Professional Documents
Culture Documents
Abstract 2
Introduction 3
Background 4
Lithic Typology 4
Group Mobility 9
Site Function 10
Mobility 29
Site Function 31
Discussion 32
Conclusion 35
Bibliography 37
1
Abstract
Research on the Epipaleolithic in the Southern Levant has led to the identification of a large number of
regional lithic industries. These industries are differentiated from one another almost exclusively on the
basis of proportional differences in specific microlith types (Henry 1995; Goring Morris 1995). The
validity of such a typology-based approach has been challenged on numerous occasions, both for the
Epipaleolithic of the Southern Levant as well as elsewhere (Neeley & Barton 1994; Bisson 2000; Barton
1991; Clark 1997; Dibble 1995). Because of criticisms of typological systematics over the past couple
decades, research on stone artifact assemblages has changed considerably, especially in the New World.
This shift emphasizes factors like raw material availability and use, mobility as related to land-use patterns,
artifact portability, site function, and risk as it relates to technological needs. These situational and
contextual factors play important roles in hunter-gatherer adaptations everywhere. They strongly influence
the characteristics of lithic assemblages. Using surface distributions associated with two large Levantine
Epipaleolithic rockshelters as a test case, lithic variability is investigated by examining the relationships
among (1) raw material use, (2) mobility, and (3) site function, while controlling for raw material
availability and local environmental factors. A standard typology is incorporated to provide a point of
reference. It is shown that explanations based on these general contextual variables differ markedly from
2
Introduction
In our effort to understand past human behavior, we assign meaning to contrasts observed in the
archaeological record. A first step is to distill variability from rocks or any other type of material remains
present at a site or in a collection. This can be done through a variety of procedures, one of which is
typology the organization of lithic artifacts into retouched stone tool types. The categorization, however,
is not an end in itself. A second step is required, one that explains this variability in the context of an
explicit conceptual framework. Historically the most common explanation for variation has been
implicitly or explicitly culture or some other form of identity consciousness expressed materially. The
Bordes-Binford debate of the mid 1970s (Binford & Binford 1966; Bordes & de Sonneville-Bordes 1970;
Binford 1973) stimulated archaeologists to examine more critically the causes of lithic variability, resulting
in the realization that many factors play a role. Two of these raw material availability and group mobility
are considered important in explaining observed variability. Other factors, such as the buffering of risk, site
function, provisioning strategies, style, and land-use patterns also figure in current approaches. It is
important to recognize that these factors do not exist in isolation from one another, but are interrelated in
complex ways, making the task of teasing them apart extremely difficult. This study will focus on three
sources of variability deemed particularly important in the organization of lithic technologies: (1) raw
material availability and use, (2) mobility, and (3) site function. Each of these will be discussed in turn
below. Insights gained from this discussion will be used to generate hypotheses and then be applied to a
case study which analyses surface samples from two Epipaleolithic rockshelter sites in west-central Jordan.
A subset of the artifacts collected will be scored on raw material type, debitage-, or tool-type, artifact
completeness, length, width, thickness, and weight. Pattern searches on these variables are then used to
assess expectations derived from the discussion of raw material availability and use, group mobility, and
site function to arrive at an initial understanding of these sites that goes beyond categorization of the
3
Background
This study is based on the view that typological and technological constructs are a means to
capture lithic variation in assemblages, not research ends in themselves. Standard typologies like those of
Bordes are only useful insofar as they help us to address specific research questions (and perhaps for
general communicative purposes). However, a general typology is probably inadequate to tackle particular
research questions, as it might confound elements of multiple causality. For these reasons it is important to
attempt to go beyond standard typology if the goal is to further our understanding of hunter-gatherer
lifeways. However, since most Levantine research relies on standard typology, it is important to incorporate
this into any study in order to make results comprehensible to other workers.
Lithic Typology
In the Southern Levant, as in many regions of the Old World, industries are described in terms of
the presence or absence of particular archaeological index tool types or fossiles directeurs, differential
proportions of those tools, and modes of retouch (Byrd 1998; Sellars 1998; Neeley 1998; Goring Morris
1995; Al-Nahar 2000). This practice has led to the definition of several industries for the Levantine
Epipaleolithic (ca 20-10.5 kyr BP) (Byrd 1998; Neeley et al. 1998) focusing almost exclusively on
microlith frequencies. The Epipaleolithic of the southern Levant can be divided in three periods: Early (20-
14.5 kyr BP), Middle (14.5-12.5 kyr BP), and Late (12.5-10.5 kyr BP) (Neeley et al. 1998; Byrd 1998;
Sellars 1998; Al-Nahar 2000; Olszewski n.d.). Within and occasionally crosscutting these broad
subdivisions are a number of primary complexes, for example the Harifian, Mushabian, and Geometric
Kebaran. Unfortunately there is little consensus as to how these complexes should be defined, which occur
in particular sites and levels, and what their regional and temporal distributions are like. The result is a
complicated web of terms, overlapping time ranges and spatial distributions. Adding to the complexity,
these industries are often regarded as the tangible material remains of identity-conscious social units of
According to Brian Byrd (1998) the primary cultural complexes for the Early and Middle
Epipaleolithic are the Kebaran (Early), the Geometric Kebaran (Middle) and the Mushabian (Middle).
However, many researchers, especially those working in Jordan, do not use the Western Levantine
4
classificatory scheme and have proposed a number of alternative complexes in order to accommodate
Jordanian data (Olszewski n.d.). Donald Henry uses Hamran, Qalkhan and Madamaghan to refer to
different microlithic industries in southern Jordan (1995). Except for the Late Epipaleolithic Natufian, there
is virtually no agreement on how the material should be divided. As a result there are nearly as many
classificatory schemata as there are specialists interested in the Levantine Epipaleolithic. There is also little
consensus about index type definitions. Although subject to much debate, it can be shown that this
typological approach and the use of culture to explain typological differences is deeply problematic (e.g.
Neeley 1997; Neeley & Barton 1994 and the discussion following their article; Goring Morris et al. 1996;
Barton and Neeley are not alone in this critique. For the Paleolithic in general , and throughout the
Old World, the basis for inferences about pattern in the remote past, however, has been derived from
French typological systematics (Clark 1997). It has been noted that these lithic typologies focus attention
on a very small percentage of lithic artifact assemblages (#5%), the retouched stone tools, and ignore
variations in the much larger debitage & core components. Bisson (2000) has made an argument that they
should be replaced by more specific, question-oriented means to record lithic variability. Further, the
tendency to explain retouched tool variations in cultural terms has effectively precluded consideration of
Explaining typological differences by invoking cultural differences has been challenged from a
number of different perspectives. The validity of tool types as emic and static has been questioned (Barton
1991; Dibble 1995). Indeed, lithic artifacts should be viewed as dynamic, engaged in a continuous process
of use, remodeling and reuse (Dibble 1987; Neeley and Barton 1994). As was demonstrated several
decades ago, the final state of an artifact at discard is potentially far removed from its original morphology
(Frison 1968; Jelinek 1976), supporting the argument that stone tool types cannot be understood in terms of
The initial success of typological systematics as the way to study stone tools has put a significant
burden on our potential for further development. Although the Bordes-Binford debate led to increased
scrutiny of typological systematics, this approach remains firmly entrenched in much research throughout
the world and across time. The American Archaeologist Steven Kuhn writes that, Bordess typological
5
method increasingly takes a back seat to other analytic approaches (Kuhn 1995: 15), but this is clearly not
the case in the Levant, where Bordesian systematics remain by far the most important frame of reference.
