifacts were
produced and those carried out on the spot. Aside from some
quantitative differences derived from the selection criteria, such as
the higher percentage of dbordant akes in the transported

171

assemblage, most of the imported blanks t perfectly well within the


expedient reduction strategies documented in level J. In particular,
blanks showing the typical characteristics of Levallois products are
uncommon. Similarly, no typological differences have been observed
between the transported tools and the few tools produced and
retouched in the shelter. The main difference concerns the size of the
knapping products. In general, the aim of the core reductions performed in the Abric Roman was directed at producing small and very
small akes. Both rets and RMUs provide numerous examples of
this kind of sequence (Fig. 7). This suggests that small akes were
intentional and sought after products. In some cases, this small-ake
production appears at the end of long reduction sequences entirely
carried out in the rockshelter, which previously produced larger
akes. However, this is the less common and it is more usual to nd
sequences whose only purpose was the production of small akes.
This intentional production of small akes has recently been documented in other Middle Paleolithic assemblages (Goren-Inbar, 1988;
Moncel and Neruda, 2000; Moncel, 2003; Dibble and McPherron,
2006). In the Abric Roman, this small-ake-oriented production is
related to the domestic activities carried out around hearths,
whereas large akes tend to be linked to the activities performed
during trips.
These two assemblages exhibit different, and even contradictory,
economical behaviors. Core reduction shows an economizing
behavior. Cores were reduced until exhausted and reduction strategies were aimed at maximizing their protability. On the other
hand, transported artifacts were not intensively used, as indicated by
the absence of strongly reduced artifacts and the low percentage of
tools showing use traces. It seems therefore that economizing
patterns were not a determining factor in the management of these
artifacts. If we take into account that most retouched artifacts

Fig. 8. Retting map of sublevel Ja.

172

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

correspond to the transported assemblage, this suggests that core


reduction and artifact retouch were unrelated phenomena subject to
different constraints.
This has also important consequences for the interpretation of
spatial patterning. One of the main characteristics of the spatial
distribution in level J is size sorting. In sublevel Ja, the interior
hearth-related areas are dened by the dominance of small
remains, while the outer areas around the large blocks show
a higher presence of large remains. Following the drop/toss zone
dichotomy, this was interpreted in previous works (Vaquero, 1999;
Martnez Molina and Rando, 2001) as the opposition between
domestic areas, where the knapping activities were carried out, and
refuse areas, where large artifacts were discarded. However, this
interpretation has not been fully supported by retting, since
connections between the inner and outer areas are scarce. Most
large artifacts located in the exterior areas are unrelated to the
knapping activities carried out at the inner domestic areas.
The differential spatial distribution of the two lithic assemblages
dened according to the provisioning method may provide an
alternative explanation for size sorting. Reduction sequences are
clearly concentrated in the hearth-related domestic areas located in
the interior of the shelter. However, the transported artifacts are
more evenly scattered and are well represented in the outer lowdensity areas. This suggests that knapping activities and the discarding of transported blanks were unrelated events, subjected to
different constraints and perhaps corresponding to different types
of occupation episodes. Domestic areas associated with an effective
settlement in the shelter and showed therefore a careful selection
of their spatial location, searching for the best protected areas. On
the other hand, if the discard of transported blanks was associated
to short visits that did not imply a real dwelling in the shelter, their

spatial locations were less constrained by the natural structure and


were therefore more evenly distributed. Temporal dynamics would
be therefore characterized by the alternation of two different
depositional contexts. Size sorting would be produced by the
differential spatial pattern of these different depositional contexts.
The short-visit context would tend to create a relatively homogeneous scatter of transported artifacts, without clear accumulations.
This scatter would be a sort of intra-site and continuous veil of
stones (Roebroeks et al., 1992), upon which the discrete accumulations formed during residential events would be superimposed.
Resource provisioning provides a rst clue to the temporal
nature of the lithic assemblage. This assemblage is the product of
a sequence of technical events that followed one another
throughout time. This temporal dynamics is also suggested by the
retting spatial pattern. Connection-lines cover all the excavated
surface and, although short rets are dominant (65.8% are shorter
than 2 m), long rets are not uncommon, especially in sublevel Ja,
in which 6.8% of the connections are longer than 8 m. At rst sight,
the movement of artifacts between different areas may suggest that
the activities represented in these areas were contemporaneous
and carried out during the same occupation in the ethnographic
sense of this term (Fig. 8). This hypothesis was proposed in previous
spatial analyses of this assemblage (Vaquero, 1999; Vaquero et al.,
2001a). However, with a closer look at the character of these
rets, the evidence seems less straightforward. The direction of
the intentional movements shows that unidirectional patterns
are prevailing (Fig. 9). Only bidirectional connections can be used
to argue that two activity areas or clusters of remains were contemporaneous. Unidirectional connections cannot be used to
support contemporaneity, especially in technical contexts characterized by expedient reduction sequences in which there is not

