You are on page 1of 18

Journal of Anthropological Archaeology 23 (2004) 431448

www.elsevier.com/locate/jaa

Upper Paleolithic raw material economies


cagzl cave, Turkey
at U
Steven L. Kuhn*
Department of Anthropology, Bldg. 30, University of Arizona, Tucson, AZ 85721-0030, USA
Received 15 September 2004

Abstract
cagzl cave
This paper addresses variation in lithic raw material economy within the early Upper Paleolithic at U
(south-central Turkey). The stratigraphic sequence documents some 12,000 years of the early Upper Paleolithic, entailing changes in lithic technology, raw material exploitation, and game use. Although the same lithic raw materials were
exploited throughout the sequence to make quite similar ranges of products, there are marked changes in the ways raw
materials from dierent source areas were treated, including patterns of transport and raw material consumption. The
concept of technological provisioning is used to understand changing strategies for procuring and managing supplies of
int from dierent source locations. Shifts in raw material economy are argued to represent responses to changes in
residential mobility and the scale/duration of occupations at the cave itself: data on cultural features and foraging strategies provide independent evidence for these shifts in land use. Results have implications for more nuanced approaches
to investigating of lithic raw material economies and the signicance of raw material transfers.
2004 Elsevier Inc. All rights reserved.
Keywords: Lithic raw material economy; Upper Paleolithic; Turkey; Technological provisioning; Mobility

The study of lithic raw material economies has assumed an increasingly prominent role in lithic studies
and research on prehistoric hunter-gatherers. Evidence
for the procurement, transport, and management of
chipped stone raw materials has served as the basis for
inferences about mobility (Ambrose and Lorenz, 1990;
Amick, 1996; Blades, 1999; Feblot-Augustins, 1993;
Thacker, 1996), territoriality (Delage, 2001; Demars,
1998; Hovers, 1990; Wengler, 1990), exchange (Bourque,
1994; Hughes, 1994; Meltzer, 1989), colonization patterns (Rockman, 2003; Tankersley, 1994), cognitive
capacities (Boesch and Boesch, 1984; Roebroeks et al.,

Fax: +1 520 621 2088.


E-mail address: skuhn@email.arizona.edu (S.L. Kuhn).

1988; Stiles, 1998; Wynn and McGrew, 1989), as well


as social and reproductive networks (Feblot-Augustins,
1993, p. 251; MacDonald, 1999) among prehistoric foragers and human ancestors alike.
Although there is broad agreement about the importance of knowing where raw materials came from and
how far they were moved before becoming part of the
archaeological record, there is little consensus on the signicance of so-called transfer distances. Many archaeologists use the distance from nd spot to source as a
proxy for raw material cost. Transfer distance is treated
as proportional to the time and energy that people were
willing to or constrained to expend in getting access to a
particular type of stone. However, the distances stone
was moved in the past have no intrinsic meaning. Such
facts are simply one frame of reference against which

0278-4165/$ - see front matter 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jaa.2004.09.001

432

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

we can recognize and attempt to understand variation in


human behavior. Artifacts can be transferred from
source to site in many dierent ways, and the relationship between distance and cost varies according to
how the transfer took place. More than just where things
came from, we want to know how members of particular
communities, populations, or species managed to get
lithic raw materials from source to places where tools
were made and used, given the particular constraints
they experienced. The strategies people used to keep
themselves supplied with stone were inuenced by a wide
range of factors, from territoriality and social networks
to practical limits on travel and time allocation. Analytically, the relationships between material transfer distances and characteristics of artifact life histories (or
chanes operatoires) are of special interest, as they can inform us about the stages of transformation and use that
artifacts underwent during the journey from source to
point of discard.
This paper presents a case study in changing raw
material economy among prehistoric foragers using a
single location. The Upper Paleolithic sequence at
cagzl cave, Turkey, documents shifts in lithic technolU
ogy, raw material exploitation, and foraging occurring
over a period of more than 12,000 years. The same lithic
raw materialspossibly the very same outcropswere
exploited throughout the sequence, and these ints were
used to make quite similar kinds of artifact. At the same
time, changes in how various raw materials were treated
provide evidence of shifts in strategies for procuring and
managing supplies of int from dierent source locations. The concept of technological provisioning serves
as a useful framework for evaluating strategies of raw
material use, which are argued to have been responses
to changes in the ways the cave itself was used. The lithic
evidence suggests that there was continuity in provisioning strategies for much of the sequence, but that major
changes occurred in the most recent Upper Paleolithic
layers. Evidence from cultural features and the zooarchaeological record suggests that the shift in provisioning strategies coincided with increasing intensity and
duration of occupations at the site.

Technological provisioning, mobility, and site use


The term provisioning, often invoked in studies of
lithic raw material economies, has more than one meaning. Among Continental European researchers it often
refers to systems for the acquisition of raw materials
suitable for making tools (e.g., Demars, 1982; Geneste,
1988a,b; Perle`s, 1990; Trassierra et al., 2002; Turq,
1996). Such studies focus primarily on locations where
materials were obtained and inferred patterns of transport from quarries to archaeological sites: common
products are star diagrams linking point sources with

a site at the center. A somewhat dierent set of concepts


(and methods) is embodied in the notion of provisioning
strategies (Kuhn, 1992, 1995, 2002). Provisioning strategies are idealized systems for making nished tools and/
or necessary raw materials available when and where
they are needed. The concept of provisioning strategies
encompasses a variety of dierent pathways by which
artifacts may nd their way from quarry to archaeological deposit. These are strategies only in the abstract
sense, repeated patterns of behavior that limit and direct
the ow of artifacts and raw materials through a technological system, and it is not assumed that they existed as
conscious goals in the minds of ancient people.
Three alternative strategies have been identied for
making artifacts or artifact-making potential when
and where needed. Since technological activities are
inevitably associated with specic locations, one option
is to stockpile tools and/or raw materials at places on
the landscape where activities are likely to take place.
Many, if not most activities involving stone tools take
place at residential sites, but manufacture or tool use
may also occur at other kinds of locations, such as specialized resource procurement or processing sites. As
artifacts must be used by people, a second potential
strategy entails keeping individuals supplied with the
artifacts and raw materials they are likely to use. These
rst two strategies involve a measure of planning, and
as such, they represent two alternative forms of curation behavior. Provisioning of individuals ts most
closely with common understanding of the term (Binford, 1979; Shott, 1997), but both entail supplying individuals or places according to anticipated needs. A
third provisioning strategy actually involves little or
no planning. This is termed provisioning activities, producing artifacts on an ad hoc basis as needs for them
arise.
The three provisioning strategies operate under different sets of constraints, and are expected to result in
contrasting artifact life histories, that is generalized trajectories of raw material procurement, tool manufacture, maintenance and discard. In provisioning
individuals, who are by denition mobile, transport cost
is an important constraint. Strategies of provisioning
individuals should be based on maximizing potential
utility of artifacts relative to weight. Just how this is best
accomplished remains a topic of some disagreement
(Goodyear, 1989; Kuhn, 1994; Morrow, 1995; Nelson,
1991; Roth and Dibble, 1998; Shott, 1986), but it is generally agreed that those individuals will most often provision themselves with nished tools rather than raw
materials. In order to extend their utility and minimize
the amount that must be carried, artifacts used to provision individuals should often be resharpened and reshaped as their edges wear out. In some cases,
signicant eort may be devoted to obtaining raw materials or achieving artifact forms (e.g., bifaces or Levallois

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

akes) that maximize utility per unit weight (Goodyear,


1989; Kelly, 1988; Kuhn, 1994).
Provisioning places requires carrying artifacts or raw
materials from the point of procurement to the specic
place(s) being provisioned. Substantial quantities of
goods can be amassed in permanently occupied or frequently visited locations and used as needed, and it is expected that places will more often be provisioned with
raw material in various states of manufacture, from unworked raw materials to blanks and nished tools.
Amassing a ready supply of stone in places where it will
be used also relaxes need to squeeze the last bit of use
out of every artifact. Provisioning of places therefore
should be marked by less investment in artifact manufacture and less extensive reduction and reworking of
tools than the strategy of provisioning individuals.
The provisioning of activities as needs present themselves is subject to a somewhat dierent set of limitations
and advantages. With any sort of planned strategy there
is always the chance of over-anticipating needs, collecting, manufacturing and carrying artifacts that are never
actually used (Surovell, 2003), so making artifacts only
when there are immediate needs for them eliminates
risks of overproduction. Because making tools in reaction to an immediate need places the toolmaker under
more severe time constraints, we expected provisioning
of activities to involve minimal investment in manufacture and to focus on locally plentiful materials. Reliance
on ad hoc manufacture also places people at the mercy
of the geological environment. Where raw materials
are not ubiquitous ad hoc provisioning of activities is
risky or potentially costly, as there is a strong possibility
of being left without the artifacts needed to carry out a
task eectively. Ad hoc production is reliable only where
raw materials are known to be available close to where
tools are used. Alternatively, it is a low-cost option for
someone prepared to do without tools at all.
No human group is expected to rely exclusively on a
single mode of technological provisioning, and no two
groups are expected to show exactly the same mix of
strategies. We can also expect individuals within groups
to have adopted dierent mixes of strategies, although
much less attention has been paid to intra-group variation to date. The relative importance of each strategy depends on the interaction of several variables, including
the distribution and qualities of raw materials, residential mobility, labor scheduling, and the nature of the
activities conducted. Raw material distributions form
an essentially passive backdrop, a xed set of constraints
that can be reconstructed to a large degree based on
modern geology. The human variables are more interesting from an anthropological perspective.
Patterns of labor allocation and the characteristics of
individual activities favor dierent strategies for maintaining a supply of usable artifacts. Artifacts used in
tasks conducted frequently and at unpredictable times