The above-mentioned Levantine Epipaleolithic complexes, although useful to some extent for
chronological purposes, should be handled with extreme care when it comes to their interpretation. It is by
no means clear what they mean, or represent, in behavioral terms. Much is gained by approaching lithic
variability by considering factors like group mobility, raw material constraints, land-use patterns, site
function, and factors of risk as related to technological choices made. These situational and contextual
factors play important roles in hunter-gatherer adaptations everywhere and strongly influence the
characteristics of lithic assemblages. Below I suggest some ways in which we can test for some of the
Differences in raw material availability and quality, and their effects on technological organization
as related to forager mobility have received increased attention in the literature (Turq 1992; Kuhn 1991;
1995; Dibble 1995; Fblot-Augustins 1993; Olszewski et al. 2000). The focus has been on elements we can
control for, such as the differential availability of stone resources and the spatial distribution and quality of
those resources. Goring Morris argument (1995: 143) that raw material constraints for chipped stone tools
are rarely of significance in the Levant, since sources are widely available throughout the region fails
to take into account differentials in quality and availability. Raw materials are not always readily available
and because the quality of the raw materials varies considerably, more attention should be paid to this issue
on a case-by-case basis (Henry 1989; Neeley 1997). Further, as Goring-Morris and others have pointed out,
different types of raw material are utilized differentially in different periods, even within the Levantine
This brings us to an important point: the necessity to differentiate between raw material
availability and use (Neeley 1997). Raw material availability refers to the types of raw material available in
a region, regardless of use. Raw material use specifically identifies those raw materials that are found on a
particular site or in a specific level of a site. Raw material use is thought to be governed by technological
6
constraints and choices made by mobile hunter-gatherers (as well as availability per se). As Kuhn points
out, we are not so much interested in raw material availability itself, but rather want to account for it so we
I emphasize here that models that identify raw material as a factor typically do so from the
perspective of raw material scarcity, rather than abundance. The notion that there might be differences in
the maintenance and recycling of stone tools (e.g. Goodyear 1989) and that these in turn depend on raw
material value implies added transport costs. In other words raw materials are sometimes valuable because
they were brought into the area from elsewhere. Along similar lines, Bamforth probably went a little too far
when he claimed that maintenance of stone tools would occur only if material was not readily available
(Bamforth 1986). The degree to which stone tools will be maintained or recycled is a function not only of
the regional distribution of raw material, but of the mobility characteristic of the settlement system of
which the artifacts are a part (Kelly 1988). When raw material is ubiquitous and sites close to one another,
one can assume that similar raw material sources will be used on those sites. In case the same resources are
not used, differential use of raw material in order to accomplish different goals can be one explanation for
observed variability (e.g., fine grained cryptocrystalline exotics versus coarse-grained local quartzites).
Another plausible explanation for differences in raw material use could be that highly mobile people are
bringing better quality raw material from elsewhere rather then using poorer quality local sources.
Therefore the first and most important step is to compare raw material types (Table 1) and their use. Groups
frequenting a region were probably well aware of raw material sources, but might have used them
differently. One way to evaluate this might be to check for evidence for differential raw material values,
Table 1: Variables to Compare Raw Materials Used (modified from Neeley 1997: 38).
High raw material value Low raw material value Test variables
Raw material types (color and texture)
Smaller cores Larger cores Core size
Formal cores More informal cores Core type
More retouched blanks to Fewer retouched blanks to Ratio of retouched blanks to
unretouched blanks unretouched blanks unretouched blanks
Less blades and flakes with cortex More blades and flakes with cortex Blades and flakes with cortex
Smaller-sized blanks and tools Larger-sized blanks and tools Tool and blank size
7
How hunter-gatherers get their raw material can be modeled in a number of ways. Binford
suggested that raw material procurement should be viewed as embedded in other activities, probably mostly
subsistence related (Binford 1979) and/or associated with predictable social activities (Gould and Saggers
1985). Kuhn distinguishes two alternative approaches for a group of hunter-gatherers supply of raw
material needs. One is to supply individuals with a portable, flexible toolkit useful under conditions of
uncertainty, the other is to supply places in the landscape where activities requiring stone are likely to be
conducted (Kuhn 1992; Kuhn 1995). These alternative strategies fit into a larger scheme relating them to
Further, procurement might be governed by different requirements for different needs. Depending on the
nature of anticipated activities, different constraints on tool or blank form might be important. An example
is preparing for the hunt compared with preparing to process cadavers from a kill. In the first case there is a
need to have the hunting gear ready to go before unpredictable encounters with mobile animals (Nelson
1991; Kuhn 1992; Bamforth 1985). In the second, such readiness in not as important, as the kill will not run
away while the necessary processing tools are made. These differences might reflect not only raw material
acquisition and use, but also the strategy by which each of these toolkits might be prepared. When good
raw material is readily available, hunting gear is expected to be highly portable whereas processing tools
should be manufactured using more expedient technologies. I stress that the situation sketched here
depends on raw material availability, not just constraints pertaining to the activity and its context. Therefore
getting a handle on availability has to be an important first step in any investigation pertaining to these
issues. At the same time, the example hints at the potentially complex interaction between raw material
8
Group Mobility
A second major factor influencing lithic variability is group mobility. As has been pointed out
repeatedly, mobility plays a large part in determining the organization of hunter gatherer lithic
technology (Kelly 1988; see also Nelson 1991; Kuhn 1995). Crucial to studies of group mobility are the
concepts of expedient and curated technologies (Neeley 1997). A curated technology is one in which
artifacts/tools (the raw material) are conserved and cared for (Nelson 1991). This can be manifested
through reshaping and resharpening, transport and advanced tool manufacture. By contrast, an expedient
technology minimizes production effort (Nelson 1991). Implements are made on the spot, used and
discarded after use. It is generally thought that more mobile groups emphasize curated technologies while
Table 2: Mobility Expectations and Test Variables (after a table in M. Al-Nahar's PhD proposal).
more sedentary groups tend to employ more expedient technologies. An example is the universally
documented change in lithic technology that coincided with the appearance and development of
horticulture (Parry & Kelly 1987). Documenting expedient and curated technologies, and how they vary,
with mobility, is made more reliable when raw material availability is either held constant, or accounted
for. Group mobility itself minimizes risk related to local depletion of critical resources. It is important to
recognize that characterizing an entire assemblage as either curated or expedient is probably a gross over
generalization, as technological subsystems are more or less curated depending on how that technological
subsystem functions in the context of the overall adaptation. As an example, Neolithic arrowpoint
technologies are typically highly curated, while those associated with plant processing usually have a more
9
expedient character. Mobility strategies can also vary with group size and season. These forms can be best
understood in terms of the forager/collector continuum (Binford 1980; Neeley 1997). Table 2 presents
expectations and test variables related to mobility patterns. It is important to stress that none of these
variables should be used in isolation, rather it is the convergence of several test implications that leads to
strong inference.
Site Function
A third factor of generally recognized importance to lithic variability is the functional aspect of
technology. There are three levels of analysis relating to function, one of which will be emphasized here.
These levels are (1) the function of individual artifacts; (2) the functional aspects of specific activity sets,
which might entail or require multiple toolkits, and finally (3) the function of sites within a larger
settlement subsistence system reflective of hunter-gatherer land-use. As individual artifact function and
the functional aspects of activity sets require microwear analysis and/or intrasite spatial analysis, the focus
here is on site function, ideally supplemented with information relating to the two other levels of analysis.
Binfords subsistence - settlement model contrasts two idealized alternative general strategies:
foraging and collecting (1979). These are best conceptualized as end points on a continuum of structural
poses characteristic of all hunter-gatherers. Foragers follow a strategy in which the entire group moves to
resource patches. Residential mobility is consequently high and special purpose sites are not expected. In a
collector strategy, logistical mobility is high while residential mobility is not. In the latter case, a variety of
different kinds of sites will result, (i.e. task groups are dispatched from residential bases) many of which
are functionally specialized, reflecting a limited number of activities and a reduced diversity in tool forms,
see Table 3. Differences are also expected between residential camps because in a foraging strategy
residential moves will be frequent, and will result in less spatially differentiated sites with comparable
ranges of artifact diversity. Collectors, on the other hand, will exhibit significant intrasite spatial
differentiation because the relatively prolonged stay in one place would create secondary refuse deposits as
well as spatial segregation of certain activities. Table 3 gives the variables and the corresponding test
implications expected for two site types residential bases and limited activity stations. It is crucial that
none of these variables be taken in isolation. Also, at this stage it is necessary to include information from
the raw material and mobility parts of the study, for these elements cannot be entirely separated. Indeed, the
10
intent of this study is to contribute to a better understanding of the undoubtedly complex interplay amongst
these conceptually separate aspects affecting lithic variability, requiring that they eventually be brought
The case study uses samples of controlled surface collections at two Early Epipaleolithic
rockshelter sites from the Karak Plateau, west-central Jordan. Both sites were investigated in the summer of
1999 as part of the Karak Resources Project directed by Gerald Mattingly. Part of this multidisciplinary
research team focused on an initial prehistoric survey directed toward investigating the prehistoric
occupation of the Karak Plateau. The survey located and sampled a total of 81 sites representing all major
prehistoric periods (Schurmans et al. in prep.) and constitutes, together with the prehistoric sites recorded
by the Limes Arabicus Project (Rollefson 1987; Koucky 1987) and the occasional prehistoric site recorded
by the Archaeological Survey of the Kerak Plateau (Miller 1991), the only record of prehistoric occupation
on the Karak Plateau of which we are aware. Among the sites recorded are two large, dense Epipaleolithic
scatters, each associated with a rockshelter. These sites, KPS 36/4 and KPS 75, are only some 3 km apart,
which allows for control of two important causes of lithic variability: raw material availability and local
environmental conditions. However, as argued earlier, we cannot simply assume similar raw material
availability without first closely examining this preposition. Raw material availability will be discussed
first. Next, both sites will be compared using a standard typological assessment and, finally, we will
11
examine the relationship amongst raw material use, group mobility, and site function and their potential
KPS
36/4
KPS
75
Figure 1: Satellite image of the Karak Plateau and the Wadi al-Hasa to the south. Dots represent
prehistoric sites from surveys in the area. Notice the graben scarring the southeast portion of
the plateau and the location of the Epipaleolithic sites studied (KPS 36/4 and KPS 75).
