University of Washington

1994 Fayum Project

NARCE Report

by

Anthony Cagle

Field Director

This report summarizes research conducted in the Fayum region of Egypt between April and July of 1994 as an ARCE Fellow.

1.0 Research goals.

The purpose of this research was to study certain aspects of artifact variability during the so-called "Epipaleolithic-Neolithic transition" in the Fayum region of Egypt (Figure 1). The Fayum Epipaleolithic or 'Qarunian' (also known as 'Fayum B' [Caton-Thompson and Gardner 1934]) has been dated securely from 8220 +/- 105 BP (Site FS-2; Wenke et al. 1983) to 7140 +/- 120 BP (Site E29G1 Area E; Wendorf and Schild 1976). Qarunian sites appear to reflect small groups of hunter-gatherers and contain no ground stone or pottery (Wenke et al. 1988; Wendorf et al. 1984). Extensive examination of plant remains indicates little or no use of cereal grains (Wenke et al. 1988:39) and faunal analyses indicate that the Qarunian subsistence base was centered around lacustrine resources, particularly fishing (Brewer 1986). The lithic assemblages, which compose the bulk of the artifacts recovered from these sites, represent bladelet technologies, with backed bladelets as the main components, numerous notches and denticulates, and a few geometric microliths.

The succeeding Neolithic (the 'Fayum A' of Caton-Thompson 1926) is separated from the Qarunian by an apparent abandonment of the Fayum, corresponding to low lake levels and general aridification (Hassan 1986a). Neolithic settlements are larger, indicating long-term, perhaps seasonal, occupations, though without any evidence of substantial architecture. Numerous grain silos with domesticated cereals (Caton-Thompson and Gardner 1934), grinding stones, and sickle blades indicate a substantial reliance on domesticated cereals. Faunal evidence (Caton-Thompson and Gardner 1934; Gautier 1976; Brewer 1986), while showing the presence of various domesticated forms (sheep, goats, and cattle), also indicates substantial use of fish, particularly Nile catfish (Clarias). Thus, there seems to be some degree of continuity in subsistence practices with the Qarunian.

The Neolithic tool kit represents a characteristic Neolithic bifacial flake industry, with the addition of worked tabular slabs (Holmes 1989). Typologically, the Neolithic assemblages are very similar to the Nabta-Bir Kiseiba assemblages (Wenke et al. 1988), and also to the Merimden Phase II-V lithics (Holmes 1989). Since these sites were all occupied roughly contemporaneously, the similarity in artifact styles raises the possibility of common cultural origins and perhaps similar origins for domesticated species and/or agricultural practices.

The problem of explaining the origin of the Neolithic -- whether it derives directly from the Epipaleolithic or is intrusive and unrelated to earlier cultural manifestations -- has continued to perplex scholars for most of the past century (Jomard 1918; Caton-Thompson 1926; Sandford and Arkell 1929; Wendorf and Schild 1976; Hassan 1986b; Wenke et al. 1988). Lithic artifacts (stone tools) are the only artifact category that appear in both Epipaleolithic and Neolithic sites, and therefore are central to resolving this question. Traditionally however, typologies have been applied to the small portion of lithic artifacts that are retouched to form specific shapes, i.e., the "tool" portion of the assemblages. Differences between assemblages described by such typologies are then used to asses the degree of cultural similarity between different areas and time periods.

While these typologies are adequate for establishing broad regional and temporal trends in tool form, they do not adequately address the source of change in stone tool industries. That is, they do not enable one to ascertain whether change results from new people moving into an area, from a new technology being developed to create the tools, from new and different functions for which the tools were used, or some combination of these. Furthermore, a vast amount of material, the debitage, or waste material from stone tool production, has been virtually ignored in most studies involving Epipaleolithic and Neolithic stone tools.

Recent research on lithic artifact variability has indicated that tool form is controlled by a number of factors including the quality and availability of raw materials (Bamforth 1986; Beck and Jones 1990) and the settlement/subsistence patterns and social structure of the people utilizing them (Teltser 1988; Wiant and Hassen 1984). Parry and Kelly (1987), for example, have noted a correspondence between the adoption of sedentism and 'expedient' (simple flake and core tools) lithic technologies in wide areas of North and Mesoamerica. McDonald (1991) has noted a similar pattern in the later Paleolithic of Dakhleh Oasis in Egypt. In order to explain changes in lithic variability then, one must describe these various sources of variation in detail to determine which factor(s) are controlling the observed change.

With this in mind, the methodological goals of this research involved that portion of lithic variability determined by raw material quality and availability and technology. More specifically, I am attempting to establish which portion of the variability in stone tool form that results from the acquisition of different raw materials and the technology used to reduce this raw material to usable tools. It is essential that this variability be accounted for and controlled for before further investigation of possible cultural and functional factors be addressed.