433

and places must be available at all times, and should often form parts of toolkits carried by individuals. In contrast, the requirements of activities associated with
specic times and locations, such as the processing of
seasonally or spatially restricted resources, can eectively be serviced with tools that have been provisioned
to that place. Tasks such as hunting large game are characterized by high levels of time stressnarrow windows of opportunity(Torrence, 1983, 1989), and
must be carried out with pre-made artifacts provisioned
to individuals or places. Tools for tasks with less severe
logistical constraints can be supplied in a variety of
ways; if suitable material is found nearby, artifacts
may even be produced on the spot. For example, among
early Paleoindian groups, successful hunting of large
game depended on spears with stone points that were
transported and maintained over long periods, probably
as part of individuals personal gear. If suitable raw
materials were present, the more relaxed job of butchering could be carried out using quickly produced and
quickly discarded tools made from local stones (e.g., Frison and Stanford, 1982).
Mobility and land use can have pronounced inuence
on the viability of dierent strategies provisioning tools
and raw materials. The relationship between mobility
and technological provisioning is approached in terms
of an informal sort of optimality model, identifying
strategies or strategic mixes that should be most advantageous under dierent conditions. Residential mobility
directly inuences the predictability of the locations
where activities will be conducted, and hence the utility
of alternative provisioning strategies. The more sedentary a group is, the more predictable are the loci of activities, and the greater are potential advantages of
provisioning places. The inverse relationship between
levels of investment and maintenance of stone tools
and degrees of sedentism in the Americas (Parry and
Kelly, 1987) is a general expression of this tendency.
Where the frequency of residential mobility is very high
and occupational events very short, it is much more
practical to provision individuals. At the same time variability is expected within any group. The duration and
predictability of residential locations and levels of individual mobility vary over the course of a year and across
the territory of all but the most sedentary peoples. Consequently, strategies for managing supplies of tools and
raw material may be more dependent on local factors
such as the nature of a particular occupation than on
broader strategies of moving around the landscape.
Both general levels of residential mobility and the
duration of occupations should also aect how distance
to source relates to the actual cost of artifacts and raw
materials, and how both variables correlate with artifact
life histories. In the case of extremely brief occupations
people may have to rely entirely on transported toolkits,
but as the length of a residential stay lengthens it be-

434

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

comes more practical to provision the residential site itself with chippable stone. All other things being equal,
we can expect foraging catchmentsand by extension,
the catchment area for provisioning a place with
stoneto expand as occupations become more prolonged and local resources are used up (Surovell,
2000). Thus, in the case of provisioning places transfer
distances for raw materials should reect duration of
occupation to some degree. Moreover, because raw
materials are moved more-or-less directly from source
to the place of use there should be a direct relationship
between transfer distances and raw material cost, and,
all other things being equal, we can expect economizing
behavior of toolmakers to reect this.
Artifacts used to provision individuals may also be
transported long distances as a function of individual
mobility. Such artifacts are not carried to locations so
much as carried along with people, so the distance to
source may say much more about scales of individual
mobility than about use of the site itself. Moreover, because the tools are carried along as part of transported
toolkits, distance to source and cost are not equivalent.
Generally speaking, we might expect artifacts from more
distant sources to show greater evidence for reduction
and reworking since those artifacts would have been in
use for a long time before being deposited in the site.
However, in the case of provisioning individuals the
relationship between artifact life history and distance
to source depends more on the route of movement than
actual distance (e.g., Hofman, 1991) and less on the actual cost of procurement.

cagzl cave
A case study: U
cagzl (three mouths) cave is located on the
U
Mediterranean coast of the Hatay region of southern
Turkey, around 15 km south of the mouth of the Asi
(Orontes) river, in the extreme northeast corner of the
Mediterranean basin (Fig. 1A). The area around
cagzl cave is characterized by dramatic relief, and
U
topography would not have been radically dierent during the Pleistocene. The coast is very steep in the vicinity
of the cavethe sea-oor reaches a depth of 200 m within 5 km of the present-day shorelineso even during
periods of very low sea level associated with Oxygen Isotope stage 3 the site would always have been within a
few kilometers of the shore.
cagzl cave collapsed at some time in
The vault of U
the past, resulting in the loss of roughly 3 m of archaeological deposits to erosion. Intact archaeological deposits are preserved in two areas within the site (Fig. 1B).
The narrow, tubular chamber at the south end of the site
contains at least a meter of heavily cemented Upper
Paleolithic sediments: this area was excavated by a previous investigator during the late 1980s (Minzoni-Der-

oche, 1992). A deeper stratigraphic sequence is


preserved at the north end of the site along what was
formerly the back wall of the cave, and this is where
the current project has concentrated its eorts (Kuhn
et al., 1999, 2004). Sediments at the extreme north end
of our excavation trench reach a depth of more than
4.5 m. Early Upper Paleolithic cultural materials are
abundant in the uppermost 3 m or so of the sequence,
while artifacts and bone are quite scarce below approximately 350 cm below datum.
The lithology of the stratigraphic sequence at
cagzl cave (Fig. 2) is fairly homogeneous. The main
U
geogenic sedimentary component is reddish clay or silty
clay, the well-known terra rosa sediment typical of karstic terrain in the Mediterranean basin. Individual layers
are dierentiated primarily on the basis of the abundance and character of anthropogenic elements such as
ash and charcoal, artifacts, and bone, and only secondarily based on subtle variations in the clay matrix. Some
layers consist mainly of terra rosa clay with a relatively
minor amounts of ash. Other units are marked by a
much more evident anthropogenic component, ranging
from massive deposits of ash to closely superimposed
ash and charcoal lenses. These distinctions certainly reect a combination of factors. Layers composed mostly
of terra rosa with little ash may represent less intensive
occupations, but they may also derive from periods
when more clay was owing into the cave from outside.
Patterns of human occupation appear to have varied
even among the layers with the strongest anthropogenic
signals. At the bottom of the sequence, in layers H2-3
and the upper part of layer I, for example, ash, charcoal,
artifacts, and bones are concentrated in relatively thin,
discrete lenses, separated by bands or stringers of clay.
This nely divided structure suggests that human occupation consisted of a series of many relatively brief
events. In contrast, layer B1-B4 is dominated by a massive dump of ash, lithics, and bone that suggests a
much more continuous deposition of anthropogenic sediment and perhaps a more prolonged occupation. This
contrast will be important in understanding changing
cagzl cave.
technological provisioning strategies at U
During excavation, many layers were further subdivided into a series of lenses or facies with designations
such as B1, B2, or H3, based on variation in the amount,
character, and distributions of ash and other material.
Because they reect human actions, these sub-units seldom extended over more than a few square meters: some
at least partially overlap in time as well. In the analyses
below some of these ner units have been combined into
units B1-B4, Fa/Fb, and H2/H3 to provide greater
stratigraphic and chronological separation of the analytical units. Results presented also represent a subsample
cagzl
of the total artifact assemblage derived from U
cave. The materials discussed below come from a block
of 8 m2 units at the north end of the excavation trench.

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

435

cagzl cave. Shaded areas show int-bearing limestones closest to the site. Geological data
Fig. 1. Location (A) and layout (B) of U
lcekli Turkiye Jeoloji Haritas, Hatay, and Adana quads. Maden Tektik ve Arama Genel Mudurlugu, Ankara, 2002.
from 1:500,000 O

These squares, which yielded the majority of archaeological materials from the site, sample all or most of the
stratigraphic column, and this horizontally continuous
subsample was considered most appropriate for the purposes of examining change over time.
In terms of common chronostratigraphic units, the
cagzl cave sequence documents a transition between
U
an Initial Upper Paleolithic (IUP) technological system
and the Ahmarian, a more classic early Upper Paleo-

lithic industry. Assemblages from layers B, B1-B4, and


C are typically Ahmarian in character, whereas layers
F, Fa, Fb/Fc, G, H, H2/H3, and I fall within the Initial
Upper Paleolithic (Fig. 2). Due to the small sizes of the
assemblages and the atypical nature of many artifacts,
the assemblages from C/D, D, and E are somewhat
more dicult to characterize, but they are certainly closer to the Ahmarian than to the IUP. The range of
cagzl cave is very similar
assemblages represented at U

436

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

cagzl cave stratigraphy, showing approximate depth ranges of major chronostratigraphic units.
Fig. 2. U

to that found in Ksar Akil shelter, layers XXII-XVI


(Azoury, 1986; Ohnuma and Bergman, 1990). A suite
cagzl indiof AMS radiocarbon determinations from U
cates that the entire suite of Upper Paleolithic assemblages spans a period from roughly 41,000 to 28,000
(uncalibrated) radiocarbon years BP. A small Epipaleolithic sample, stratigraphically unconnected with the rest
of the sequence, is not discussed here.
cagzl cave (layer F
The Initial Upper Paleolithic at U
and below) is characterized by wide, at blades and
points with broad facetted platforms. Few specimens
t a strict denition for Levallois products but the general conguration of both blanks and cores is highly
reminiscent of Levallois method, particularly in layers
G through I. Dorsal scar patterns on blanks reect predominantly unidirectional parallel or convergent orientation of detachments, and most cores have a single
striking platform. The large platforms and pronounced
bulbs of percussion suggest that hard-hammer percussion was the dominant technique. Endscrapers, especially short, thick specimens with faceted butts, are the
most common retouched tool forms. Burins are also
moderately abundant (Fig. 3, 1117). Chanfreins, an
artifact type characteristic of the IUP in Lebanon (Azoury, 1986; Ohnuma and Bergman, 1990), are clearly
present only in layer I.