Surface scatters typically are considered of little scientific value because of the possibility of
chronological mixture. In xeric, deflated environments, like the Levant, they are often viewed as
considerable time intervals. However, these considerations also apply to buried deposits. Therefore, the use
of surface material should not be rejected a priori because it might be mixed, just as a buried deposit
should not be assumed to be unmixed. The realization that collections often represent a palimpsest of
12
occupations, especially at rockshelter sites (Barton and Clark 1993), is an important consideration that must
Both sites are located in the immediate vicinity of the Fajj al Usaykir, a northwest/southeast
trending graben scarring the southeast portion of the Karak Plateau (Fig. 1). A multitude of suitable fine-
grained cryptocrystalline rocks are present in the immediate vicinity of the sites that could have, and did,
serve as raw material sources for Epipaleolithic hunter-gatherers. Although no systematic survey for raw
materials was conducted during the Karak Prehistoric Survey (KPS), relevant information can be extracted
from the systematic raw material survey conducted by the Eastern Hasa Late Pleistocene Project (EHLPP)
immediately to the south, in the Wadi al-Hasa. Information gathered during the raw material survey
(Olszewski et al. 2000; Olszewski et al. n.d.) is used to assess raw material availability on the southeast
fringe of the Karak Plateau because the geological formations in this area of the plateau are the same as
those found in the Wadi al-Hasa. Information derived from the raw material survey in the Hasa can be
compared to the information on the geological map of the southeastern part of the Karak Plateau (Adir
Section, RGS, 1990), an accompanying publication of the Natural Resources Authority of the Hashemite
Kingdom of Jordan (Shawabkeh 1991), and observations made during the Karak Prehistoric Survey itself.
Three Upper Cretaceous formations blanket the study area; each contains usable raw material
(Olszewski et al. 2000). Formations from oldest to youngest are the Wadi Umm Ghudran Formation (WG),
the Amman Silicified Limestone Formation (ASL), and the Al Hisa Phosphorite Formation (AHP) (Fig. 2).
Each yields a wide variety of both matte and glossy gray to brown, brecciated cherts occurring in extensive
beds, and as nodules. The latter both in wadi beds and interspersed between the chert beds. Other
knappable raw material includes siliceous coquina, quartzite, and siliceous phosphorite. The WG formation
consists of sandstone, limestone, and quartzite beds, surfaces mainly in the Wadi al-Hasa and is of lesser
importance to this study. The ASL formation contains numerous chert and limestone beds and a variable
number of distinctive oyster shell coquina beds. The AHP formation comprises three members: the Sultani
Phosphorite (SP), the Bahiya Coquina (BC), and the Qatrana Phosphorite (QP). The BC is noteworthy as
rockshelters throughout the area, including the two investigated here, are formed at the contact between the
BC and the underlying softer deposits of the Qatrana Phosphorite. A fourth formation occurs more than 10
13
Figure 2: Stratigraphic section of the southern half of the Karak plateau (from Shawabkeh 1991).
14
km from the sites, the Muwaqqar Chalk-Marl Formation (MCM). This formation of Upper Maestrichtian to
Paleocene age contains predominantly marly beds and oilshale. Some phosphorite and phosphatic cherts
have been recovered from boreholes in the MCM Formation (Shawabkeh 1991). In addition to bedrock
formations, Pleistocene fluviatile and lacustrine gravels, alluvial fans, and mudflats mantle depressions
throughout the plateau (Shawabkeh 1991). Each yields usable raw material eroded from one or more of the
formations mentioned and could equally have served as sources for raw material. Finally, small basalt plugs
are found immediately to the north northwest of KRP 75 and 36/4, at some 6-7 km near the site of al-
Mudaybi. Basalt served as raw material for the groundstone found at KPS 75.
The raw materials found on the sites closely resemble those found in their immediate vicinity,
although it is extremely difficult to pinpoint exact sources because of the sheer diversity of raw materials
on any one location and the similarity of raw materials present at different locations. The approximate
distance from each site to each relevant geological unit is given in Table 4. Clearly each of the different
units is readily accessible and distances to them are very similar for each of the rockshelter sites. Hence we
can conclude that a similar range of raw materials was available to the occupants of both sites. Raw
material availability is different from raw material use. As an example there is a clear distinction in the raw
materials used at Middle Paleolithic sites in the area as opposed to the raw materials used during
Epipaleolithic times. Although it could be argued that raw materials were not always as visible/accessible
as they are now because of vegetation and soil cover, I argue that the majority of raw materials available
today were readily available in the past. Regardless, sites from the same prehistoric time period and close
proximity certainly should have the same raw materials available to them. Therefore, differential raw
15
material use is governed by technological choices made by hunter-gatherer groups. Such choices will be
The Kerak Prehistoric Survey collected surface samples from KPS 36/4 and KPS 75, and excavated a small
0.5 by 0.5m testpit at KPS 36/4. The surface collections consist of random grab samples and large
quantitative collections from grid squares. This study exclusively uses random samples of the quantitative
surface collections. Collections at KPS 36/4 were made using 5 by 5m and 2 by 2m grid squares. Only the
latter are used in this study because collecting all lithics visible on the surface from a 5 by 5m unit is
extremely difficult. Such large collection units tend to result in collections biased towards larger artifacts as
is demonstrated by collections of small pieces resulting from checks we did on previously collected 5 by
5m squares. Unfortunately such checks were not done formally for the smaller units. At KPS 75 we used
exactly the same collection strategy, collecting all lithics visible on the surface from grid cells while sitting
and/or laying down, but used 1by 1m units rather than the larger units at KPS 36/4. Because collection
units are smaller at KPS 75 and lithic scatters much denser at the same site it is possible that we missed
A small testpit excavated to a depth of 40cm in 10cm arbitrary units at KPS 36/4 was designed to
establish if subsurface deposits are present at KPS 36/4. All excavated sediment was screened resulting in
the retrieval of more and smaller pieces than those recovered from the surface deposits. This study
compares the general characteristics for both surface collections and the excavated collection treated as a
single assemblage. The main component of this study on raw material use, mobility and site function do not
incorporate the subsurface deposits, but focus instead on the comparison between the surface collections of
Random samples of all pieces collected for each site were selected for typing and of that sample a
random selection of three squares was made from each site. All artifacts from the latter sample except
shatter, trimming flakes and some flakes were measured and weighed. While each of these collections is
random it needs to be pointed out that the selections of grid squares collected at the sites were not selected
randomly. Rather at both sites we chose to collect grid cells along a trench running from the entrance of the
rockshelter downslope (mild slopes, estimated between 0-5 degrees, are present at both sites). At KPS 75
16
some additional squares were selected off this transect one of which ended up in the sample investigated
more closely.
Random samples of the controlled surface collections from KPS 36/4 (36 m2) and KPS 75 (13 m2),
and the assemblage from the testpit at KPS 36/4, are compared in Tables 5, 6, and 7. In Table 5 standard
debitage categories are compared. The table indicates that the assemblages are fairly similar although some
interesting differences show as well. Cores are common in both surface assemblages, but are rare in the
Table 5: Counts and Percentages for the Debitage Categories and Tools from KPS 36/4 and KPS 75 as well
as Adjusted Counts and Percentages for KPS 36/4.