Fig. 9. Directionality of the connection-lines corresponding to intentional movements. A dominant unidirectional pattern can be observed.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

a preparing-core stage. On the contrary, they can provide a good


argument in favor of a temporal gap between the formations of
both accumulations. Rets cannot be therefore used as evidence of
the contemporaneous occupation of the different activity areas
identied in level J.
Moreover, the temporal nature of the lithic assemblage is fairly
clear if we consider two interrelated questions: the changes in
provisioning strategies during the formation of level J and recycling. Concerning the rst question, an example from sublevel Ja is
presented. Based on archeostratigraphy and differences in the
scattering of knapping events, three successive formation periods
can be recognized in the middle of the site, each one showing
a different provisioning strategy:
1) The earliest one corresponds to the RMU showing more
dispersed distributions. Products of these reduction sequences
were found scattered over a wide area. These are entire knapping sequences performed on entire or almost entire nodules of
mediocre quality. Most reduction sequences on limestone and
quartz would correspond to this occupation period. Most chert
nodules exploited during this formation period correspond to
the SMT formation (Fig. 10).
2) The second stage is principally associated to nal reduction
sequences and is focused around square P51. These reduction
events exhibit much clustered scatters and most of them
correspond to chert from the VLD formation (Fig. 11).
3) Finally, archaeostratigraphy has identied a third formation
stage represented by a small accumulation found in O49. This
was located at the top of the layer and corresponds to one of the
latest deposition events of level J. These artifacts show a very
selective introduction pattern. They are mainly transported

173

blanks, although a short reduction event producing large akes


was also carried out (Fig. 12).
Recycling is one of the clearest expressions of the temporal
dynamics affecting assemblage formation. Some long-distance rets
indicate intentional movements associated with the recycling of
previously refused artifacts. In some cases, recycling has been
inferred from knapping sequences showing different degrees of
scattering for successive reduction stages, which suggests that these
stages had different taphonomic histories. Two kinds of recycling are
documented: recycling of cores or akes for producing short series of
small akes, and recycling of cobble fragments for their use as
hammer-stones. The scatter of remains corresponding to most
recycling events, together with the location of some of them at the
top of the layer, suggest that this behavior was more common in the
later phases of occupation. Some examples are provided in this
paper, which are fairly illustrative.
The RMU represented in Fig. 13 presents two spatially separated
reduction stages, which are characterized by very different dispersion radii. Artifacts from this nodule were scattered over most of the
excavated surface. This RMU was introduced as a complete or nearly
complete nodule and the rst reduction stages show the widest
scatter. The artifacts from this stage were clustered around O48,
which seems to correspond to the knapping focus, but some
remains were dispersed throughout the central sector of the site.
Nevertheless, the end of the reduction sequence, aimed at
producing very small akes, showed a marked cluster in N59. Five of
the six very small akes detached in this terminal stage were in N59
and the sixth in P59. The core was found in the collection from
Ripolls excavation, which indicates that it was moved again to the
area of Pit 1.

Fig. 10. Spatial distribution of RMU on SMT chert corresponding to the rst formation stage discussed in the text. Drawing by P. Saudo.

174

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 11. Spatial distribution of the RMU on VLD chert from the second formation stage. Drawing by P. Saudo.

Another example from sublevel Jb can be seen in Fig. 14. It shows


the same difference between the scattering of the rst and last
reduction stages. The rst one was widely scattered throughout the
middle of the shelter. The second one formed the bulk of the lithics
found in L49-50. The rst stage, including the decortication of the
nodule, provided a wide array of products, while the nal stage was
exclusively aimed at producing small and very small akes. This is
shown by the rets found in L49-50 and one of them in particular
which is made up of 19 very small and small akes that were
conjoined on the core. The differences between the two stages were
not only related to production goals, but also to the degree of
dispersion. On the one hand, the remains from the rst reduction
event were very scattered. The mean length of the connection lines
found in this zone was 154 cm. On the other hand, the second stage
was clustered in L49-50, showing a principal accumulation of only
50 cm in diameter. Mean length of connection lines was 33 cm. The
assemblage of L49-50corresponds to knapping events carried out
from blanks produced during the rst reduction stage. One of the
cores exploited in L49-50 corresponded to a cortical product
detached during the decortication of the nodule, which was
transported to L49-50 and reduced to obtain very small akes.
Another core found in L49-50 was also transported from the rst
reduction area. The differences in scatter between the two phases
suggest that they were temporally differentiated events and the
accumulation of L49-50 was the result of the recycling of artifacts
discarded in the rst reduction stage .
Recycling of limestone fragments can be seen, for example, in
the retting from sublevel Ja shown in Fig. 15. It is a broken limestone cobble presenting two fragments conjoined by an 1125-cm
connection line between P51 and O40. In this case, a third fragment,
detached at the time of the breakage, was also recovered in P51,