The Ahmarian assemblages of layers B, B1-B4, and C


at the top of the sequence are characterized by very regular blade blanks produced from bidirectional prismatic
cores. The morphologies of both cores and blades suggest that soft hammer or indirect percussion was used
at least in the nal stages of blank production: platforms
on blades are small, sometimes invisible, and often show
evidence of extensive preparation by grinding. Elongated simple endscrapers on blades are the most common tool forms in the Ahmarian assemblages. Other
relatively common tool forms include retouched and/
or pointed blades (Ksar Akil or el Wad points) (Fig.
3, 110). Burins are very rare, as are typical Aurignacian
forms such as Dufour bladelets and carenated pieces.
These changes in blade production technology are
quite signicant in terms of Levantine Upper Paleolithic
cagzl cave sequence
sequences. Overall, however, the U
is characterized by a notable degree of continuity. With
the sole exception of the earliest layer (I), the majority of
tools are made on blade blanks, and although the technological attributes of these blades changed over time
their basic sizes remained relatively constant. Retouched
tool inventories are also remarkably consistent. Endscrapers are the dominant tool for in every stratum except layer I. Pointed blades increase over time whereas
burins become scarcer, but neither artifact class ac-

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

437

cagzl cave: 110, Ahmarian (layers B-C); 1117, Initial Upper Paleolithic (layers F-I).
Fig. 3. Artifacts from U

counts for more than 20% of retouched tools in any


layer.
cagzl cave
Conditions of organic preservation at U
are very good, and a range of artifacts and materials beyond chipped stone is represented. Simple bone tools,
mainly polished points, needles, and rods, are present
in small numbers in most levels. Shell ornaments are

abundant throughout the Upper Paleolithic sequence,


although the composition and diversity of the ornament
assemblages changed over the time (Kuhn et al., 2001).
Most layers also yielded rich faunal assemblages.
Changes in foraging over the long sequence at
cagzl cave are also subtle but signicant. Despite the
U
sites proximity to the sea, medium to large terrestrial

438

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

ungulates were the predominant prey throughout the


Upper Paleolithic occupation. This emphasis on large
game may be a function of local topography: steepwalled, narrow canyons on either side of the site would
have made excellent natural traps for game. The most
common prey species varied over time, probably as a response to changing terrestrial environments. In layers I
through F, Capra aegagrus, wild goat, was the main large
prey animal, followed in abundance by roe deer (Capreolus capreolus): pig (Sus scrofa) and bear (Ursus arctos)
are also presenting smaller but still appreciable numbers.
In the middle layers (D, E) fallow deer (Dama mesopotamica), roe deer, and red deer (Cervus elaphus) were the
most common large prey. In the uppermost layers (B,
B1-B4, and C) roe deer is the most common prey item,
followed in abundance by fallow deer: larger species such
as red deer, Bos primigenius, and Sus are quite rare.
The use of resources other than large game show
somewhat more striking trends. Below layer D it ap cagzl cave seldom
pears that the inhabitants of U
exploited small game or marine resources, or at least
that they did not return these foods to the cave. Layers
B and B1-B4 in particular contain substantial quantities
of small animal remains. Two varieties of shellsh that
inhabit rocky shores (Patella and Monodonta) are especially common, but bones of rabbits, birds, and large
sh are also abundant. The emphasis on small game
and marine foods reached its peak in the small Epipaleolithic sample. A similar tendency towards increasing diet
breadth and intensied focus on small game over the
course of the Upper Paleolithic characterizes Upper
and Epipaleolithic sequences throughout the Mediterranean basin (Stiner et al., 1999, 2000). We have no direct
cagzl
evidence for the use of non-animal resources at U
cave. However, several pitted anvil stones which could
have been used to crack open nuts were recovered from
layers B, B1-B4, and E. Some kind of technological
application for these artifacts cannot be excluded,
although bipolar technique, which would also produce
pitted anvils, was seldom used.

Lithic raw materials


The great majority of artifacts from the Upper Paleo cagzl cave was made of int and allied
lithic layers at U
forms of crypto-crystalline silicate rock. The cryptocrystalline silicate materials show considerable variety
in color, graininess, and fossil inclusions. We have been
able to identify a number of potential primary and secondary sources for several of the most common int
raw materials within a radius of approximately 30 km
of the site.
cagzl cave are
The primary int sources closest to U
associated with Upper Cretaceous limestone bedrock on
the high plateau north and east of the site. Surface expo-

sures near the town of Yayladag, roughly 15 km straightline distance from the cave, contain spherical and ellipsoid nodules ranging in size up to 40 cm in length. The
most common variety of int is a light gray or brown,
semi-translucent material containing numerous small,
round white fossils. Textures range from extremely ne
to rather coarse grained. Cretaceous ints like those in
the Yayladag area are the most common raw materials
cagzl cave. No usable ints or cherts
in all layers at U
have been identied in the somewhat older Cretaceous
limestone into which the cave itself is eroded.
A second set of bedrock int sources occurs in younger (Eocene, Oligocene or Miocene) limestones around
cagzl cave. Surface expo30 km distance away from U
sures near the village of S
enkoy yield relatively at,
irregularly shaped nodules of high-quality int as large
as 30 cm. The S
enkoy ints vary in color and texture
in dierent exposures, perhaps reecting dierent limestone beds: the geological stratigraphy of the area is
not well worked out. The most abundant material is very
ne grained, opaque, brown to black int. Although it
has better aking properties than the Cretaceous ints
described above, the dark brown S
enkoy int is much
cagzl cave.
less abundant in the Upper Paleolithic at U
Other materials from the S
enkoy area, including a mottled brown, semi-translucent variety of int, are more
cagzl cave.
common in the assemblages from U
Secondary deposits of heavily rolled int pebbles and
cobbles, associated with fossil beaches some distance
above the modern shoreline, occur much closer to
cagzl cave. We have identied two such deposits
U
within a few km of the site as well as several others located at slightly greater distances. Most pebbles found
on the surface today are less than 1012 cm in length,
smaller than nodules from the primary sources, but
archaeological specimens indicate that much larger pebbles could be found in the area during the Pleistocene.
The geological sources represented in these secondary
deposits are diverse. The characteristic translucent, fossiliferous Cretaceous materials are abundant in these
pebble deposits, but other cypto-crystalline silicates,
including radiolarites, are also present. Active cobble
cagzl cave do not contain silicate
beaches around U
rocks, but are composed exclusively of limestone and
dolomite. Still, we cannot rule out the possibility that
deposits containing siliceous materials are located even
closer to the site beneath the current level of the sea.
Because some materials, including the most common
types of Cretaceous int, occur in both primary and secondary contexts within 20 km of the site, chemical or
mineralogical criteria would be of little utility in determining where a given specimen was actually collected.
A simpler but nonetheless eective approach to assessing
the origins of specic artifacts is based on the nature of
cortex preserved on archaeological specimens (see also
White, 1995, 1998). Several dierent types of cortex were

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

recorded during data collection. The term fresh nodule cortex refers to soft white chalky or opaline rind
preserving its original irregular surface. Specimens with
rolled nodule cortex exhibit soft, white outer layer
that has been lightly smoothed or eroded by some
mechanical process, presumably water transport. Pebble cortex is a distinctively abraded, pitted outer surface showing evidence of extensive water transport and
reworking: none of the original chalk or opal cortex is
retained. In our experience in the area, pebbles from
beach deposits exhibit only pebble cortex, whereas both
fresh and rolled nodular cortex can be found only on
specimens collected around the primary source areas.
Dierences in the nature of cortex on artifacts allow
us to infer only minimum transport distances. Generally
speaking, artifacts with nodular or rolled nodule cortex
cannot have been collected closer to the site than the
closest primary deposits. Of course, int-bearing limestone outcrops over a large area and it is possible that
some of the specimens with nodular cortex came from
greater distances. Likewise, individual specimens with
pebble cortex could have been collected within a few
km of the cave, though they could also have come from
more distant pebble beds. All of the raw materials which
we can attribute to a geological source could have been
collected within around 30 km from the site. Distances
of 2030 km lie outside the normal expected foraging radius of hunter-gatherers (Kelly, 1995, pp. 136141;
Surovell, 2000), but are still within one to two days travel of the site (given the steep topography). Specimens
from more distant locales may well be present but the
current state of knowledge about int sources in Turkey
does not permit us to identify them. Given the fact that
the most common variety of chipped stone raw material
was obtained from both primary secondary sources, we
cannot assess the origins of specimens that do not bear
cortex. In the analyses below it is assumed that cortexbearing pieces are representative of the larger population
of artifacts from the site.