Artifact type Testpit KPS 36/4 Surface KPS 36/4 Adjusted Surface 36/4 Surface KPS 75
n % n % n % n %
Core 7 0.5 50 2.1 50 1.4 85 1.3
Blade 51 3.6 155 6.5 155 4.4 655 10.1
Bladelet 244 17.4 222 9.3 444 12.5 1211 18.6
Flake 149 10.6 873 36.6 873 24.6 1619 24.9
Trimming flake (<2cm) 841 60 830 34.8 1660 46.8 2165 33.3
Retouched tool 54 3.9 131 5.5 183 5.2 364 5.6
Microburin 8 0.6 4 0.2 9 0.3 25 0.4
Rejuvenation piece 2 0.1 9 0.4 8 0.2 74 1.1
Shatter 45 3.2 109 4.6 164 4.6 296 4.6
Total 1401 100 2383 100 3546 100 6494 100
KPS 36/4 testpit. Microburins are present in all three assemblages. When we look at microburin
percentages relative to retouched tools, there are some differences as the surface deposit of KPS 36/4 has
3%, KPS 75 6% and the subsurface at KPS 36/4 13% microburins. Microburins are deemed important as
they appear early (c. 20 kyr BP) in sites east of the Dead Sea, and are used extensively in the preparation of
The biggest overall difference is that between the surface and subsurface collections. The
excavated assemblage from KPS 36/4 consists mainly of smaller artifacts; bladelets (elongated blanks with
a maximum width of 12 mm) and trimming flakes (flakes smaller than 2 cm in any direction) comprise
77.4% of the total. Most of the tools consist of non-geometric microliths (Table 6), increasing the number
of small artifacts to >80% of the assemblage. By contrast the surface material from KPS 36/4 has far fewer
small pieces. I argue that this difference is caused by two contributing factors: collection methods and
17
taphonomic agents. As mentioned earlier the excavated dirt was screened allowing us to retrieve much
more and smaller artifacts. Tahponomical agents thought to cause the size differences are twofold: (1)
possible sheet wash removing smaller artifacts from the surface while leaving the larger pieces and (2)
preferential downward migration of smaller artifacts in the soil. To account for this winnowing I adjusted
the counts and percentages by counting every small artifact twice. I do not suggest that by counting every
small piece twice we represent the original pattern in any sense, rather I suggest we can use these adjusted
percentages as a heuristic means to evaluate how the assemblage could have looked like if size based
So far as the comparison of the surface material is concerned, however, the percentages for the
major debitage categories are similar. One important difference is the incidence of bladelets, which is
(much) higher at KPS 75 than at KPS 36/4, even when adjusted for size. Although this pattern at first sight
could be explained by geomophological processes (e.g., removal by erosion of the smaller artifacts from
the surface at KPS 36/4), this hypothesis can be discounted when it is noted that trimming flakes comprise
nearly identical fractions for both surface deposits. Therefore, if winnowing took place, which it probably
did, still the difference between the surface deposits in terms of the trimming flake:bladelet ratio would
remain. Hence geomorphological processes do not account for observed differences. Blades are also more
numerous at KPS 75, and the resulting blades & bladelets: flakes & trimming flakes ratio is 1:4.5 at KPS
36/4 and 1:2 at KPS 75. The percentage of tools relative to the entire collection is consistent for each of the
assemblages.
When we look at the relative frequencies of the retouched pieces for each assemblage, some basic
differences become apparent (Tables 6, 7). Typologically, despite differences, all three assemblages can be
described as Early Epipaleolithic: non-geometric microliths constitute the largest tool class in each
assemblage, the microburin technique is used, endscrapers are common, burins moderate, and geometric
microliths
18
Table 6: Composition of Tools from Two Collections from KPS 36/4 and One from KPS 75.
virtually absent (Byrd 1988; Olszewski 2000). At a more detailed level, however, there are some significant
differences in the tool components of these assemblages. Except for a single burin and two scrapers, all
tools from the KPS 36/4 testpit are non-geometric microliths. This pattern is in sharp contrast to the
relatively moderate 39% non-geometric microliths from the KPS 36/4 surface collection, and the 65% non-
geometric microliths at KPS 75. Further, endscrapers are particularly common on the surface at KPS 36/4,
somewhat less common at KPS 75, and rare for the KPS 36/4 excavated sample.
When we look more closely at the non-geometric component, we can get a basic idea of the
temporal positioning of the assemblage. An Early Epipaleolithic (20 14.5 kyr BP) date seems
unquestionable. Narrow backed microliths are common (mainly backed, curved, pointed, and backed and
truncated pieces) (Table 7). All three assemblages resemble one another when it comes to the composition
of the non-geometric artifacts. As noted above, this is not necessarily true for the assemblages as a whole
19
Table 7: Composition of the Non-Geometric Tools from KPS 36/4 and KPS 75.
Even if we accept their existence, and overlooking problems with ambiguous conflicting
definitions, assigning the assemblages to particular Early Epipaleolithic industries and/or complex(es) is a
much more difficult task. For our purposes, it is probably not necessary to do this, and is almost certainly
premature as only limited excavation and surface data are presently available. Having described the main
debitage and retouched tool components, the next step is to investigate raw material use, mobility, and site
function as these might be reflected in the lithic assemblages and available contextual information.
As noted earlier, suitable raw material is ubiquitous in the area. It remains to be determined
whether the same raw material types were actually used and if patterns of raw material choice are reflected
in differential discard patterns at the study sites. In order to do so, samples from both surface collections
were typed using raw material categories described below. Short wave UV light and a simple method to
test for translucence were adopted to test the validity of raw material categories distinguished after
Distinguishing raw material types on the Karak Plateau as well as in the Wadi al-Hasa is not a
straightforward task. The problem stems from the fact that there are many varieties of raw material and
boundaries between chert types (e.g., phosphatic chert, brecciated chert, bedded, and nodular cherts) are
ambiguous and overlapping, reflecting the variability in local circumstances when chert is formed
(Klawiter 2000; Bush & Sieveking 1986; Knauth 1994). These problems led Neeley to distinguish no less
20
than 28 different types of raw material for his dissertation work on the Epipaleolithic site of Tor al-Tareeq
in the Wadi al-Hasa (Neeley 1997). Because of the possibility that Neeleys chert typology was overfine,
I decided to group raw material varieties in more general, broad categories. This seems warranted until
more precise geochemical characterization is done as it is clear that much variability can be found even
within a single chert bed. Categories used are given in Table 8 and are described below.
Brown glossy chert: fine-grained glossy dark brownish to light brown in color,
typically with a thin smooth orange cortex; often translucent, especially the
lighter brownish and reddish varieties; typically occurs in nodular form, both in
the ASL and especially the AHP Formations and their derivatives;
Brown matte chert: fine-grained (albeit somewhat coarser than the glossy
variety) dark to light brown/cream colored chert, typically with a thin smooth
orange cortex. This category includes fine-grained phosphatic cherts, typically
dark brown in color and with phosphatic inclusions; occurs as bedded and
nodular flint in the ASL and AHP Formations and their derivatives. It is difficult
to distinguish between these and non-phosphatic brown matte cherts;
Gray glossy chert: fine-grained, glossy, translucent chert, light to dark bluish
gray in color; cortex varies considerably, but dark reddish brown is common for
the nodular variety; occurs in nodular and bedded forms and, in the latter case, is
often attached to coarser raw material. Swirling patterns of gray are common.
Gray glossy chert is found in the AHP, but seems especially ubiquitous in the
ASL Formation;
Gray matte chert: fine to medium fine-grained, dull gray to bluish gray in color;
swirly and banded patterns are common; generally inferior in quality to the
above. Gray matte cherts occur in both the ASL and AHP Formations as both
nodular and bedded varieties;
Whitish chert: relatively fine-grained cherts, whitish to light gray in color and as
such often hard to distinguish from white patinated surfaces. This raw material
type is not as easily located in the landscape as the preceding four, however, it
should not be interpreted as introduced from elsewhere;
Siliceous coquina: medium coarse grained silicified coquina, with bluish gray,
often patinated surface. Siliceous coquina has many inclusions that are typically
very small linear bits of shell visible on close inspection without magnification.
Shell particles seem bounded by a translucent silicate gel. Siliceous coquina can
be found in association with the Bahiya Coquina (within AHP) or one of several
smaller coquina beds within the ASL Formation;
21
Other: This category accommodates a wide variety of cherts that occur in low
frequencies. Brecciated, and patinated cherts that could not be assigned to the
preceding categories are included. Very small bladelets and trimming flakes
account for much of this category as the pieces are sometimes too small for
adequate categorization;
To assess the validity of the chert categories a sample of artifacts was tested for translucence and
fluorescence. A simple test for translucence consists of holding the artifact up to a standard 75 Watt lamp to
see if the artifacts raw material allows light to pass through. Translucence is, of course, a question of
degree, but for our purposes it was decided to distinguish between translucent and not translucent in which
the outer 2 mm have to light up for it to be categorized as translucent. A second test determined whether
chert types fluoresce under short wave UV light (for an archaeological application using UV, see e.g.