which suggests that this was the breakage locus. The direction of
movement was therefore from P51 to O40, which indicates intentional transport. The two pieces located in P51 were burned. The
fragment in O40 was not burned and showed percussion marks
posterior to the fracture. These patterns show that original cobble
was broken in the middle of the site, after a rst use as hammerstone. One of the fragments was moved toward the area around
O40, where a second use as hammer-stone took place. As another
evidence of the temporal depth of the assemblage, a burning
episode affecting the fragments in P51 happened after the recycling
event.
The last example, from sublevel Ja, shows also the use as
hammer-stone of a limestone fragment from the breakage of a large
cobble after a rst event of use. The artifacts are very scattered, as
can be seen in the ret of Fig. 16, which conjoins artifacts located in
J62, K58, K61, M59 and P52. One fragment presented percussion
marks that extended over the fracture plane, which indicates that
the use as percussor was subsequent to the breakage event. Two
artifacts of this ret were burnt, while the rest of the set did not
show any evidence of re damage. This indicates a temporal
succession of at least four different events: the cobble breakage, the
spatial dispersion of the fragments derived from this breakage, the
use as hammer-stone of one fragment, and the exposure to re of
two other fragments. A new dispersion event can be also proposed,
since the burnt fragments were not located inside a hearth.
6. Discussion
The study of the lithic assemblage of level J has yielded interesting insights on Neanderthal technical behavior. This behavior
combines some structural features characterizing level J as a whole

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

175

Fig. 12. Spatial distribution of the artifacts attributed to the third formation stage.

and other aspects showing strong variability. In the rst place,


exibility is evident in several aspects, from raw material provisioning to artifact use, and is largely the result of the expediency
prevailing in the technical system. This expediency allows for
permanent adaptation to changing circumstances. Variability is the
rule in most behaviors, although certain aspects exhibit a marked
stability, like the manufacture of retouched artifacts, which is
focused on denticulates and notches. Another structural trend is
the preference for chert in knapping, although it was less available
than other materials in the immediate surroundings. However, this
preference was less pronounced during some occupation periods,
which underlines the temporal variability of provisioning strategies. The event-focused approach adopted in this paper has yielded
a broadest picture of technical behavior, highlighting some variability phenomena that would go unnoticed in an assemblage-as-awhole approach. This shows how palimpsests tend to minimize
technical variability that becomes evident at the event level.
Lithic variability is related to the temporal dynamics of assemblage formation. The lithic assemblage dened by stratigraphic
criteria is not a behavioral unit. It is a palimpsest formed by an
unknown number of events, subjected to different constraints, and
showing a wide range of, sometimes contradictory, behaviors. Some
events were associated with knapping, but others to the introduction
and discard of single artifacts. Two different lithic assemblages can
be distinguished, each of them formed in a different depositional
context. These different depositional contexts were largely correlated to different types of occupation events: residential campsites
associated with hearth construction and short non-residential visits.
These occupation types are basic components of Middle Paleolithic
settlement patterns. Residential campsites may be one of the
features dening the Middle Paleolithic as a developmental stage

(Rolland,1999), as indicated by the apparent correlation between the


beginning of the Middle Paleolithic and the rst well-dened
hearth-related assemblages. Although residential occupations were
not necessarily the most common, they produced the bulk of the
lithic remains and exhibit the repeated use of the best-sheltered
areas.
The second depositional context e short non-residential visits e
has been also well documented throughout the Middle Paleolithic,
although it is present since Early Pleistocene times. It has been well
dened at sites in which it was not mixed with a residential
component and the single-artifact-transport pattern was therefore
easier to identify. These assemblages are mainly formed by large
blanks, including high percentages of retouched artifacts. Remains
derived from in situ knapping sequences tend to be scarce (Brugal
and Jaubert, 1991; Deeur and Crgut-Bonnoure, 1995; Costamagno
et al., 2006). This would be a common use of cave and rockshelters,
and this component would be present in practically all lithic
assemblages. The residential use of caves would determine the
differences between sites, since it would be less generalized, both in
diachronic and synchronic terms.
The factors conditioning technical behaviors worked at the event
level and some technical features were event-specic. Certain
behaviors are associated with some events, but not to others. The
assemblage variability is therefore conditioned by the kind and
number of events that contributed to its formation. As the number
of events is a function of formation length, differences in the amount
of time represented in the assemblages are a relevant factor in
explaining inter-assemblage variability. Knapping of limestone and
quartz is not randomly distributed during the formation of the
assemblage, but it is mainly concentrated in a specic area of
sublevel Ja e the central area  suggesting that it occurred during

176

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 13. Retting of the two reduction stages and spatial distribution of Chert-007. Drawing by P. Saudo. Photo by G. Campeny.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 14. Spatial distribution of Chert-01, showing the location of the two reduction
stages. Drawing by P. Saudo. Photo by G. Campeny.

the same occupation event or the same formation phase. These poor
quality but immediately available materials were not equally
exploited throughout the level J formation, but tended to be
restricted to a specic time period. A similar phenomenon was also