Changing patterns of raw material exploitation


The rst order of business is to examine where the raw
cagzl cave might have been colmaterials found at U

439

lected. Table 1 shows the combined frequency of fresh


and rolled nodular cortex (i.e., cortex reecting acquisition at or near primary sources) among dierent artifact
classes in the assemblages from 13 combined stratigraphic units. Only cortex-bearing specimens are included in the calculation as only these specimens can be
assigned to a general source area with any degree of reliability. Several general trends are apparent. First, the
general frequency of cortex representing primary sources
increases over time (Fig. 4), the very small sample from
layer D representing an apparent interruption of the
trend. However, the assemblages from the two most recent layers, B and B1-B4, contain a good deal more fresh
and rolled nodular cortex than any of the others. This
increasing use of primary sources is also expressed within
the more common raw material class, the Cretaceous
int: in the earlier layers this kind of int seems to have
come mainly from pebble sources, whereas later on the
Cretaceous material came mainly from primary deposits.
As would be expected, proportions of nodular or
rolled nodule cortex, derived from more distant sources,
are almost always higher for retouched tools than for
akes, debris, or cores. In all layers, roughly half or
more of cortex on retouched tools is of the fresh or
rolled nodular varieties. Figures for unretouched akes,
debris, and cores show much wider ranges of variation
but, with two exceptions, layers B and B1-B4, the percentages of nodular cortex are always substantially lower than for retouched tools. This observation is
consistent with a general tendency observed across Middle and Upper Paleolithic sites for the most distant
sources of raw materials in a given site to be represented
more often by retouched tools than by debitage (FeblotAugustins, 1993; Geneste, 1988a,b), and ts with
theoretical expectations about relationships between
transport distance and artifact utility. We expect longdistance transport, whether in the context of provisioning places or individuals, to involve artifacts with the
highest ratio of utility to mass (Kuhn, 1994; Shott,
1986). Finished tools or prepared blanks contain more
potential utility and less waste than do cores or unworked chunks of stone. More formal arguments about
tradeos between transport and processing decisions
have been made for food resources (e.g., Metcalfe and
Barlow, 1992; OConnell et al., 1988).

Table 1
Frequencies of non-pebble (fresh and rolled nodule) cortex in various artifact and raw material classes

Retouched tools
Unretouched >2.5 cm
Debris
Cores
Cretaceous int only

B1-B4

C/D

Fa

Fb/Fc

H2/H3

83.6
87.8
91.0

91.3

91.7
80.0
90.1
70.0
85.6

76.0
61.4
42.6

72.2

(0/11)
51.9
42.6

(4/8)

(1/3)
18.7

73.7
47.0
47.7
45.6
57.6

65.2
32.2
38.2
40.7
31.0

62.5
44.2
35.6

38.5

48.2
27.6
35.0

22.1

45.5
20.2
40.6
9.1
7.9

58.1
29.7
34.4
16.7
12.7

65.3
18.2
43.9
20.0
5.1

60.0
22.3
35.8
10.0
16.7

Only cortical specimens counted.

440

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

When plotted against the assemblage-wide proportion of non-pebble cortex, cortex representing nodules
collected from near primary sources (Fig. 5), the disparity values suggest that, during layer B and B1-B4 times,
cagzl cave made a rather dierent
the occupants of U
set of decisions about where to collect raw materials
and what to transport to the cave compared with other
periods in the caves occupation. Not only did the sites
occupants use more material from primary sources at
this time, they apparently reduced cobbles of these materials at the cave, leaving behind many reduction byproducts with fresh or rolled nodular cortex and resulting in
low disparity values. In the other layers, artifacts bearing cortex from primary (non-pebble) sources are not
only scarcer, but they are represented mainly by retouched tools, appearing much less frequently as akes,
cores, and debris, resulting in higher disparity indices.
This second pattern indicates much more selective or
limited transport of material to the site from the primary
deposits. Judging by the cortex frequencies alone, the
preponderance of nodules worked in situ in layers C
and below came from secondary pebble beds, probably
located in close proximity to the site. In these earlier levels, material from more distant sources found its way

Fig. 4. Frequency of cortex representing primary sources (nonpebble), by assemblage.

A more suggestive trend can be observed from the


data presented in Table 2, the disparity in cortex proportions between retouched tools and unretouched
akes and debris. This disparity gure, modied after
the Robinson-Brainerd index of similarity (Robinson,
1951), is calculated summed absolute dierences in percentages of the three cortex types (nodular, rolled nodule, and pebble) among two groups of artifacts;
retouched tools and unretouched akes, cores and debris
combined. Potential values for the index vary from zero
(complete similarity) to 200 (complete dissimilarity).
Unlike the overall frequency of nodular cortex, the trend
over time in the disparity values is not smooth or monotonic. Instead, the index dierentiates the two most recent layers form the rest of the sequence. In
assemblages from layers B and B1-B4, the proportions
of dierent cortex types for unretouched and retouched
materials are very similar, with disparity values of 25.0
and 17.0, respectively. Values for layers I through C
are considerably higher than this for the most part, ranging between 43.2 and 136.9, with no particular directional stratigraphic trend. Only the disparity index for
layer G (29.1) approaches that for B and B1-B4.

Fig. 5. Frequency of non-pebble cortex plotted against disparity index.

Table 2
Disparity index comparing proportions of dierent cortex types among retouched tools and unretouched artifacts (cores, akes, and
debris)
Disparity index

B1-B4

C/D

Fa

Fb/Fc

H2/H3

25.0

17.0

72.8

137.9

50.0

53.1

65.6

73.7

43.2

29.1

60.5

66.2

74.3

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

into the site mainly in the form of retouched tools or


blanks that were subsequently retouched. Excluding layers B and B1-B4, it even appears that there is a positive
association between the disparity index and the amount
of non-local cortex in the assemblages, though the correlation value is not statistically signicant (r = 0.534,
p = 0.091, n = 11). We might expect such a relationship
if non-local raw materials were entering the site primarily in the form of nished tools. Fig. 5 also shows that
the low index of disparity for layer G means something
rather dierent than the low indices for layers B and B1B4. In the case of layer G, the disparity value is low because rolled and fresh nodular cortex are rare in all artifact classes, just the opposite of in the two most recent
levels.
Another perspective on the dierences in raw mate cagzl cave is provided by Fig.
rial economy within U
6, which plots the frequency of non-pebble (i.e., primary) cortex against the proportion of large blanks
(>2.5 cm) that were retouched. Retouch frequency is a
commonly used, if imperfect, proxy measure for the
intensity of raw material exploitation within archaeological assemblages. For any provisioning strategy we
might expect raw materials collected from distant
sources to be more extensively consumed, whether because of greater cost or as a result of the longer life histories of artifacts transported long distances (e.g.,
Andrefsky, 1994; Dibble, 1991; Odell, 2000). What
Fig. 6 shows is that there is a very clear positive trend
in the relationship between frequency of fresh or rolled
nodular cortex and retouch frequency among most of

Fig. 6. Frequency of non-pebble cortex plotted against proportion retouched pieces (retouched pieces/(retouched pieces
+ large blanks)).

441

cagzl assemblages: in general, the more exotic


the U
stone is represented, the more often blanks were retouched into tools. For layers C through I, this relationship is very strong, with a Pearsonss correlation
coecient of 0.814 (p = 0.002, N = 11). Here again,
however, the assemblages from layers B and B1-B4 represent conspicuous outliers. Although the cortex data
indicate that these assemblages are far and away the
most strongly dominated by materials from primary
sources, they do not have especially high frequencies
of retouched tools. What this suggests is that the cost
and/or the life histories of artifacts made on nodular
ints in layers B and B1-B4 were rather dierent from
other assemblages. The high frequencies of non-local
cortex on unretouched as well as retouched artifacts
from these layers reect something similar.
Fig. 7 provides yet another perspective on the economics of raw material use. Here, the Y-axis variable
is ratio of blanks (retouched tools and large akes) per
core: the X-axis again is the percentage of non-pebble
cortex in each assemblage. The full complement of
assemblages appears to form a roughly linear arrangement, although values for B and B1-B4 are conspicuously higher. There is even a weak but statistically
signicant correlation between the two variables (Pearsons r = 0.67, p = 0.022, n = 13). This would not be
an unexpected result if cost were proportional to distance to source, as we might expect cores of more costly
exotic stone to be more extensively exploited. However, the relationship disappears entirely when the
assemblages from B and B1-B4 are eliminated
(r = 0.048, p = 0.888, n = 11). In this case a relationship
appears only by virtue of the contrast between these two

Fig. 7. Frequency of non-pebble cortex plotted against ratio of


blanks per core ((tools + large blanks)/cores).