Hofman et. al. 1991). A simple battery operated, handheld UV lamp offering short and long wave UV light
was used. Only short wave UV light was used when it became clear that very few artifacts fluoresced under
long wave UV light. The degree of fluorescence also varies dramatically. Two types of fluorescence were
observed, only one of which is used here. These are yellow, which is fairly distinctive, and a much less
distinctive deep purple. Some artifacts fluoresce strongly, while others get a yellowish shine. Further, many
artifacts fluoresced only in part, again underscoring the diversity of raw materials present on the plateau,
and variation in the composition of a single rock. I coded fluorescence as present or absent and
included artifacts that fluoresced lightly or only partially (> 25% of the surface) as present. Table 8
shows counts of translucent artifacts and artifacts that fluoresced yellow under short wave UV light by sites
and raw material categories as defined macroscopically. Percentages indicate the relative frequencies of
Table 8 shows that glossy chert varieties are almost always translucent. However, numerous matte
artifacts are translucent as well, precluding translucence to distinguish between matte and glossy varieties.
Relative frequencies of translucent chert used at both sites are similar. In contrast, the number of
fluorescent artifacts is much higher at KPS 36/4. Most of the fluorescent artifacts are categorized as gray,
particularly glossy gray chert. The table again reminds us of the diversity of raw materials present in the
local geological formations and the fact that a large number of them were utilized prehistorically.
22
Table 8: Counts of Translucent and Fluorescent Artifacts by Raw Material Type and Site.
Differences in fluorescence between the sites indicates that at least some different locales for raw material
extraction were used at each, although some were evidently frequented more at one site than the other.
Translucence and fluorescence characterizations could be used when a further breakdown of raw materials
is desired, and actual raw material sourcing to specific extraction locations is intended. However, the
following analysis will only use raw material types as they are identified on macroscopic grounds. This is
warranted as macroscopic characteristics are probably more closely related to elements evaluated by
Raw material comparisons of both surface collections are given in Table 9. Inspection of the table
shows that brown and gray varieties of primarily nodular chert make up the bulk of the raw material,
although there are some differences in raw material use at the two sites. At KPS 75, c. 60% of the
assemblage consists of brownish colored cherts, while this category only comprises 44% of the assemblage
at KPS 36/4. Raw material use at KPS 36/4 is more diverse than it is at KPS 75. Whitish cherts are very
rare at KPS 75 and siliceous cherts entirely absent. Overall, percentages of the more rare chert categories
23
Table 9: Counts and Percentages of Raw Materials Used at KPS 36/4 and KPS 75 (Surface Material Only).
Although Table 9 indicates how different macroscopically characterized raw materials were used,
it does not necessarily imply that these raw materials are sufficiently different to have an important impact
on the lithic variability observed at these sites. As it is clear from the description of some of the raw
material types (gray and brown varieties), the major distinctions are color and luster, rather than grain-size
(likely to have been more important in decisions about which raw material to use). To investigate whether
raw material use also has implications for lithic variability, the distributions of raw material categories by
artifact class are examined, along with metrical attributes considered important in raw material utilization.
This comparison was done using expected values obtained through Chi-square. Although the Chi-squared
measure itself is not useful especially given large sample sizes (Shennan 1988) I simply used the expected
values. 2 expected values are derived for each cell in the table by multiplying the corresponding row and
column marginal totals (for details see Shennan 1988). Values were computed using Keith Kintighs
Twoway program. Table 10 gives the observed minus expected values (O-E) by raw material groups and
artifact categories. The sum of all values in the table equals zero.
Table 10: Observed Minus Expected Values for all Raw Material Categories and Artifact Classes. Columns
with data on KPS 36/4 are gray, while those for KPS 75 are white.
Chisquared Brown glossy brown matte gray glossy gray matte whitish S. coquina other
(O-E) 36/4 75 36/4 75 36/4 75 36/4 75 36/4 75 36/4 75 36/4 75
Core 2.3 3.6 0.9 -3.1 -1.3 9.1 -2.9 -7.1 -0.4 0.3 0.2 0 1.2 -2.8
Blade -2.8 -0.7 4.6 4.8 -3.5 -4.1 3.3 1.6 -0.5 4 -1.9 0 0.8 -5.6
Bladelet -1.8 6.1 -6.8 6.9 7.4 6.3 0 -13.9 -1 -3.9 -2.4 0 4.7 -1.5
Flake -3.5 -0.1 -6 -2 -2.6 -5 3.3 6.5 -2.5 -1.2 7.1 0 4.3 1.8
Trimming flake 6.3 -6.8 1.5 -13.9 -5.2 -9.8 -2.1 19.8 10.2 1.1 -2 0 -8.7 9.6
Tool -0.4 -2.2 5.8 7.4 5.1 3.5 -1.5 -7 -5.8 -0.2 -1 0 -2.4 -1.5
24
A multitude of patterns can be observed from the table. This at two levels: (1) patterns within each
site, and (2) patterns between the two sites. At the site level (1) more blades than expected are made from
varieties of matte chert, this pattern holds for both sites. (2) More bladelets and tools than expected are
produced from glossy gray chert at KPS 36/4, while fewer bladelets and flakes than expected are made of
brownish chert. Tools made from brown matte chert are more numerous than expected. A somewhat
different pattern is found at KPS 75. Bladelets are more common than expected for both brown chert
varieties and for gray glossy chert. However, for the gray matte chert the observed value is 23% below the
expected value. The pattern for KPS 75 trimming flakes is exactly opposite the pattern for bladelets. Tools
at KPS 75 made on brown matte and gray glossy chert blanks show higher than expected frequencies. It
seems clear that differences in raw material use are certainly present, but these differences are not
consistent between sites. Interpreting these patterns individually is beyond the scope of this paper and
might prove difficult as each raw material type distinguished consists in turn of multiple raw material types.
However, interpretations of these types of patterns could yield insight into where different raw materials
and their reduction fit in the sequence of moving into a site, full stay, or near the end of occupation. For
example a group could have brought some worked raw material from elsewhere when first occupying the
site. When that material is dumped it is going to show different stages in the reduction sequence than
material that is being flaked in preparation for the next residential move. To do this kind of analysis
successfully I suggest MAN (minimum analytical nodule) (Odell 2000) divisions are necessary. A raw
material division much more fine-grained and labor intensive than the one used here.
Some simple comparisons between the artifacts from KPS 36/4 and KPS 75 allow for a more
detailed consideration of raw material differences and similarities between the two sites. One important
element here are the cores, more particularly their types and sizes (Table 11). In the table, the cores are
grouped into two categories, one consisting of formal core types (e.g., single platform & opposed
platform cores), the other comprising multiple platform and other informal core types. Inspection of the
table indicates that formal cores are more common at KPS 75 than they are at KPS 36/4. When expected
and observed values for this table are generated, there are more single and opposed platform cores at KPS
75, while the reverse is true for KPS 36/4 (i.e., less than expected single and opposed platform cores, more
than expected multiple platform and other cores). Core sizes are very similar at both sites (Table 12).
25
Table 11: Counts and Percentages for Combined Core Types at KPS 36/4 and KPS 75.
Table 12: Metrics for Length, Width, Thickness and Weight of Cores at KPS 36/4 and KPS 75.
The higher incidence of more formalized core types at KPS 75 hints that use of raw material is less
expedient. When comparing the weights of formal to informal cores it is clear the latter represent more
expedient rather than exhausted cores because the expedient cores are larger on average than the formal
cores (Table 13). If KPS 36/4 material is more expedient we expect cortical pieces at KPS 36/4 to be
Table 13: Weights in gram for Formal and more Expedient Cores at KPS 36/4 and KPS 75.
Cortex was broken down into three categories: (1) absent (<10%), (2) present (10-50%), and (3) abundant
(>50%). The ratio of absent:present:abundant is 10:4:1 at KPS 36/4 and 8:2:1 at KPS 75 indicating
that, contrary to expectation, cortical pieces are more abundant at the latter site. However, caution is urged
because the nodules used at KPS 75 are typically fairly small and often retain a cortical surface at discard.
If cortex tends to be more prevalent on cores at KPS 75, the above pattern would be difficult to interpret in
terms of extent of raw material reduction. Indeed, while core sizes are the same, the ratios of core cortex
26
classes are 2.5:4:1 at KPS 36/4 and 1.3:2.4:1 at KPS 75. This underscores Odells point that cortex
presence is a variable of doubtful interpretive significance (2000). Thus, cortex on cores is much more
common at KPS 75. The abundance of cortex has much to do with the shape of raw material used and the
reduction strategy. At KPS 75, cores are often made from flat nodular chert from which a large blade-like
primary flake is struck. This blow prepares the striking platform for the removal of bladelets along the
narrow edge of the core. When necessary, a core tablet is removed, often in the form of a blade,
rejuvenating the striking platform. Many of these rejuvenation pieces were recovered at KPS 75, while few
were recorded at KPS 36/4, again pointing to a more formal reduction process at KPS 75.