177

observed in level I, where the reduction of limestone and quartz


cores was found to be characteristic of some accumulations
(Carbonell, 2002). From this perspective, characterizing a stratigraphic assemblage in behavioral terms makes little sense if the
different components contributing to its formation are not identied. The technical behavior of level J is the average of several
different and may be contradictory actual behaviors. The behavioral
signicance of technical strategies should therefore be sought on an
event time scale.
Far from showing repetitive patterns, behavior exhibits high
variability in the short term. The apparent monotony is the result of
the temporal depth of archeological assemblages. As the resolution
of analysis increases, Neanderthal behavior becomes more variable.
However, this approach depends on the ability to identify single
activity events. In level J, this has been possible in lithic analysis,
and our perspective on this level of variability is biased toward
technical activities. Other domains, such as the study of faunal
remains, are less suitable for this temporal analysis, and they tend
to favor the structural viewpoint of the assemblage-as-a-whole. It
remains to be proven whether this variability at the event level is
also characteristic of other behavioral realms, such as the exploitation of faunal resources, in which only the structural level of
variability can be reached due to the low resolution of the data.
The temporal dimension of variability is also shown by recycling.
Provisioning choices varied throughout the sequence of events that
formed the archaeological assemblage. Archeological sites were
dynamic entities and their appeal for human frequentation may
have changed over time. During the beginning of the formation of
level J, the rockshelter was a place devoid of lithic resources and
provisioning of bulk resources in the form of entire nodules would
have been more likely. As the formation process advanced, the site
was progressively transformed into a raw material source itself and
the need to introduce bulk resources would have been less likely.
The dispersion radiuses of reduction sequences indicate that recycling events tended to take place in the last stage of the formation
sequence. Moreover, the accumulation focused in square P51 is
characterized by a series of reduction sequences on highly reduced
cores introduced as such into the shelter. The scarce dispersion of
these sequences also suggests that they were carried out in the last
occupations of level J. This indicates that raw material constraints
changed throughout the formation of the assemblage and, consequently, lithic provisioning strategy also changed.
Temporal dynamics are relevant to the interpretation of spatial
patterns. Spatial studies should not be exclusively spatial. They
must also include a temporal analysis, since some spatial patterns
can be conditioned by the succession of different depositional
contexts showing different spatial distributions. Size sorting should
be scrutinized from this temporal perspective, especially concerning the identication of secondary refuse areas. Sublevel Ja shows
a spatial pattern characterized by the size sorting of both lithic and
faunal remains. Small remains are dominant in the inner hearthrelated areas, while the frequency of large items tends to increase
toward the outside. This was interpreted as the formation of drop
and toss zones. The inside would correspond to the activity area, in
which there is a preferential accumulation of small remains. Large
remains produced during these activities would be discarded
toward the exterior, forming a low-density belt dened by high
percentages of large items. However, the event-focused approach
indicates that there may be other processes at work in this spatial
distribution. First, rets between the knapping areas and the
purported dumping area are scarce, suggesting that tossing large
artifacts toward the outside was not common. Second, RMU
distribution indicates that most large artifacts were not produced at
the site, but transported as single blanks from outside. Two
different depositional patterns contributed to the formation of the

178

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 15. Retting formed by three fragments of a cobble that shows an intentional transport of one of the fragments. Fragment number 1 was found in O40, fragments 2 and 3 were
found in P51. Photo by G. Campeny.

assemblage. On the one hand, knapping events, which produced


the bulk of the assemblage, are clustered in the hearth-related
areas. On the other hand, artifacts introduced as single blanks tend
to be more evenly distributed, and are well represented in the lowactivity areas. Overlapping of these different depositional contexts
produces a size sorting distribution that resembles that derived
from refuse strategies.
Level J also provides insight into Middle Paleolithic technological variability. Different occupational contexts are identied
throughout the formation of level J, but there is not a clear correlation between these contexts and changes in knapping methods.
Both transported artifacts and reduction sequences carried out on
the spot can be attributed to the same reduction strategies, whose
expedient character is outlined above. Variability in reduction
strategies is more evident at the inter-assemblage rather than the
intra-assemblage level. These changes are clearer when comparing
different levels of the Abric Roman sequence e for instance, levels
E and J e but can hardly be shown when analyzing only one lithic

assemblage. This suggests that technological trends had a temporal


pattern and characterized long time spans. Expedient discoidal
methods were dominant during the formation of level J, regardless
of changes in provisioning strategies or occupation types. This
technological conception was applied in the different activity
contexts performed by human groups in their annual cycles.
Changes in the criteria used in reduction sequences would correspond to long duration dynamics (Bintliff, 1991). These criteria
dened the initial conditions upon which the variability associated
with the adaptation to the specic circumstances took place. Use of
complex or expedient reduction methods would not have depended on occupation length or mobility patterns. The technical
paradigms were in place previous to the variations caused by such
settlement factors.
This temporal dimension shines a new light on Neanderthal
behavior. It allows variability levels associated with two different
time scales to be discerned: the time of the event and the time of the
structure (Sewell, 1996; Harding, 2005). Archeology is a particularly

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

Fig. 16. Retting and spatial distribution of RMU Lim-016 that shows a temporal
succession in at least three different events in various zones of the rockshelter.
Drawing by P. Saudo. Photo by G. Campeny.

suitable domain for this kind of approach, since it allows an almost


immediate access to these temporal levels. Events are specic and
contextual, singular moments embedded in circumstance (Beck Jr.,
et al., 2007), and represent adaptation to the circumstances of
a particular place in a particular instant of time. Events can be also
considered as the actualization of structures. According to Giddens
(1979), structures are the underlying long-term behavioral patterns
and natural conditions in which the short-term events are founded.
The interplay between these temporal scales is essential in understanding social and cultural changes. However, the causal relationships between events and structures are far from being established.
On the one hand, it might be considered that what happens at the
time of the event is dened by the long-term structures. On the other
hand, it could be argued that the short-term individual engagements
determine the larger-scale entities and processes (Harding, 2005). In
any case, establishing such causal relationships is the most important
endeavor at this point. For now, it is enough to identify these two time
scales in the archeological record and to associate them with specic
features of assemblage variability.