442

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

assemblages and those from the rest of the site. Once


again, it seems that int was being exploited in a rather
dierent way when layers B and B1-B4 were laid down.
A somewhat dierent perspective is provided by the
relationship between raw material source exploitation
and the treatment of retouched tools. Table 3 shows
the mean lengths of endscrapers for the ve assemblages
that provide statistically reliable samples of complete or
nearly complete endscrapers made on blanks with residual cortex. Because the sizes of the raw tool blanks certainly varied, the length of archaeological specimens
does not necessarily inform us about the amount of utility lost to resharpening and use. Instead, the length of a
discarded endscraper is a rough measure of the amount
of residual utility it contains. As such it is more indicative of decisions about when to abandon and/or replace
tools than it is of the length of the artifacts use life.
Several general patterns are apparent from the summary statistics in Table 3. First, there are no consistent
intra-assemblage dierences between endscrapers made
on blanks derived from primary and secondary sources
of int. In other words, regardless of the source of the
raw material used, endscrapers within a particular
assemblage tended to be reduced to the same average level of residual utility before discard. If blanks obtained
from pebbles indeed started out smaller, this had little
or no eect on the sizes of discarded pieces. On the other
hand, there are pronounced inter-assemblage dierences
in mean lengths of abandoned endscrapers. Scrapers

Table 3
Mean lengths of endscrapers, by cortex type, for the ve largest
assemblages
Layer

Pebble cortex

Non-pebble cortex

All endscrapers

B
Mean
SD
N

48.8
8.5
10

46.6
9.6
26

48.0
11.2
95

B1-B4
Mean
SD
N

44.8
11.9
23

46.4
10.7
53

47.7
10.2
250

F
Mean
SD
N

38.9
14.0
6

39.6
6.9
15

38.8
10.4
80

Fb/Fc
Mean
SD
N

35.4
11.5
22

38.9
11.1
16

37.4
11.3
82

H2/H3
Mean
SD
N

38.2
7.0
19

38.5
9.4
26

37.6
8.6
105

Table 4
ANOVA results, endscraper length by assemblage, for the ve
largest assemblages
ANOVA
results

Pebble
cortex only

Non-pebble
cortex only

All
endscrapers

df
Multiple r
F ratio
Prob.

75/4
0.419
3.98
0.006

131/4
0.357
4.78
0.001

607/4
0.434
35.15
<0.001

from the most recent layers (B and B1-B4) are on average 0.61.3 cm (roughly 2030%) longer than those from
layers F, Fb/Fc, and H2/H3. An analysis of variance
shows that there are signicant inter-assemblage dierences in scraper size for each raw material category as
well as for the entire sample (Table 4).
The data in Tables 1 and 3 also suggest that inter-assemblage variation in scraper size is related to general
levels of reliance on primary and secondary sources of
int in a somewhat counter-intuitive way. Endscrapers
from the layers with the highest proportions of nodular
cortex (B and B1-B4) tend to be larger than those from
layers with more reliance on local raw secondary raw
material pebble ints. This observation is paralleled by
an analysis of the sizes of complete tools, which shows
there to be a strong positive relationship between the
lengths of retouched tools and the amount of fresh
and rolled nodular cortex in an assemblage (r = 0.867,
p < 0.001, n = 12). The correlation between artifact
width and proportion of nodular cortex is not signicant
(r = 0.308, p = 0.330, n = 12), however, suggesting that
the larger tools were not simply made on bigger blanks.
(Note that the assemblage from layer D was excluded
due to an insucient sample of unbroken retouched
cagzl cave
tools.) In sum then, when toolmakers at U
relied more on distant primary sources they discarded
endscrapers and other tools when they still had some
residual utility, whereas when they were making greater
use of local pebble ints their discarded artifacts were
closer to complete exhaustion. Once again, cost (as
a function of distance from source) does not seem to
have been a primary determinate of decisions about
when to discard retouched tools.

cagzl
Discussion: changing provisioning strategies at U
cave and beyond
The relationships between the origins of lithic raw
materials and the intensity of raw material exploitation
cagzl cave are complex. The same rock types were
at U
utilized throughout the entire sequence to make similar
kinds of artifacts. Over time we see quantitative shifts
in the reliance on primary and secondary deposits of
int. However, the general increase in reliance on more

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

distant primary sources does not seem to have occurred


in the context of a single basic kind of raw material
economy. For most of the Upper Paleolithic sequence,
objects made of int collected at primary sources seem
to have entered the cave most often in the form of nished tools and/or large blanks. Judging by the kinds
of cortex present, most of the cores, unretouched akes
and debris in these early layers are attributable to pebbles from secondary pebble deposits located much closer
to the cave. In contrast, in the two most recent Upper
Paleolithic assemblages, from layers B and B1-B4, nodular cortex dominates in all categories, from retouched
tools to cores and small debris, indicating that cobbles
cagzl
or partially prepared cores were transported to U
cave from the primary sources located 2030 km inland
and then further reduced in the cave.
Fig. 6, showing the relationship between retouch frequency and proportion of non-pebble/non-local cortex,
is especially signicant. We might expect a positive correlation between these variables in the context of any
sort of provisioning strategy, but for rather dierent
reasons. If raw materials were transported from distant
sources mainly as parts of personal toolkits, most
would arrive at a site in the form of well-used retouched
pieces. As such artifacts accumulated they would simultaneously raise the proportions of both non-local raw
materials and retouched pieces, relative to artifacts produced in situ from local raw materials. If, on the other
hand, materials from non-local sources arrived as a result of provisioning places we might anticipate them to
be more extensively exploited simply because of their
higher cost. The point is that these two pathways
should result in dierent relationships, and dierent
correlations between proportions of non-local stone
and retouched pieces. The fact that layers B and B1B4 represent conspicuous outliers to an otherwise
strong correlation for the other 11 assemblage suggests
cagzl
that this may be exactly what is going on at U
cave.
Other factors suggest that the assemblages from layers B and B1-B4 represent a distinct set of economic
cagzl
decisions from the rest of the material from U
cave. Within individual layers tools were discarded in
similar condition regardless of the material of which
they were made. However, greater reliance on potentially more costly inland primary int deposits is associated with apparently more wasteful treatment of tools,
in the sense that discarded specimens tend to be larger
where non-local sources were most heavily exploited.
On the other hand, cores from the two layers with the
highest proportion of raw material from primary
sources, B and B1-B4, do seem to have been more extensively exploited than in other assemblages. In other
words, it seems that the amount of non-local raw material aected the extent of core reduction and retouch frequencies, but not in the same ways.

443

Terms such as local, exotic, or distant are common in


the archaeological literature on lithic raw material economies, yet there is little comparability between studies.
Researchers working with late Upper Paleolithic populations on the plains of Eastern Europe or with early
Paleoindians may consider transfers of large amounts
of int over several hundred km as routine events
(e.g., Amick, 1996; Hofman et al., 1991; Tankersley,
1994). In contrast, analysts concerned with the behavior
of Plio-Pleistocene hominids (e.g., Stiles, 1998; Toth,
1985) or the technologies of sedentary Neolithic groups
may view any material coming from more than a days
walk from its nd spot as exotic As a matter of practice, the lines between exotic and local are often drawn
from observations of the archaeological material itself,
and the criteria used to dene exotic or local materials
depend more on how material was treated than on
how far it actually had to be moved. Any raw material
that is represented by a full range of products and
byproducts seems local. Types of stone that occur as
a restricted range of products, whether standardized
forms (e.g., objects of trade) or as extensively used and
reworked tools, appear more unusual, and their sources
are treated as exotic or distant (e.g., Blades, 1999, pp.
113114). From this perspective, the primary sources
cagzl cave
of int located some distance inland from U
appear to have shifted from being exotic (represented
by a restricted range of products) through most of the
sequence to being more local, with a full range of
reduction byproducts present, in the most recent layers,
B and B1-B4.
It would be dicult to explain the observed changes
cagzl cave in terms of
in raw material economy at U
accessibility or distance-to-source as a proxy for the cost
of obtaining raw materials. Barring an undocumented
tectonic event of truly enormous magnitude, the geographic distances from sources to site probably did not
change. Although there could be primary sources of
Cretaceous int even closer to the site under the surface
of the sea, there is no reason to think sea levels were lower during the more recent end of the sequence. In fact,
the abundance of mollusk remains in the faunal assemblages from layers B and B1-B4 indicates that if anything sea levels were relatively high, and the coastline
relatively close, when these deposits were accumulating.
Of course, centuries of exploitation might have partially
depleted local pebble beds, forcing a greater reliance on
more distant sources of int, but this would not explain
the apparently wasteful treatment of ints from inland sources in the most recent layers.
Concepts of technological provisioning, combined
with independent archaeological evidence concerning
cagzl cave, are more
the character of occupations at U
useful for making sense of the changing roles of dierent
raw materials. Several indicators suggest that the nature
and duration of occupations changed over time at