Finally, tool and blank sizes can be compared at both sites. The box and whisker plots below
clearly show the extreme similarity in artifact sizes between the two sites. Only for the length of the
artifacts does there appear to be a consistent pattern, with the artifacts at KPS 75 slightly longer than those
at KPS 36/4. This pattern of similarity also holds for the tools, here divided into microliths and non-
microlithic (nm) categories. However, when not so divided the tools at KPS 75 are smaller on average than
those at KPS 36/4, reflecting the higher percentage of microlithic tools on the surface of KPS 75. Small
artifacts seem to be underrepresented at KPS 36/4 possibly due to reasons described above. As a result it is
extremely difficult to evaluate the difference between microlithic tools at both sites.
To summarize the results of the raw material study, we can conclude that raw material types are
used differently at both sites. However, the overall pattern and metrics for both are very similar. Several
elements seem to point to a more formalized use of raw material at KPS 75 when compared with KPS 36/4.
27
100 70
90
60
80
70 50
60
LENGTH
40
WIDTH
50
30
40
30 20
20 SITE10 SITE
10 KPS 3
KPS 36/4
0 KPS075 KPS 7
de t re ke lith m de let re ke lith m
bla ele co fla o l- n bla de co fla cro l- n
bla d
m icr too b la
mi too
COMPARE COMPARE
50 80
70
40
60
THICKNESS
30 50
WEIGHT
40
20
30
20
10 SITE SITE
10
KPS 36/4 KPS 3
0 KPS075 KPS 7
de let re ke lith m
de let ke lith
bla de co fla cro l- n de fla l- n
m
la
mi too bla cro
b b la
mi too
COMPARE COMPARE
Figure 3: Boxplots for Length, Width, Thickness, and Weight for Various Artifact Classes by Site with
KPS 36/4 on the Left and KPS 75 on the Right.
28
Mobility
Raw material diversity is greater at KPS 36/4 and also more evenly distributed there than at KPS
75 (Table 9). According to the test implications generated in Table 2, that might indicate higher mobility at
KPS 36/4, although this is counterintuitive given the more formalized nature of the artifacts at KPS 75. An
argument could be made that when raw material is ubiquitous less effort will be spent when any material
will do the job, while more formalized technological organization might require specific kinds of raw
materials. There seems to be some evidence for this at KPS 75, where bladelets are more common than
expected for three raw material types and where a single type, brown matte chert, constitutes 43% of all
bladelets. At KPS 36/4 bladelets are distributed much more evenly across raw material types. I argue we
should re-evaluate the validity of expected raw material diversity patterning shown in Table 2 in areas with
ubiquitous raw material since the pattern might be the exact opposite of expectations in areas with scarce
raw materials. In the latter case materials brought from elsewhere play an important role and as such cause
The incidence of cortical pieces also must be evaluated with caution, as just noted. There are more
cortical pieces at KPS 36/4 than at KPS 75, but this is probably related to raw material type more than to
intensity of reduction, as would be expected when raw material is curated. That it is related to raw material
type seems clear from the incidence of cortex on cores, which is higher at KPS 75 even though the core
sizes for both sites are very similar. The incidence of cortex is a relatively crude and unreliable measure,
however (Odell 2000). Further it needs to be stressed that the differences between KPS 36/4 and KPS 75
are relatively minor and situated not on a scale from expedient to curated rather than one from expedient to
more formalized technologies. The importance of this distinction is important in the evaluation of areas
where raw material is ubiquitous and as such renders evaluating group mobility extremely difficult.
However, I argue we can say something about the importance of mobile versus more sedentary activities.
The incidence of retouched pieces can also be investigated. Relatively more retouched pieces are
expected at sites where higher mobility and a more curated technology are the norm. Tool:blank ratios are
the same for both sites. Most of the tools at KPS 75 are microliths, in contrast with the pattern at KPS 36/4
where most of the tools are large. Although the ratio of retouched tools to blanks could shed light on
relative mobility, such a measure can be difficult to interpret. There are large and small retouched tools at
29
both sites and it is the relation between these two that is particularly informative, not so much the incidence
of retouched tools relative to debitage per se. As argued earlier there is a distinction between mobile and
sedentary activities and it is the relative importance of each at a particular locale, adjusted by the rate of
discard, that should determine the relationship between the relative incidence of the large and small
components of the retouched assemblage. As small artifacts at KPS 36/4 are probably underrepresented we
cannot derive any definitive statements from the tools themselves. Further, it should also be kept in mind
that the retouched tools only identify a portion of the actual tools, since many were probably not retouched.
core
250
200
150
WEIGHT
100
50
0
KPS 36/4 KPS 75
SITE
Figure 4: Core weight compared for KPS 36/4 and KPS 75.
Although the tools themselves are smaller at KPS 75, this is not necessarily true for the blanks. However,
when we look at all artifacts and compare sizes it seems clear that those at KPS 75 are smaller overall than
those at KPS 36/4. In terms of core weights, both assemblages are fairly similar (KPS 36/4 has slightly
Ratios of tools to cores are 2.6:1 for KPS 36/4 and 4.3:1 at KPS 75, while the ratio of cores to
blanks is 1:25 at KPS 36/4 and 1:9.6 at KPS 75. Again these measures conflict with the expectations
derived from Table 2. While the tool to core ratio should indicate a more mobile/curated technology at KPS
75, the core to blank ratio indicates less intense core reduction at KPS 75, which in turn was linked to a
more sedentary/expedient technology. Again, the problem might be situated at the level of raw material
availability. In contrast to a situation of raw material scarcity, when raw material is ubiquitous, as it is on
the Karak Plateau, there is no need to further reduce (i.e., maximally utilize) cores when a more formalized
technology is followed. How the contrast between core:blank ratios can be interpreted is not clear, but it
30
could vary by the type of raw material used. At KPS 36/4, a lower core:blank ratio is typical of the
brownish chert varieties, while the others have a higher ratio. At KPS 75 brownish cherts are relatively
Finally, artifact density and site size are other potentially valuable indicators, but these are
extremely hard to evaluate for surface material. At KPS 36/4 surface density (66 artifacts per m2 value
not adjusted) is much lower than at KPS 75 (500 artifacts per m2). Estimated areas for these sites are
>10,000 m2 for KPS 36/4 and slightly over 1,000 m2 for KPS 75. However, estimating site area is very
difficult, especially at KPS 36/4, because of the extensive, but uneven distribution of artifacts. Repeated use
of a single area also confounds easy interpretation of site features and structures. At KPS 36/4, two bedrock
mortars are located near the entrance to the rockshelter, but it is currently impossible to assess whether
these are associated with the lithic scatter. Other features in the immediate vicinity, but likely not related to
the scatter, are 3 stone circles probably pertaining to the Chalcolithic or Bronze Age period, rock art, and a
small stone structure on top of the bedrock outcrop in which the rockshelter is found. The latter two date to
the post Roman period. At KPS 75 three bedrock mortars are situated on top of the bedrock outcrop
containing the shelter, but again, there is no way to ascertain whether they are associated with the lithic
scatter. KPS 75 also has some basalt groundstone scattered over the site surface. While these could be
associated with the site, it seems more prudent to await excavation to verify this. These elements are
particularly important as they underscore the danger of possible intermixture from post-Epipaleolithic
occupations. However, there is no evidence of non-Epipaleolithic lithics. The exceptions are single
Neolithic points at both sites. All in all these are very minor intrusions and the fact that they are made of
exotic raw material further indicates that the points probably represent ephemeral use of the area by passing
groups.