7. Conclusions
Technical variability and settlement patterns should not be
approached without taking into consideration the temporal nature
of the archeological assemblages. The archeostratigraphical units
identied in level J do not correspond to occupations in the ethnographic sense, but they are palimpsests produced by a succession of
occupation episodes. This temporal dimension is fundamental to
approach some structural features used to characterize residential

179

occupations: occupation length, special activity areas and disposal


areas. The behavioral interpretation of palimpsests is often biased by
the episodes that produced more vestiges, at the expense of other
behaviors that may be more common but generate few remains. This
is evident by comparing the archeological consequences of the
reduction sequences carried out entirely at the site with the introduction and disposal of single blanks. The latter behavior is more
common and can be considered as more signicant in characterizing
the technical behavior of the Neanderthals that visited the Abric
Roman. However, this behavior tends to be blurred in an assemblage-as-a-whole perspective. These problems can only be confronted by focusing attention on the events and organize the
archeological record according to them. The construction of an
archeological discourse based on events should be considered as
a challenge for future research on the evolution of human behavior.
The temporality of the assemblage limits the ability to characterize the occupations beyond the distinction between residential
and non-residential events. Even in a high-resolution context like
the Abric Roman, it seems illusory to reach a time scale e that
of occupation in the ethnographic sense e in which these kinds of
questions may be answered. This is partly due to the equinality of
processes acting at different time scales but producing the same
archeological outcomes. The archeological consequences of longterm occupations are similar to those produced by the overlapping
of different occupations during long periods. Spatial patterns normally associated with long occupations may also be the result of the
succession of different kinds of depositional events over time.
Some factors used to explain the variability of Middle Paleolithic
lithic assemblages e raw material provisioning, artifact transport,
exploitation intensity e operate at the time of the event. However,
level J also provides some insights into variability factors acting on
a long-term structural time scale. These factors are dened by the
characteristics that remained unchanged during the formation of
the archeological assemblage in spite of the short-term adaptations. For example, the dominance of denticulates and the general
criteria applied in core reduction would be structural trends. Core
morphology exhibits high variability due to the use of expedient
methods aimed at producing as many akes as possible, but characterized by little concern for blank shape and size. This is an
example of the interplay between the different temporal scales of
variability. The specic characteristics of each reduction sequence
depend on the circumstances operating at the event level, such as
nodule shape or transport mode. It is at this level that the variability
of core forms should be explained. However, the expedient nature
of technical behavior is structural and denes the general framework in which this large variability of cores emerges. The lack of
correlation between occupational contexts and changes in knapping methods suggests that the expedient or elaborated nature of
the technical system (e.g. discoidal vs. Levallois) does not depend
on short-term adaptations. The changes in techno-psychological
criteria correspond to long-term process that can be observed in
a geological time scale, but not in an ethnographic time scale. These
technical criteria dened the initial conditions upon which the
adaptation to specic circumstances took place. Use of complex or
expedient reduction methods would not have depended on occupation length or mobility patterns. This pattern differs from that
identied at other sites (Munday, 1979; Henry, 1995) that show
a correlation between reduction strategies and occupational
context. Explaining these differences will be an interesting line of
research.
The limits of the ethnographical explanation are the consequence of the differences between the ethnographic time scale and
the archeological time scale. The basic time scale of ethnoarcheological models, occupational time, is the most difcult to
deduce from the archeological record. There is easy access to the