444

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

cagzl cave. Where thermal structures are present at


U
all in the earlier part of the sequence (below layer C)
they tend to be small, discrete hearths or thin scatters
of ash. This suggests that they formed as result of a series of relatively brief occupational events. Interestingly,
the vertebrate faunal assemblages from these layers consist almost entirely of bones from medium to large game
animals. From the perspective of general ecological theory, such heavy reliance on high-ranked resources such
as medium/large terrestrial herbivores generally implies
relatively low consumer demand relative to the sizes of
prey populations. Demands on resources can be kept
small either by keeping populations small or by limiting
time spent in any one foraging area. Holding environment constant, cross cultural studies show that high levels of dependence on hunted game are usually associated
with frequent residential mobility (Kelly, 1995, pp. 111
160). Thus, both patterns of game use and the archaeo cagzl cave
logical structures in the earliest layers at U
point to relatively frequent residential moves and short
episodes of occupation during the earlier part of the
sequence.
In contrast, layer B1-B4 is characterized by a thick,
dense midden-like accumulation of ash, stone, bone,
and other materials. This alone points to a more continuous occupation of the site, perhaps by a larger group.
No such midden is apparent in layer B, but this could
simply be a matter of changing use of space within the
cave. The faunal assemblages from both B and B1-B4
also provide evidence for expansion of diet breadth to
include signicant amounts of shellsh, birds, rabbits,
sh, and perhaps even vegetable foods (in the form the
pitted anvils). Hints of this dietary expansion are present
earlier, in that these foods were occasionally incorporated into the diet in layers C and below, but their roles
expanded signicantly in B and B1-B4. Holding environment more or less constant, such broadening of the diet
is commonly taken to be a response to excessive pressure
on high-ranked resources such as hunted game, brought
on by larger resident consumer groups and/or a more
permanent human presence (Kelly, 1995, pp. 111160;
Kuhn and Stiner, 2001; Stiner et al., 1999, 2000). It is
tempting to suggest that the composition of residential
groups could also have changed over this span. In the
early part of the sequence the cave could well have been
used mainly as a specialized hunting camp, occupied by
a restricted range of individuals concerned mainly with
procuring game. The more prolonged occupation(s) in
the most recent layers imply a larger and more diverse
group of occupants, some of whom (women and children, for example) may have specialized in collecting
shellsh, birds, and other relatively small prey.
In light of this evidence for changing nature and
duration of occupations, the shifts in raw material econ cagzl cave are more comprehensible. For
omy at U
most of the sequence occupations were short and epi-

sodic, perhaps accompanied by high overall levels residential mobility. Many retouched tools and blanks
came to the cave from distant sources as parts of transported toolkits used to provision individuals. Some of
these artifacts were subsequently abandoned and replaced on site using local ints. The fact that scrapers
tended to be reduced to relatively small size suggests that
tool users tried to get the most out of artifacts before
abandoning them, a characteristic of mobile toolkits
used to provision individuals (Kuhn, 1992, 1994). Early
in the sequence Upper Paleolithic tool makers also provisioned the site with raw materials, but, as would be expected for short occupations, they focused on materials
found closer to the site, which in this case happened to
be beach pebbles. Because the strategies of provisioning
places and individuals tended to involve raw materials
from dierent sources their respective products stand
out clearly.
In layers B and B1-B4 at the top of the sequence there
is evidence for more intense, longer occupations, and
provisioning the place with raw material became a more
viable and important strategy. Along with daily foraging
radius, catchments for raw material procurement would
have expanded with longer occupations (Kelly, 1995, pp.
135137). The later Upper Paleolithic foragers were able
to provision the cave with large nodules or partially prepared cores of better materials from primary sources
two or three days travel away: this could have occurred
as part of long-distance hunting or collecting forays or
as specialized trips to collect raw material. The fact that
discarded scrapers retained greater residual utility than
in the earlier layers suggests that constraints of raw
material cost were actually somewhat reduced through
amassing raw materials at the residential hub, even when
most of the int came from fairly far away. Individual
transported toolkits were undoubtedly still part of the
strategic mix, but because even in situ manufacture involved raw materials obtained from primary sources,
elements of transported toolkits are largely
undistinguishable.
cagzl cave, the eects of raw material cost,
At U
as measured by distance to source area, are also expressed dierently on cores, blanks, and tools. There
seems to be little relationship between the treatment of
tools and transfer distances for the raw materials of
which they were made. Within levels, endscraper reduction does not vary according to raw material source, and
discarded tools are actually larger (less reduced) in those
layers characterized by greater reliance use of inland primary int sources. The frequency of retouched blanks
(akes and blades) shows more of the eects of raw
material costs and provisioning strategies. There are
no clear patterns in the treatment of cores overall, but
the two assemblages with the most exotic raw material
(B, B1-B4) also have the highest blank/core rations.
These dierences between artifact classes could be a

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

function of cores, blanks, and tools entering the site


through dierent pathways, as a result of dierent mixes
of provisioning strategies. For most of the sequence it
appears that elements of transported toolkits used to
provision individuals would be signicant components
of the archaeological assemblages. Such transported
toolkits are expected to consist mainly of retouched
tools and blanks (Kelly and Todd, 1988; Kuhn, 1994;
Morrow, 1995). Because transported toolkits are not
brought to places but are carried along with mobile individuals, the cost of transport is not a direct function
of distance to source. At the end of the sequence, in layers B and B1-B4, it seems that the site was heavily provisioned with raw materials from inland sources,
relaxing constraints on portability and on getting the
maximum use out of every endscraper. In contrast, unretouched akes and blades could have come to the site
through a variety of channels, either as parts of transported toolkits or through on-site production. It is not
surprising therefore that the intensity of blank utilization shows a stronger relationship with raw material
sources than does the treatment of tools, though again
layers B and B1-B4 stand out.
These characterizations of occupations and provisioning strategies are highly generalized. Certainly, the
lengths of occupations and strategies of technological
provisioning varied within the earlier part of the sequence: variation in amounts of exotic material and
retouched tool frequencies among layers C through I is
probably evidence of this. By that same token, they
may well be very brief occupational episodes represented
in layers B and B1-B4, even if the signal is overwhelmed
by the more intense occupation. Likewise, individuals or
classes of individual within residential groups may have
dealt with lithic raw materials in dierent ways according to activity regimes, mobility, and time constraints.
However, such ne-grained intra-assemblage variability
is dicult to sort out given the constraints on this
analysis.
It is interesting that the major shift in lithic raw mate cagzl cave is independent of the
rial economy at U
archaeological culture represented. The Initial Upper
Paleolithic component ends at the boundary between
layers F and E. The assemblage from layer C is clearly
Ahmarian in character, and the sparse material from
C/D, D and E is closer to the Ahmarian than to the
IUP. Nonetheless, in terms of treatment of lithic raw
materials these assemblages t much better with the earlier part of the sequence than they do with layers B and
B1-B4. On the other hand, changing provisioning strategies may explain a technological anomaly in the
cagzl sequence. Dorsal scar patterns on akes and
U
tools indicate an increasing use of bidirectional/opposed
platform cores beginning with layer E, and yet opposed
platform cores are common only in layers B and B1-B4
(Kuhn, 2004). This anomaly may reect changing locus

445

of blade production, and eorts to get the most out of


cores of int from distant sources in the most recent
Upper Paleolithic occupations. As Fig. 7 shows, the B
and B1-B4 assemblages do have the highest blank/core
ratios by far in the entire sequence.
cagzl cave is instructive, it
Although the case of U
presents no hard and fast rules. It is likely that the relationships between raw material transfer distances and
the treatment of dierent classes of artifact will vary in
dierent contexts. The most remote sources of int that
we currently are able to identify are not particularly far
cagzl cave (around 30 km) and if more distant
from U
sources are represented we lack the geological information needed to identify them. Much greater distances
of raw material movement have been documented in
other Upper Paleolithic sites (Demars, 1998; Dobosi,
1991; Geneste, 1988a,b; Schild, 1987), southern African
MSA localities (Ambrose and Lorenz, 1990; McBrearty
and Brooks, 2000), and even some Middle Paleolithic
cases (see Feblot-Augustins, 1993; Roebroeks et al.,
1988). Provisioning strategies notwithstanding, there
will almost certainly be signicant relationships between
distance to source and the treatment of even retouched
tools in cases where transfer distances were much
greater.
The results reported in this paper may nonetheless be
relevant to studies of earlier time periods. A number of
researchers have addressed generic dierences between
patterns of lithic raw material exploitation in the Middle
and Upper Paleolithic of Eurasia as evidence of changes
in basic cognitive abilities of hominids, though there is
little consensus as to either the empirical facts or their
interpretation (e.g., Mellars, 1996; Roebroeks et al.,
1988; Soer, 1989). Because we are presently unable to
determine actual distance to source for individual arti cagzl cave it is dicult to compare the results
facts at U
reported here directly to those of other studies. However, the changes in raw material economy within the
cagzl do have implicaUpper Paleolithic sequence at U
tions for how one might investigate hypothetical contrasts between Middle and Upper Paleolithic.
It seems well established now that Eurasian Middle
Paleolithic hominids habitually transported artifacts over
distances of up 50 km, and occasionally much farther
more (Feblot-Augustins, 1993; Roebroeks et al., 1988).
If there is any qualitative dierence between the Middle
and Upper Paleolithic, it has to do with what was moved.
During the Middle Paleolithic, raw material transfers in
excess of 1020 km almost always involved prepared
tools, tool blanks, and sometimes Levallois cores (Feblot-Augustins, 1993; Geneste, 1988a,b). Evidence for
moving materials in bulk more than a dozen kilometers,
while certainly not universal, is more common in Upper
Paleolithic contexts (Mellars, 1996, pp. 163165; Soer,
1989). What this suggests is that Mousterian hominids
regularly engaged in provisioning of individuals (Kuhn,