Site Function
A third major element influencing lithic variability is site function. It is argued that both sites
represent residential bases where extensive tool manufacturing and other activities took place. Both scatters
can be characterized as large, even though KPS 36/4 is ten times the size of KPS 75. Both also have a
wide range of tools and artifacts representing all major artifact classes. That activities took place at
31
particular places within each is indicated by the aggregation of retouched pieces at particular locales. For
example, one square at KPS 36/4 has 28% of all the tools from nine squares, while another at KPS 75
yielded 23% of all the tools from 13 squares. That the tool counts are highly spatially variable is evidenced
by high coefficient of variation values (=sigma/mean) for tool counts across all squares: KPS 36/4, CV=
68.5; KPS 75, CV= 68.6. However, assuming that we can reconstruct site activities from these patterns is
not warranted. That some aggregation is present seems acceptable, but interpreting the aggregation is
simply impossible as these sites have been on the surface for thousands of years and the current
arrangement of lithics at the surface is probably far removed from the original situation. Further, we are
probably dealing with a palimpsest of occupation events. To evaluate whether core reduction took place at
the site we can refer to the indices of artifact density and core:artifact ratios. All stages of reduction are
present at each site, although the nature of that reduction is very different. At KPS 36/4 cores were not
extensively prepared and rejuvenated during use while they were at KPS 75 (see rejuvenation pieces in
Table 5). Further KPS 75 has several groundstone artifacts and there are bedrock mortars on both sites.
These elements seem to indicate that both sites had similar functions on at least a very coarse analytical
scale.
Discussion
In what follows, I will (1) address differences between the surface and subsurface samples at KPS
36/4, (2) discuss general issues with the use of surface collections, (3) offer a general interpretation of the
patterning at each site, and (4) examine some potential problems with the expectations suggested at the
The differences between the surface and excavated samples at KPS 36/4 are striking. The vast
majority of the excavated material is very small, but fits typologically in the Early Epipaleolithic. By
contrast, the surface sample has a more balanced distribution between larger and smaller elements. Several
hypotheses could account for this pattern. (1) The subsurface deposit represents part of a much more
extensive deposit that was deflated to the current surface, mixing lithics from several occupations on the
surface, most of which were Early Epipaleolithic. This hypothesis seems unlikely, as there is no significant
admixture of fossiles directeurs from other time periods. (2) The surface material consists of a different
industry. This is a possibility even though the material is very similar in respect of microlith typology, the
32
defining element for the Levantine Epipaleolithic. (3) The subsurface material comprises (smaller) artifacts
that have been displaced downward from the current surface. Although there has been much debate about
artifact movement in soils and sediments, it is not clear whether size-sorting is correlated with downward
movement, nor whether it is possible to generalize about how substantial downward displacement might be
(Barton 1987; Amour-Chelu 1994). I believe that further in-depth analysis on this issue is absolutely
necessary. (4) The subsurface assemblage represents the original cultural accumulation and small artefacts
have been differentially weathered from the surface assemblage (Barton personal communication; Dibble
et. al. 1997). As larger pieces are nearly absent from the subsurface it seems unlikely the surface and
subsurface represent the same original assemblage closely resembling the composition of the subsurface
material. This would presume that the surface material only represents a very small fraction of the original
assemblage, as there are only ca. 15% larger artifacts subsurface versus 44% larger artifacts on the surface.
The use of surface material in lithic studies has often been criticized because of problems
associated with context. Earlier I alluded to the potential for deciphering the information contained in
palimpsests and countered the argument that they were analytically useless with the suggestion that
subsurface deposits could very well suffer from the same problems (Bernabeu Auban et. al. in press).
Indeed, there is an urgent need for in-depth post-depositional analysis and geomorphological study at any
(prehistoric) excavation. However, this does not solve the problem at hand. Do the surface deposits
represent multiple occupations and if so, does that compromise this study? I am convinced that the answer
to the first part of the question is yes; however this does not invalidate the research. The assumption that
we are dealing with single occupation events is probably invalid at most prehistoric sites. At both
rockshelters investigated here, there is good evidence to demonstrate repeated use over time. At KPS 75,
for example, portions of the bedrock some 50 m from the site were recently tested to evaluate possible
quarrying, and part of the deposits in the rockshelter itself were dug out by Bedouins. Landscapes and
particular areas within them are used over and over again, and we should expect disturbances and multiple
occupations, especially at focal points in the landscape (Barton and Clark 1993). Two Neolithic points
indicate later visits at both sites from people who could have contributed to the lithic assemblage. However,
the absence of other fossiles directeurs suggests that most of the repeated visits occurred during the Early
Epipaleolithic. These sites were likely part of an annual round, or were used sporadically at intervals over a
33
long period of time. They might have been way stations in a cycle of seasonal transhumance involving
movement between the Hasa floodplain and lowlands, and the plateau itself.
Controlling for raw material availability and environmental factors we learned several things about
raw material use, mobility patterns and site function at KPS 36/4 and KPS 75. At KPS 36/4 there is a wide
array of different raw material types and evidence for a more expedient technology. Tool sizes are larger at
KPS 36/4 than at KPS 75, indicating that portability was probably not a primary concern for its occupants.
However, this finding does not imply that all of the technology was more expedient. As the high incidence
of narrow backed bladelets showed, part of the activities there were fully geared towards the constraints
imposed by high mobility. When the site was vacated, the heavier artifacts could be left behind and
manufactured when needed at a new campsite from local raw material. When a scatter in areas of high raw
material availability points to more mobility, this does not necessarily imply residential mobility. It might
indicate a limited activity station connected with a range of maintenance and refitting activities probably
connected with hunting. Where raw material is scarce however, this clearly does not apply because raw
material for all kinds of activities must be hauled around from place to place.
In contrast with KPS 36/4, KPS 75 shows more evidence of a formalized technology. Cores are
more standardized in a technological sense. Single platform bladelet cores make up the majority.
Microliths, particularly narrow backed, backed and truncated, pointed, and curved microliths, make up
most of the tool types. Using the logic outlined above, this suggests that preparing for mobile activities
constituted a major emphasis here. As a consequence a more restricted range of raw materials were selected
for use in the production of blanks. One interesting result of the study is that the bladelets, described as the
major Epipaleolithic technology, do not in fact constitute most of the blanks on either site. Indeed, flakes
are more common. Although both sites probably housed the same major groups of activities and both are
interpreted as residential bases, specific kinds of activities varied between them. Again, more mobile
While the preceding interpretation is consistent with most of the patterns observed, it is not
consistent with all of them. In fact, a large number of elements pointed to the exact opposite pattern where
KPS 36/4 would be the site with high mobility. Two measures that should indicate high mobility were
more raw material types, and fewer cortical pieces (Table 3). Both point to KPS 36/4 as the site with the
34
more mobile foragers. As I argued above, each of these criteria might have to be re-evaluated. I will use
First, the notion that fewer cortical pieces should be correlated with more mobility, and with more
curated technologies arises from the idea that available raw material is used more extensively. As a result
more pieces without cortex will be produced. As noted, the problem with this assumption is that it does not
take the morphology of available raw material into account and it assumes the need to further reduce cores.
Second, the expectation to see a more diverse set of raw materials associated with highly mobile
foragers (Table 2) might be incorrect in the case of ubiquitous raw material. When raw material is scarce,
overall more mobile groups probably carry and discard raw materials from a variety of sources. When raw
material is abundant, however, the opposite pattern seems more intuitively likely. Because a more portable
toolkit goes hand in hand with a more formalized technology, more stringent raw material selection is
expected. This seems to be the pattern at KPS 75, where most of the bladelets are made of high quality
fine-grained brownish matte chert (43 %). These two examples illustrate (1) that the factors affecting lithic
variability are closely and complexly interrelated, and (2) that expectations about pattern under conditions
of high mobility are biased in favor of conditions of raw material scarcity. We need to develop alternative
sets of test implications for high mobility foragers under conditions of raw material abundance. This paper
Conclusion
It is clear that the traditional typological systematics have many limitations. Not only do they focus on a
very small percentage of the artifacts at a site (usually <5%), but also the behavioral meaning of the
resulting units is equivocal (e.g. Clark & Lindly 1990). By assuming that all patterned variation is due to
culture (ways of making stone tools leaned in a social context), other important sources of variation are
ignored. This study shows that general contextual factors with which all hunter-gatherers must come to
terms (e.g., mobility, where to find raw material) play significant roles in the composition of lithic
assemblages. Evaluating raw material availability and use, mobility, and site function at two Early
Epipaleolithic sites indicated more evidence for expedient technologies at KPS 36/4, whereas KPS 75 is
35
characterized by more formalized technologies. These patterns were in turn linked to activity mobility. It
was argued activities involving high mobility were emphasized at KPS 75, while more sedentary activities
probably constituted the main activity set at KPS 36/4. A conventional typological analysis of each
assemblage showed only minor differences between the two sites. The study showed that a re-evaluation of
material expectations for mobility patterns should start from the all-important observation of raw material
availability. It was argued that many of the conventional views on mobility and reduction intensity proceed
from an initial assumption of raw material scarcity. However, when suitable raw material is ubiquitous,
very different constraints on the organization of technology come into play. More explicit expectations
should be about relative mobility developed to take situations where raw material is abundant into account.