180

M. Vaquero et al. / Quaternary International 247 (2012) 162e181

time scale of the single event and the geological time scale represented by the stratigraphic unit, but the occupation time scale
remains archeologically invisible. All that is available are events and
relationships between events, but there are serious constraints to
dene these relationships in occupational terms. Through the
spatial association of events, activity areas can be identied, to
establish relationships between activity areas by means of retting.
Nevertheless, it seems unlikely that this chain of relationships will
expand to the scale of an occupation in the ethnographic sense. This
does not mean that all the ethnoarcheological information is
useless for archeological interpretation. It implies that archeologists must be more conscious of the consequences of temporality in
the use of such information. The ethnoarcheological evidence corresponding both to an event time scale and a structural time scale
would be fully adjusted to the kind of data immediately available to
archeologists and would therefore maintain its reliability.
Acknowledgments
Excavations at the Abric Roman are carried out with the
support of the Departament de Cultura de la Generalitat de
Catalunya, Ajuntament de Capellades, Ocina Patrimoni CulturalDiputaci de Barcelona, Tallers Grcs Romany-Valls, BercontrsCentre de Gesti Medioambiental SL, and Constructora de Calaf
SAU. The Generalitat de Catalunya provides nancial support to the
Research Group in Quaternary Human Autoecology (2005SGR00702). We also thank the anonymous reviewers for their helpful
comments. Research of one of the authors (M.G.C.) is supported
by a postdoctoral grant from the Juan de la Cierva Subprogram (JCI2010-07863) of the Spanish Ministry of Science and Innovation.
References
Bailey, G.N., 1981. Concepts, time-scales and explanations in economic prehistory.
In: Sheridan, A., Bailey, G.N. (Eds.), Economic Archaeology. BAR International
Series 96, pp. 97e117. Oxford.
Bailey, G.N., 1983. Concepts of time in Quaternary Prehistory. Annual Review of
Anthropology 12, 165e192.
Bailey, G., 2007. Time perspectives, palimpsests and the archaeology of time. Journal
of Anthropological Archaeology 26, 198e223.
Beck Jr., R.A., Bolender, D.J., Brown, J.A., Earle, T.K., 2007. Eventful archaeology. The
place of space in structural transformation. Current Anthropology 48 (6),
833e860.
Beyries, S., Boda, E., 1983. Etude technologique et traces dutilisation des "eclats
dbordants" de Corbehem (Pas-de-Calais). Bulletin de la Socit Prhistorique
Franaise 80, 275e279.
Binford, L.R., 1977. Forty-seven trips: a case study in the character of archaeological
formation processes. In: Wright, R.V.S. (Ed.), Stone Tools as Cultural Markers.
Australian Institute of Aboriginal Studies, Canberra, pp. 24e36.
Binford, L.R., 1981. Behavioral archaeology and the Pompeii premise. Journal of
Anthropological Research 37, 195e208.
Binford, L.R., 1986. In Pursuit of the Future. In: Meltzer, D.J., Folwer, D.D., Sabloff, J.A.
(Eds.), American Archaeology Past and Future. Smithsonian Institution, Washington, DC, pp. 459e479.
Bintliff, J., 1991. The contribution of an Annaliste/structural history approach to
archaeology. In: Bintliff, J. (Ed.), The Annales School and Archaeology. Leicester
University Press, London and New York, pp. 1e33.
Bischoff, J.L., Juli, R., Mora, R., 1988. Uranium-series dating of the mousterian
occupation at Abric Romani, Spain. Nature 332, 68e70.
Brochier, J.E., 1999. Couche archologique, sol archologique et distributions spatiales: quelques rexions (go)archologiques sur un vieux problme. In:
Rossell., V.M. (Ed.), Geoarqueologia I Quaternari Litoral: Memorial Maria Pilar
Fumanal. Universitat de Valncia, Valncia, pp. 91e95.
Brooks, R.L., 1982. Events in the archaeological context and archaeological explanation. Current Anthropology 23 (1), 67e75.
Brugal, J.-Ph., Jaubert, J., 1991. Les gisements palontologiques plistocnes indices
de frequentation humaine: un nouveau type de comportement de prdation?
Palo 3, 15e41.
Carbonell, E. (Ed.), 2002. Abric Roman nivell I. Models docupaci de curtadurada
de fa 46.000 anys a la Cinglera del Capell (Capellades, Anoia, Barcelona).
Universitat Rovirai Virgili, Tarragona.
Costamagno, S., Meignen, L., Beauval, C., Vandermeersch, B., Maureille, B., 2006. Les
Pradelles (Marillac-le-Franc, France): a mousterian reindeer hunting camp.
Journal of Anthropological Archaeology 25, 466e484.