446

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

1992) but that they less often provisioned sites with highquality materials from more distance sources.
Even if this general assessment proves correct, it does
not necessarily demonstrate that Middle Paleolithic
hominids had fundamentally dierent behavioral capacities from Upper Paleolithic populations. I am certainly
not prepared to argue that the changes in raw material
cagzl cave
economy within the Upper Paleolithic at U
are due to cognitive evolution in the populations that
occupied the site (although that is not impossible). Instead, they can be understood as responses to changes
in mobility and the nature of the occupation at the site
itself. By that same token, one would not want to argue
that shrinking catchments of raw material procurement
across the Paleoindian to Archaic transition in North
America represented a decline in humans abilities of
anticipate future needs. Dierences in habitual raw
material transfers between Mousterian and Upper
Paleolithic could reect shifting patterns of land use
resulting from long-term climatic, demographic, and social changes rather than changes in cognitive abilities.
Using raw material transfers to help gauge the planning
abilities of hominids would require comparing technological responses to similar mobility regimes or the provisioning of similar kinds of occupations.
The ndings discussed here reinforce the assertion
that distance to source is an imperfect, sometimes misleading, measure of raw material cost. As a consequence
distance to source is a poor predictor of human behavior, even when people are responding mainly to economic constraints. Artifacts can have widely varying
histories, and they can follow many dierent pathways
from quarry to archaeological context. Dierent measures of cost may be more appropriate for dierent kinds
of artifact life histories. In the case of transported personal gear, portability or utility per unit weight may be
the most appropriate measure of cost, whereas the cost
of material used to provisioned sites is more directly a
function of how far it had to be carried. Conversely,
however, the existence of unexpected relationships, or
the absence of apparent relationships between distance
to source and treatment of artifacts (e.g., Close, 1999;
Gould and Saggers, 1985) does not necessarily imply
that cost is irrelevant. Considering the dierent pathways along which artifacts and raw materials can move,
the dierent strategies that result in things being transferred from procurement spots to the archaeological
sites where they are eventually found, such cases suggest
instead that the economics of raw material use are simply more complex, and ultimately more interesting.

Acknowledgments
cagzl cave are a collaboration beExcavations at U
tween the University of Arizona and Ankara University.

The success of the project is attributable in large part to


the eorts of Professor Erksin Gulec, leader of the Turkish team, as well as her associates and graduate students,
including Aysen Acikol, Ismail Baykara, Dr. Ismail
zer, and Hakan Yilmaz. Dr. Mary Stiner contributed
O
innumerable insights on the Paleolithic occupations of
cagzl, as well as the faunal analyses that are briey
U
summarized here. My understanding of lithic raw material economies has been very much improved as a result
of working with current and former graduate students
Jesse Ballenger, Dr. P.J. Brantingham, Kris Kerry, and
Dr. Todd Surovell. Kris Kerry also did the artifact illustrations. I am also grateful to two anonymous reviewers
for insightful and constructive comments. This research
reported in this was supported by the National Science
Foundation under Grants SBR-9804722 and BCS0106433.

References
Ambrose, S.H., Lorenz, K.G., 1990. Social and ecological
models for the Middle Stone Age in Southern Africa. In:
Mellars, P. (Ed.), The Emergence of Modern Humans.
Edinburgh University Press, Edinburgh, pp. 333.
Amick, D.S., 1996. Regional patterns of Folsom mobility and
land use in the American Southwest. World Archaeology
27, 411426.
Andrefsky, W., 1994. Raw material availability and the
organization of lithic technology. American Antiquity 59,
2134.
Azoury, I., 1986. Ksar Akil, Lebanon: A technological and
typological analysis of the transitional and early Upper
Palaeolithic Levels of Ksar Akil and Abu Halka. B. A. R.
International Series 289. British Archaeological Reports,
Oxford.
Binford, L., 1979. Organization and formation processes:
looking at curated technologies. Journal of Anthropological
Research 35, 255273.
Blades, B., 1999. Aurignacian lithic economy and early modern
human mobility. Journal of Human Evolution 37, 91120.
Boesch, C., Boesch, H., 1984. Mental maps in wild chimpanzees: an analysis of hammer transports for nut cracking.
Primates 25 (2), 160170.
Bourque, B.J., 1994. Evidence for prehistoric exchange on the
Maritime peninsula. In: Baugh, T., Ericson, J. (Eds.),
Prehistoric Exchange Systems in North America. Plenum
Press, New York, pp. 2346.
Close, A.E., 1999. Distance and decay: an uneasy relationship.
Antiquity 73, 2432.
Delage, C., 2001. Les resources lithiuques dans le nord dIsrael:
la question des territoires dapprovisionnement Natouens
confronta la hypothese de leur sedentarite. Institut dArt et
dArcheologie. Paris, Universite de Paris I.
Demars, P.-Y., 1982. Lutilisation du silex au Paleolithique
superieur: choix, approvisionnement, circulation. Lexemple
du Bassin de Brive. CNRS, Paris.
Demars, P.-Y., 1998. Circulation des silex dans le nord de
laquitaine au Paleolithique Superieur: loccupation de

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448


lespace par les derniers chasseurs-cueilleurs. Gallia Prehistoire 40, 128.
Dibble, H.L., 1991. Local raw material exploitation and its
eects on Lower and Middle Paleolithic assemblage variability. In: Montet-White, A., Holen, S. (Eds.), Raw
Material Economies among Prehistoric Hunter-Gatherers.
University of Kansas, Lawrence, pp. 217230.
Dobosi, V.T., 1991. Economy and raw material: a case study of
three Upper Palaeolithic sites in Hungary. In: MontetWhite, A., Holen, S. (Eds.), Raw Material Economies
among Prehistoric Hunter-Gatherers. University of Kansas,
Lawrence, pp. 197203.
Feblot-Augustins, J., 1993. Mobility strategies in the late
Middle Palaeolithic of central Europe and western Europe:
elements of stability and variability. Journal of Anthropological Archaeology 12, 211265.
Frison, G., Stanford, D., 1982. The Agate Basin Site: A Record
of the Paleoindian Occupation of the Northwestern High
Plains. Academic Press, New York.
Geneste, J.-M., 1988a. Les industries de la Grotte Vaufrey:
Technologie du debitage, economie, et circulation de la
matie`re premie`re lithique. In: La Rigaud, J.-P., (Ed.), Grotte
Vaufrey: Paleoenvironnement, Chronologie, Activites
Humaines. Societe Prehistorique Francaise, Paris, pp. 441
518.
Geneste, J.M., 1988b. Syste`mes dapprovisionnement en matie`res premie`res au Paleolithique moyen et au Paleolithique
superieur en Aquitaine. In: Kozlowski, J.K. (Ed.), Lhomme
de Neandertal: Actes du Colloque International de Lie`ge:
Vol. 8, La Mutation. Etudes et Recherches Archeologiques
de lUniversite de Lie`ge (ERAUL), Lie`ge, pp. 6170.
Goodyear, A.C., 1989. A hypothesis for the use of cryptocrystalline raw materials among Paleoindian groups of
North America. In: Ellis, C., Lothrop, J. (Eds.), Eastern
Paleoindian Lithic Resource. Westview Press, Boulder, pp.
110.
Gould, R., Saggers, L., 1985. Lithic procurement in central
Australia: a close look at Binfords idea of embeddedness in
archaeology. American Antiquity 50, 117136.
Hofman, J., 1991. Folsom land use: projectile point variability
as a key to mobility. In: Montet-White, A., Holen, S. (Eds.),
Raw Material Economies among Prehistoric Hunter-Gatherers. University of Kansas Publications in Anthropology
19, University of Kansas, Lawrence, pp. 285303.
Hofman, J., Todd, L.C., Collins, M., 1991. Identication of
central Texas Edwards Chert at the Folsom and Lindenmeier sites. Plains Anthropologist 36, 297308.
Hovers, E., 1990. The exploitation of raw material at the
Mousterian site of Quneitra. In: Goren-Inbar, N. (Ed.),
Quneitra: A Mousterian Site on the Golan Heights. Qedem
13. Institute of Archaeology, Hebrew University, Jerusalem,
pp. 153170.
Hughes, R.E., 1994. Mosaic patterning in prehistoric CaliforniaGreat Basin exchange. In: Baugh, T., Ericson, J.
(Eds.), Prehistoric Exchange Systems in North America.
Plenum Press, New York, pp. 363383.
Kelly, R., 1988. The three sides of a biface. American Antiquity
53 (4), 717734.
Kelly, R., 1995. The Foraging Spectrum: Diversity in HunterGatherer Lifeways. Smithsonian Institution Press, Washington, DC.