First, I sought to establish the specific geological deposits from which the raw materials were obtained. Since the same technology may produce different forms when applied to different raw materials, it is essential to control for variation caused by differential raw material use. The distance to these raw material sources can also affect the technology used to produce these tools: distant material may be used much more efficiently and completely than locally available material. Second, since finished tools exhibit only the final stages of manufacture, I endeavored to gain technological information through an analysis of the debitage which, with some qualifications, preserves evidence of the entire production (or more appropriately, reduction) sequence. By addressing both of these aspects of tool manufacture simultaneously, I hoped to be able to obtain measures of the variation in technology before and after the Epipaleolithic-Neolithic transition.

2.0 Data collection.

The fieldwork for this project began in the first part of May 1994 and continued into the first part of July 1994. My work concentrated in the southwestern corner of Birket Qarun, near the village of Qasr Qarun. In this area, Dr. R.J. Wenke of the University of Washington conducted extensive surveys and excavations of Epipaleolithic and Neolithic sites in 1981 (Wenke et al. 1983, Wenke et al. 1988). Wenke's work revealed extensive remains of Epipaleolithic and Neolithic settlements along either side of a fossil beach ridge, a prominent feature of the area (see Figure 1). The Neolithic sites, designated FS-1, are found on the desert side of this ridge; the Epipaleolithic sites, designated FS-2, are found on the lake side of this ridge. All of these remains cover extensive areas of what is now desert. Because of their relative isolation, they have been largely ignored by looters and early archaeologists, and thus retain a great deal of their material. According to Wenke's initial analyses, the two sets of remains can be assumed to be independent of one another; that is, no Epipaleolithic material is found on the Neolithic side of the ridge, and vice versa.

For both the archaeological and geological samples, I made use of the Global Positioning System for locational purposes. This system uses satellites to determine the locations of the user's latitude and longitude to an accuracy of within about 25 meters. Since many of these collections described here were made in remote areas where standard surveying methods are impractical, this was the only viable system with which I could accurately determine the location of my collections.

2.1 Debitage collection.

First I located several large areas of archaeological deposits on either side of the ridge that contained abundant remains of stone tool production. I laid out a total of twelve 2-meter squares, each oriented N-S, six on the FS-2 (Epipaleolithic) side, and six on the FS-1 (Neolithic) side. All surface material from each square, including rocks, bones, and other debris, was screened through approx. 0.60 cm mesh. All of the material was then bagged and removed to the guardhouse near the temple site of Qasr Qarun. These samples were then sorted to separate all lithic debris from the remaining matrix. This debris was then separately bagged. The units were designated as follows:

Neolithic Epipaleolithic

FS1-A FS2-A

FS1-B FS2-B

FS1-C FS2-C

FS1-D FS2-D

FS1-E FS2-E

FS1-F FS2-F

2.2 Geological collection.

There has been relatively little work done to determine accurately the sources of raw materials for this material, especially for these more southern sites. Most researchers have indicated various Eocene and Oligocene deposits above the northern rim of the Depression as the source of most lithic raw materials. That this is probably true is not in question; however, a more detailed analysis is required especially for the FS1 and FS2 assemblages as they are more distant from these northern source areas than assemblages collected by earlier workers.

Research into the archaeological and geological literature from this area revealed a number of possible deposits. These were spread throughout the Depression and included the following:

1) Fluvio-Marine series (Gebel Qatrani beds). These Upper Eocene-Lower Oligocene beds outcrop on the top of the scarp directly above Qasr el Sagha, and consist of hard cherty limestones with beds of tabular chert and flint. According to Beadnell (1905) this is the only area in the Fayum where chert occurs in primary geological position. Its use as a flint source for stone tools has been mentioned by numerous authors including Beadnell (1905), Caton-Thompson (1926, 1934), and Wendorf and Schild (1976).

Above this area on the very top of the northern scarp this series continues as a set of beds consisting of variegated sediments, including small chert and/or flint nodules. Again, these nodules have been mentioned by several authors as source materials. These beds also occur above the scarp near Qasr Qarun, only a few kilometers away from the sites of FS-1 and FS-2. These deposits have not, to my knowledge, been cited in the literature as possible lithic raw material sources. However, their proximity to the sites in this analysis obviates their study.

2) Plio-Pleistocene gravels. These gravels are found over wide areas of the Fayum, though only a few can be said to contain chert/flint adequate for stone tool production. Several of these deposits occur along the road between the Fayum and Cairo, roughly twenty kilometers north of the site of Karanis.

3) Pleistocene gravels. These heterogeneous deposits are found throughout the Fayum and occur in the area of FS-1 and FS-2. There is some indication that these materials were utilized, though to what extent in unknown. Certainly the use of local stone, if suitable for lithic production, would be preferable to distant sources.