Further, it is far less straightforward to derive measures of hunter-gatherer group mobility in the case of
raw material abundance. Rather, we can try to get at the relative importance of mobile activities versus
more stationary activities. Furthermore the difference is not so much evidenced as expedient versus curated
assemblages, but rather expedient versus more formalized assemblages, irrespective of intensity of
reduction.
36
Bibliography
37
in: A. Garrard and H.G. Gebel (eds) The Prehistory of Jordan (BAR International Series, 396, 1),
Oxford: 209-285.
-Coinman, N. (1990) Rethinking the Levantine Upper Paleolithic, PhD dissertation, Department of
Anthropology, Arizona State University.
-Coinman, N. (in press) The Upper Paleolithic of the Wadi Al-Hasa, in: N. Coinman (ed.) The Archaeology
of the Wadi al-Hasa, West-Central Jordan, Volume 2: Excavations at Middle, Upper, and
Epipaleolithic sites (Arizona Research Papers, 52), Tempe, Arizona.
-Dibble, H. (1987) The Interpretation of Middle Paleolithic Scraper Morphology, in: American Antiquity,
52: 109-117.
-Dibble, H. (1995) Middle Paleolithic Scraper Reduction: Background, Clarification, and Review of the
Evidence to Date, in: Journal of Archaeological Method and Theory, 2, 4: 299-368.
-Dibble, H., P. Chase, S. McPherron & A. Tuffreau (1997) Testing the Reality of a Living Floor with
Archaeological Data. In: American Antiquity, 62: 629-651.
-Fblot-Augustins, J. (1993) Mobility strategies in the late Middle Paleolithic of Central Europe and
Western Europe: elements of stability and variability, in: Journal of Anthropological Archaeology,
12: 211-265.
-Frison, G. (1968) A functional analysis of certain chipped stone tools, in: American Antiquity , 33, 2:149-
155.
-Goring-Morris, N. (1995) Complex Hunter/Gatherers at the End of the Paleolithic (20,000-10,000 BP). In:
T. Levy (ed.) The Archaeology of Society in the Holy Land, London, Leicester University Press:
141-168.
-Goring-Morris, N., D. Henry, J. Phillips, G. Clark, M. Barton & M. Neeley (1996) Pattern in the
Epipaleolithic of the Levant: Debate after Neeley & Barton, in: Antiquity, 70: 130-147.
-Goodyear, A.C. (1989) A Hypothesis for the Use of Cryptocrystalline Raw Materials Among Paleoindian
Groups of North America, in: C. Ellis & J.C. Lothrop (eds.) Eastern Paleoindian Lithic Resource
Use. Boulder, Westview Press: 1-9.
-Gould, R. & S. Saggers (1985) Lithic Procurement in Central Australia: a Closer Look at Binfords Idea of
Embeddedness in Archaeology. In: American Antiquity, 50: 117-136.
-Henry, D. (1989) Correlations between reduction strategies and Settlement Patterns. In: D. Henry & G.
Odell (eds.) Alternative Approaches to Lithic Analysis, Archeological Papers of the American
Anthropological Association, 1, Washington DC: 139-155.
-Henry, D. (ed)(1995) Prehistoric Cultural Ecology and Evolution: Insights from Southern Jordan, Plenum
Publishing Corporation, New York.
-Hofman, J., L. Todd & M. Collins (1991) Identification of Central Texas Edwards Chert at the Folsom and
Lindenmeier Sites. In: Plains Anthropologist, 36, 137: 297-309.
-Hours, F. (1974) Remarques sur l'utilisation de listes-types pour l'etude du Paleolithique Superieur et de
l'Epipaleolithique de Levant, in: Palorient, 2,1: 3-18.
-Jelinek, A.J. (1976) Form, function, and style in lithic analysis, in: C.E. Cleland (ed.) Cultural change and
continuity, New York, Academic Press: 19-35.
-Kelly, R. (1988) The Three Sides of a Biface. In: American Antiquity, 53: 717-734.
-Klawiter, B. (2000) An Investigation into the Potential for Geochemical / Geoarchaeological Provenance
of Prairie du Chien Cherts. M.S. Thesis, University of Minnesota
-Knauth, L. (1994) Petrogenesis of Chert. In: Silica: Physical Behavior, Geochemistry and Materials
Applications. Mineralogical Society of America: 233-260.
-Kuhn, S. (1991) Unpacking Reduction: Lithic Raw Material Economy in the Mousterian of West-Central
Italy. In: Journal of Anthropological Archaeology, 10: 76-106.
-Kuhn, S. (1992) Blank Form and Reduction as Determinants of Mousterian Scraper Morphology. In:
American Antiquity, 57: 115-128.
-Kuhn, S. (1995) Mousterian Lithic Technology. An Ecological Perspective, Princeton, Princeton
University Press.
-Miller, J. (ed)(1991) Archaeological Survey of the Kerak Plateau (American Schools of Oriental Research
Archaeological Reports), Atlanta, Scholars Press.
-Neeley, M. (1997) Assigning Meaning to Lithic Variability in the Epipaleolithic of the Southern Levant,
PhD dissertation, Department of Anthropology, Arizona State University.
38
-Neeley, M., G. Clark, J. Schuldenrein, & J. Peterson (1998) Tor al-Tareeq (WHS 1065): An Epipaleolithic
Site in its Regional Context, in: Studies in the History of Archaeology of Jordan VI, Amman, The
Department of Antiquities of Jordan.
-Neeley, M.P. & C.M. Barton (1994) A new approach to interpreting late-Pleistocene microlith industries
in southwest Asia, in: Antiquity, 68, 259: 275-288.
-Nelson, M. C. (1991) The study of technological organization, in: M.B. Schiffer (ed) Archaeological
Method and Theory, 3: 57-100.
-Odell, G. (2000) Stone Tool Research at the End of the Millennium: Procurement and Technology. In:
Journal of Archaeological Research, 8: 269-331
-Olszewski, D. (2000) The Epipaleolithic in the Wadi al-Hasa: An Overview, in: N. Coinman (ed.) The
Archaoeology of the Wadi Al-Hasa, West-central Jordan, Volume 2. Excavations at Middle,
Upper and Epipaleolithic Sites (Anthropological Research Papers, 52), Tempe: 227-244.
-Olszewski, D. (in press) The Paleolithic of Jordan.
-Olszewski, D.; N. Coinman; T. Clausen, J. Cooper, H. Jansson; M. al-Nahar; L. Saele; A. Sampson; U.
Schurmans & J. Thompson (in press) The Eastern Al-Hasa Late Pleistocene Project. Preliminary
Report on the 2000 Season. In: Annual of the Department of Antiquities of Jordan.
-Olszewski, D.; U. Schurmans, H. Jansson, J. Cooper & M. al-Nahar (2000) Chert Raw Material Survey in
the Wadi al-Hasa, Jordan: Preliminary Findings. In: Neo-Lithics, 2: 16-18.
-Parry, W.J. & R.L. Kelly (1987). Expedient core technology and sedentism, in: J.K. Johnson & C.A.
Marrow (eds.) The organization of core technology, Boulder, Westview Press: 284-304.
-Rollefson, G. (1987) Chipped Stone Artifacts from the Limes Arabicus Surveys, in T. Parker (ed) The
Roman Frontier in Central Jordan. Interim Report on the Limes Arabicus Project, 1980-1985
(BAR International Series 340, ii) Oxford: 757-792.
-Schurmans, U., G.A. Clark, C. Bartlett, A. Tarawneh (in prep) The Karak Prehistoric Survey, West-central
Jordan.
-Sellars, J.R. (1998) The Natufian of Jordan, in: D.O. Henry (ed) The Prehistoric Archaeology of Jordan
(BAR International Series, 705), Oxford: 83-101.
-Shawabkeh, K. (1991) The Geology of the Adir Area. Map Sheet No. 3252 II. Geology Directorate,
Geological Mapping Division, Amman.
-Shennan, S. (1988) Quantifying Archaeology. University Press, Edinburgh.
-Turq, A. (1992) Raw Material and Technological Studies of the Quina Mousterian in Perigord, in H.
Dibble & P. Mellars (eds) The Middle Paleolithic: Adaptation, Behavior and Variability
(University Museum Monograph 78) Philadelphia, University of Pennsylvania: 75-85.
-ADIR map, Jordan (1990) Map Sheet No. 3252 II. The Hashemite Kingdom of Jordan Ministry of Energy
and Mineral Resources. Natural Resources Authority, Amman.
39