447

Kelly, R.L., Todd, L.C., 1988. Coming into the country: early
Paleoindian hunting and mobility. American Antiquity 53
(2), 231244.
Kuhn, S., 1992. On planning and curated technologies in the
Middle Paleolithic. Journal of Anthropological Research
48, 185214.
Kuhn, S., 1994. A formal approach to the design and assembly
of transported toolkits. American Antiquity 59 (2), 426442.
Kuhn, S., 1995. Mousterian Lithic Technology: An Ecological
Perspective. Princeton University Press, Princeton.
Kuhn, S., 2002. Economies de mate`ria primera ltica, provement tecnolo`gic dina`miques sociales. Cota Zero: Revista
dArqueologia I Cie`ncia 17, 5666.
Kuhn, S., 2004. From Initial Upper Paleolithic to Ahmarian at
cagzl Cave, Turkey. Anthropologie (Brno). Special issue
U
42, 275288.
Kuhn, S., Stiner, M., 2001. The antiquity of hunter-gatherers.
In: Panter-Brick, C., Layton, R.H., Rowley-Conwy, P.
(Eds.), Hunter-Gatherers, an Interdisciplinary Perspective.
Cambridge University Press, Cambridge, pp. 99142.
Kuhn, S., Stiner, M.C., Gulec, E., 1999. Initial Upper Paleolithic in south-central Turkey and its regional context: a
preliminary report. Antiquity 73, 505517.
Kuhn, S., Stiner, M.C., Gulec, E., 2004. New perspectives on
cagzl cave,
the Initial Upper Paleolithic: the view from U
Turkey. In: Brantingham, P.J., Kuhn, S., Kerry, K.W.
(Eds.), The Early Upper Paleolithic Beyond Western
Europe. University of California Press, Berkeley, pp. 113
128.
Kuhn, S., Stiner, M.C., Reese, D., Gulec, E., 2001. Ornaments
in the Earliest Upper Paleolithic: new perspectives from the
Levant. Proceedings of the National Academy of Sciences of
the United States of America 98 (13), 76417646.
MacDonald, D.H., 1999. Modeling Folsom mobility, technological organization and mating strategies in the Northern
Plains. Plains Anthropologist 44, 141161.
McBrearty, S., Brooks, A., 2000. The revolution that wasnt: a
new interpretation of the origin of modern human behavior.
Journal of Human Evolution 39, 453563.
Meltzer, D., 1989. Was stone exchanged among Eastern North
American Paleoindians?. In: Ellis, C., Lathrop, J. (Eds.),
Eastern Paleoindian Lithic Resource Use. Westview Press,
Boulder, pp. 1139.
Mellars, P., 1996. The Mousterian Legacy. Princeton University
Press, Princeton.
Metcalfe, D., Barlow, R.K., 1992. A model for exploring the
optimal trade-o between eld processing and transport.
American Anthropologist 94 (2), 340356.
cagzl magara, un site aurignaMinzoni-Deroche, A., 1992. U
cien dans le Hatay (Anatolie). Premiers resultats. Paleorient
18/1, 8996.
Morrow, T., 1995. Bigger is better: comments on Kuhns
Formal approach to mobile Toolkits. American Antiquity
61, 581590.
Nelson, M.C., 1991. The study of technological organization.
In: Schier, M.B. (Ed.), Archaeological Method and Theory. University of Arizona Press, Tucson, pp. 57100.
OConnell, J., Hawkes, K., Blurton-Jones, N., 1988. Hadza
hunting, butchering and bone transport and their archaeological implications. Journal of Anthropological Research
44, 113161.

448

S.L. Kuhn / Journal of Anthropological Archaeology 23 (2004) 431448

Odell, G.H., 2000. Stone tool research at the end of the


millennium: procurement and technology. Journal of
Archaeological Research 8, 269331.
Ohnuma, K., Bergman, C.A., 1990. A technological analysis of
the Upper Palaeolithic Levels (XXV-VI) of Ksar Akil,
Lebanon. In: Mellars, P. (Ed.), The Emergence of Modern
Humans. Edinburgh University Press, Edinburgh, pp. 91
138.
Parry, W., Kelly, R.W., 1987. Expedient core technology and
sedentism. In: Johnson, J., Morrow, C. (Eds.), The Organization of Core Technology. Westview Press, Boulder, pp.
285304.
Perle`s, C., 1990. LOutillage de pierre taillee Neolithique en
Gre`ce: Approvisionnement et exploitation des matie`res
premie`res. Bulletin de Correspondance Hellenique 114, 142.
Robinson, W.S., 1951. A method for chronologically ordering
archaeological deposits. American Antiquity 16, 293301.
Rockman, M.H., 2003. Landscape learning in the late glacial
recolonization of Britain. Ph.D. Dissertation. Department
of Anthropology, University of Arizona. UMI, Ann Arbor.
Roebroeks, W., Kolen, J., Rensink, E., 1988. Planning depth,
anticipation and the organization of Middle Palaeolithic
technology: The Archaic Natives meet Eves Descendents.
Helinium XXVIII, 1734.
Roth, B.J., Dibble, H.L., 1998. Production and transport of
blanks and tools at the French Middle Paleolithic site of
Combe-Capelle Bas. American Antiquity 63 (1), 4762.
Schild, R., 1987. The exploitation of chocolate int in Poland.
In: Sieveking, G.G., Newcomer, M. (Eds.), The Human
Uses of Flint and Chert. Cambridge University Press,
Cambridge, pp. 137149.
Shott, M., 1986. Technological organization and settlement
mobility: an ethnographic examination. Journal of Anthropological Research 42, 1551.
Shott, M., 1997. An exegesis of the curation concept. Journal of
Anthropological Research 52, 259280.
Soer, O., 1989. The Middle to Upper Paleolithic transition on
the Russian Plain. In: Mellars, P., Stringer, C. (Eds.), The
Human Revolution: Behavioural and Biological Perspectives on the Origins of Modern Humans. Princeton
University Press, Princeton, pp. 714742.
Stiles, D., 1998. Raw material as evidence for human behaviour
in the Lower Pleistocene: the Olduvai case. In: Petraglia,
M.D, Korisettar, R. (Eds.), Early Human Behaviour in
Global Context: The Rise and Diversity of the Lower
Palaeolithic Record. Routledge, London, pp. 133150.
Stiner, M.C., Munro, N., Surovell, T.A., Tchernov, E., BarYosef, O., 1999. Paleolithic population growth pulses
evidenced by small game use. Science 283, 190194.

Stiner, M.C., Munro, N.D., Surovell, T.A., 2000. The tortoise


and the hare: small game use, the Broad Spectrum Revolution and Paleolithic demography. Current Anthropology
41, 3973.
Surovell, T.A., 2000. Early Paleoindian women, children,
mobility, and fertility. American Antiquity 65, 493508.
Surovell, T.A., 2003. The behavioral ecology of Folsom Lithic
technology. Ph.D. Dissertation, Department of Anthropology, University of Arizona. UMI, Ann Arbor.
Tankersley, K., 1994. The eects of stone and technology on
uted point morphometry. American Antiquity 59, 498
510.
Thacker, P., 1996. Hunter-gatherer lithic economy and settlement systems: understanding regional assemblage variability in the Upper Paleolithic of Portugese Extremadura. In:
Odell, G.H. (Ed.), Stone Tools: Theoretical Insights into
Human Prehistory. Plenum Press, New York, pp. 101124.
Toth, N., 1985. The Oldowan reassessed: a close look at early
stone artifacts. Journal of Archaeological Science 12, 101
120.
Torrence, R., 1983. Time budgeting and hunter-gatherer
technology. In: Bailey, G. (Ed.), Hunter-Gatherer Economy
in Prehistory: A European Perspective. Cambridge University Press, Cambridge, pp. 1122.
Torrence, R., 1989. Retooling: towards a behavioral theory of
stone tools. In: Torrence, R. (Ed.), Time, Energy and Stone
Tools. Cambridge University Press, Cambridge, pp. 5766.
Trassierra, M.D.G.-A., Sabater, N.M., Mallol Duque, C., 2002.
Lapprovisionment en matie`res premie`res lithiques au
Pleistoce`ne inferieur et moyne dans la Sierra de Atapuerca,
Burgos (Espagne). LAnthropologie 106, 4155.
Turq, A., 1996. Lapprovisionnement en matie`re premie`re
lithique au Mousterien et au debut du Paleolithique
superieur dans le Nord est du bassin aquitain: continuite
ou discontinuite?. In: Carbonell, E., Vaquero, M. (Eds.),
The Last Neandertals, the First Anatomically Modern
Humans. Universitat Rovira i Virgili, Tarragona, pp. 355
362.
Wengler, L., 1990. Economie de matie`res premie`res et territoire
dans le mousterien et lAterian maghrebins: Exemples du
Maroc oriental. LAnthropologie 94 (2), 335360.
White, M., 1995. Raw materials and biface variability in
southern Britain: a preliminary examination. Lithics 15, 1
20.
White, M., 1998. On the signicance of Acheulean biface
variability in southern Britain. Proceedings of the Prehistoric Society 64, 1544.
Wynn, T., McGrew, W., 1989. An apes view of the Oldowan.
Man (N.S.) 24, 383398.

You might also like