Deeur, A., Crgut-Bonnoure, E., 1995. Le gisement palolithique moyen de la


grotte des Cdres (Var). Editions de la Maison des Sciences de lHomme
Paris.
Dibble, H.L., McPherron, S.P., 2006. The missing mousterian. Current Anthropology
47 (5), 777e803.
Dibble, H.L., Roth, B., Lenoir, M., 1995. The use of raw materials at Combe-Capelle
Bas. In: Dibble, H.L., Lenoir, M. (Eds.), The Middle Paleolithic Site of CombeCapelle Bas (France). The University Museum, University of Pennsylvania,
Philadelphia, pp. 259e287.
Fblot-Augustins, J., 1997. La circulation des matires premires au Palolithique.
ERAUL, 75. Universit de Lige, Lige.
Garca Rodrigo, B., 1957. El Valle del Anoia. Memorias Y Comunicaciones XVI;
p. 45e80.
Geneste, J.-M., 1989. Economie des resources lithiques dans le mousterien du SudOuest de la France. In: LHomme de Neandertal, 6. La Subsistance. E.R.A.U.L,
Lige, pp. 75e97.
Giddens, A., 1979. Central Problems in Social Theory: Action, Structure, and
Contradiction in Social Analysis. Macmillan, London.
Goren-Inbar, N., 1988. Too small to be true? Reevaluation of cores on akes in
Levantine Mousterian assemblages. Lithic Technology 17 (1), 37e44.
Hall, C.T., 2004. Evaluating prehistoric hunteregatherer mobility, land use, and
technological organization strategies using minimum analytical nodule analysis. In: Hall, C.T., Larson, M.L. (Eds.), Aggregate Analysis in Chipped Stone.
University of Utah Press, Salt Lake City, pp. 139e155.
Harding, J., 2005. Rethinking the Great Divide: long-term structural history and the
temporality of event. Norwegian Archaeological Review 38 (2), 88e101.
Henry, D.O., 1992. Transhumance during the late Levantine Mousterian. In:
Dibble, H.L., Mellars, P. (Eds.), The Middle Paleolithic: Adaptation, Behavior, and
Variability. University of Pennsylvania, Philadelphia, pp. 143e162.
Henry, D.O., 1995. Prehistoric Cultural Ecology and Evolution: Insights from
Southern Jordan. Plenum Press, New York.
Holdaway, S., Wandsnider, L. (Eds.), 2008. Time in Archaeology: Time Perspectivism
Revisited. The University of Utah Press, Salt Lake City.
Julien, M., Karlin, C., Valentin, B., 1992. Dchets de silex, dchets de pierres chauffes. De lintret des remontages Pincevent (France). In: Hofman, J.L.,
Enloe, J.G. (Eds.), Piecing Together the Past: Applications of Retting Studies in
Archaeology. BAR International Series 578, pp. 287e295. Oxford.
Ketterer, I., Pigeot, N., Serra, S., 2004. Le temps de loccupation. Une histoire des
activits et des comportements. In: Pigeot, N. (Ed.), Les derniers magdalniens
dtiolles. Perspectives culturelles et palohistoriques. ditions du CNRS, Paris,
pp. 235e254.
Kuhn, S.L., 1995. Mousterian Lithic Technology. An Ecological Perspective. Princeton
University Press, Princeton.
Lake, M., 1996. Archaeological inference and the explanation of hominid evolution.
In: Steele, J., Shennan, S. (Eds.), The Archaeology of Human Ancestry: Power,
Sex, and Tradition. Routledge, London, pp. 184e206.
Larson, M.L., 2004. Chipped stone aggregate analysis in archaeology. In: Hall, C.T.,
Larson, M.L. (Eds.), Aggregate Analysis in Chipped Stone. University of Utah
Press, Salt Lake City, pp. 3e17.
Lemorini, C., 2000. Reconnatre des tactiques dexploitation du milieu au Palolithique Moyen. La contribution de lanalyse fonctionnelle. In: Etude fonctionnelle des industries lithiques de Grotta Breuil (Latium, Italie) et de La Combette
(Bonnieux, Vaucluse, France). British Archaeological Report International Series
858, Oxford.
Martnez, K., 2005. Anlisis funcional de industrias lticas del Pleistoceno Superior.
El Paleoltico Medio del Abric Roman (Capellades, Barcelona) y el Paleoltico
izli (Hatay, Turqua) y del Mol del Salt (Vimbod, Tarragona).
Superior de ag
Cambios en los patrones funcionales entre el Paleoltico Medio y el Superior. Ph.
D. thesis, Universitat Rovira i Virgili.
Martnez Molina, K., Rando, J.M., 2001. Organizacin y funcionalidad de la produccin ltica en un nivel del Paleoltico Medio del Abric Roman. Trabajos de
Prehistoria 58 (1), 51e70. Nivel Ja (Capellades, Barcelona).
Moncel, M.-H., 2003. Lexploitation de lespace et la mobilit des groupes humaines
au travers des assemblages lithiques la n du Plistocne moyen et au dbut
du Plistocne suprieur. La moyenne valle du Rhne entre Drme et Ardche.
In: BAR International Series 1184, Oxford.
Moncel, M.-H., Neruda, P., 2000. The Klna level 11: some Observation on the
Debitage rules and Aims. The originality of a middle Palaeolithic microlithic
assemblage (Klna cave, Czech Republic. Anthropologie, Brno 38 (2),
219e247.
Munday, F.C., 1979. Levantine mousterian technological variability: a perspective
from the Negev. Palorient 5, 87e104.
Murray, T., 2002. Evaluating evolutionary archaeology. World Archaeology 34 (1),
47e59.
Odell, G.H., 2004. Lithic Analysis. Kluwer Academic/Plenum Publishers, New York.
Ort, F., 1990. Las formaciones evaporticas del Terciario continental de la zona
decontacto entre la Cuenca del Ebro y los Catalnides. In: Ort, F., Salvany, J.M.
(Eds.), Formaciones evaporticas de la Cuenca del Ebro y cadenas perifricas,
yde la zona de Levante: Nuevas aportaciones y gua de supercie. Enresa-Universidad de Barcelona, Barcelona, pp. 70e75.
Ros, J., 2005. Caractersticas de la produccin ltica al nal del Paleoltico Medio
en el Pas Vasco. El caso del Nivel B de Axlor (Dima, Bizkaia). In: Montes, R.,
Lasheras, J.A. (Eds.), Neandertales cantbricos, estado de la cuestin. Monografas, 20. Museo de Altamira, Museo de Altamira, Santander, pp.
333e348.

M. Vaquero et al. / Quaternary International 247 (2012) 162e181


Roebroeks, W., 1988. From Find Scatters to Early Hominid Behaviour: a Study of
Middle Palaeolithic Riverside Settlements at Maastricht-Belvdre (The
Netherlands). In: Analecta Praehistorica Leidensia, 21. University of Leiden,
Leiden.
Roebroeks, W., de Loecker, D., Hennekens, P., van Ieperen, M., 1992. "A veil of
stones": on the interpretation of an early Middle Palaeolithic low density
scatter at Maastricht-Belvdre (The Netherlands). Analecta Praehistorica Leidensia 25, 1e16.
Rolland, N., 1999. The middle Palaeolithic as development stage: evidence from
technology, subsistence, settlement systems, and hominid socio-ecology. In:
Ullrich, H. (Ed.), Hominid Evolution. Lifestyles and Survival. Archaea, Gelsenkirchen/Schwelm, pp. 315e334.
Schiffer, M.B., 1985. Is there a Pompeii premise in archaeology? Journal of
Anthropological Archaeology 41, 18e41.
Sewell Jr., W.H., 1996. Historical events as transformations of structures: inventing
revolution at the Bastille. Theory and Society 25, 841e881.
Shott, M.J., 2008. Lower paleolithic industries, time, and the meaning of assemblage
variation. In: Holdaway, S., Wandsnider, L. (Eds.), Time in Archaeology:
Time Perspectivism Revisited. The University of Utah Press, Salt Lake City, pp.
46e60.
Smith, M.E., 1992. Braudels temporal rhythms and chronology theory in archaeology. In: Knapp, A.B. (Ed.), Archaeology, Annales, and Ethnohistory. Cambridge
University Press, Cambridge, pp. 23e34.
Sol Sabars, L. (Ed.), 1958e1964, Geograa de Catalunya, vol. 4. Aedos, Barcelona.
Stern, N., 1993. The structure of the Lower Pleistocene archaeological record:
a case study from the Koobi For a formation. Current Anthropology 34 (3),
201e224.
Stevenson, M.G., 1985. The formation of artifact assemblages at workshop/habitation sites: models from Peace Point in northern Alberta. American Antiquity 50
(1), 63e81.
Stevenson, M.G., 1991. Beyond the formation of hearth-associated artifact assemblages. In: Kroll, E.M., Price, T.D. (Eds.), The Interpretation of Archaeological
Spatial Patterning. Plenum Press, New York, pp. 269e299.

181

Turq, A., 1992. Raw material and technological studies of the Quina Mousterian in
Prigord. In: Dibble, H.L., Mellars, P.A. (Eds.), The Middle Paleolithic: Adaptation,
Behavior, and Variability. University of Pennsylvania, Philadelphia, pp. 75e85.
Vaquero, M., 1997. Tecnologa ltica y comportamiento humano: organizacin de las
actividades tcnicas y cambio diacrnico en el Paleoltico Medio del Abric
Roman (Capellades, Barcelona), Ph. D. thesis. Universitat Rovira i Virgili.
Vaquero, M., 1999. Intrasite spatial organization of lithic production in the Middle Palaeolithic: the evidence of the AbricRoman (Capellades, Spain). Antiquity 73, 493e504.
Vaquero, M., 2005. Les stratgies de transport doutils dans un contexte rsidentiel:
un exemple du Palolithique moyen. In: Vialou, D., Renault-Miskovsky, J.,
Patou-Mathis, N. (Eds.), Comportements des hommes du Palolithique moyen
et suprieur en Europe: territoires et milieux. Actes du Colloque du G.D.R. 1945
du CNRS, Paris, 8e10 janvier 2003. E.R.A.U.L, Lige, pp. 121e132.
Vaquero, M., 2008. The history of stones: behavioural inferences and temporal
resolution of an archaeological assemblage from the Middle Palaeolithic. Journal of Archaeological Science 35, 3178e3185.
Vaquero, M., Chacn, G., Cuartero, F., Garca-Antn, M.D., Gmez de Soler, B, Martnez, K., in press. The Lithic Assemblage of Level J. In: Carbonell, E. editors. High
Resolution Archaeology and Neanderthal Behavior: Time and Space in Level J of
Abric Roman (Capellades, Spain). Springer.
Vaquero, M., Chacn, G., Fernndez, C., Martnez, K., Rando, J.M., 2001a. Intrasite
spatial patterning and transport in the Abric Roman middle Paleolithic site
(Capellades, Barcelona, Spain). In: Conard, N.J. (Ed.), Settlement Dynamics of the
Middle Paleolithic and Middle Stone Age. Kerns Verlag, Tbingen, pp. 573e595.
Vaquero, M., Past, I., 2001. The Denition of spatial units in middle Palaeolithic sites: the
hearth-related assemblages. Journal of Archaeological Science 28 (11), 1209e1220.
Vaquero, M., Rando, J.M., Chacn, G., 2004. Neanderthal spatial behavior and social
structure: hearth-related assemblages from the Abric Roman middle Palaeolithic site. In: Conard, N.J. (Ed.), Settlement Dynamics of the Middle Paleolithic
and Middle Stone Age, vol. 2. Kerns Verlag, Tbingen, pp. 367e392.
Vaquero, M., Vallverd, J., Rosell, J., Past, I., Allu, E., 2001b. Neandertal behavior at
the middle Palaeolithic site of Abric Roman, Capellades, Spain. Journal of Field
Archaeology 28 (1e2), 93e114.

You might also like