I was able to investigate some aspects of each set of deposits, though financial constraints prevented a more thorough analysis this season. I surveyed each deposit on foot to establish the variability within the deposit in terms of its chert and flint content. Then a random sample of nodules was collected in grab sample fashion. Each sample constituted of roughly 20-30 nodules, depending on the homogeneity of the deposit and the size of the nodules themselves. I collected two samples each from either side of the beach ridge separating FS-1 and FS-2 (Pleistocene gravels, above); two samples from the Plio-Pleistocene deposits that outcrop on either side of the Fayum-Cairo road (designated PP1 and PP2); and one sample from the scarp north of the lake near FS-1 and FS-2 (part of the Fluvio-Marine series). All of these are secondary deposits, consisting of water-worn nodules of chert and flint.

3.0 Analysis.

3.1 Debitage analysis.

The goal of this analysis was to determine the completeness and complexity of core reduction in each set of assemblages and how these differ according to time period and rock type. Completeness of reduction indicates how far each core was reduced before being discarded. Complexity of reduction refers to the number of reduction events performed prior to an individual flake's removal. Each flake was measured using a paradigmatic classification consisting of attributes that directly reflect completeness and complexity. Besides basic size attributes, these attributes, or dimensions, include: the amount of cortex on the dorsal surface; platform configuration (cortex, number of facets); the presence of platform crushing and lipping; the number of dorsal scars; and the type of flake termination.

Each attribute was coded to a particular value, thus generating a class for each flake describing its configuration along these attributes. Different core reduction strategies will thus result in different proportions of classes for each assemblage. For example, a very incomplete reduction strategy will result in a large number of relatively large flakes with cortex and small numbers of dorsal scarring. A more complete reduction strategy would produce a wider range of flake sizes, most without cortex, and abundant dorsal scarring (see Teltser 1991 for a more detailed analysis of this kind of classification scheme).

I am also conducting a functional analysis of the debitage using an additional paradigmatic classification. To perform this analysis I examine the edges and surfaces of each flake for signs of use in the form of wear. Since these materials have been subject to sandblasting for the past several thousand years, a micro-wear analysis is not useful. However, a macro-scale analysis of wear is appropriate, since the larger-scale instances of wear will remain visible. These include chipping, crushing, deep abrasions, and some kinds of polish. Each instance of wear on a flake edge or surface is treated as a separate tool. Consequently, a single flake may have several instances of wear and thus be composed of several tools.

While this kind of analysis will not demonstrate what specific "activities" these objects were used for, it will show the kinds of actions that they were subjected to. These wear patterns indicate the mechanical parameters that were required of raw material, and is thus vitally important determining why particular materials were used over others.

3.2 Debitage results.

The analyses reported here are based on a preliminary analysis of two collection units, FS1-A (Neolithic) and FS2-A (Epipaleolithic), and the conclusions drawn herein are necessarily quite tentative.

Table 1 contains descriptive statistics for some of the variables from the technological classification. Both units have approximately the same proportions of whole and broken flakes/chunks, roughly half, and also about the same proportions of platform cortex. However, FS1-A has a slightly larger percentage of complete flakes with no cortex and a slightly smaller percentage of flakes with full dorsal cortex than does FS2-A. This indicates that FS1-A has a more complete reduction sequence than FS2-A, and also that initial core reduction may have occurred off-site during Neolithic (FS1) times.

Figures 2a and 2b show the distribution of flake sizes in the FS1-A and FS2-A assemblages using complete flakes only. I use weight as a measure of flake size, since the log of flake weight in these data and others (Teltser 1991) has a strong linear relationship with both (log10)length and (log10)thickness. There is a striking difference between the units. The FS1-A debitage has a skewed distribution indicating that the majority of flakes are small sized. The FS2-A debitage, in contrast, has a much flatter and even distribution of flake sizes. This trend is also noted in the descriptive statistics of the mean and standard deviation accompanying each graph: the FS1-A debitage has both a smaller mean flake size and smaller standard deviation. All other things being equal, this is to be expected given the difference in dorsal cortex measures: flakes removed from the inner parts of the core (and having no cortex) will be smaller than those removed from the outside.

Figure 3 shows the distribution of number of dorsal scars in each unit. Again, the FS1-A assemblage is somewhat more skewed toward a smaller numbers of dorsal flake scars, while the FS2-A debitage has a more even distribution of this variable.

The overall differences can be summarized thus: 1) FS1-A (Neolithic) debitage has a somewhat larger percentage of flakes with no cortex; 2) FS1-A debitage has a wider range of flake sizes with most flakes occurring in small sizes, while FS2-A (Epipaleolithic) debitage has a more even distribution of flake sizes; 3) FS1-A debitage has a wider range of numbers of dorsal scars and is again positively skewed toward more flake scars than those in FS2-A, which has a more even distribution. These factors, taken together, indicate that the reduction process at the Neolithic site of FS1 was more complete and complex than that of FS2.

The use wear data presented something of a surprise. Given the proclivity of settled agriculturalists to develop more expedient technologies, I had expected to find relatively more simple utilized flakes among the FS1 debitage. Looking again to Table 1, we see that this is not the case at all: the presence of use wear is nearly identical in both units. Further, the number of instances of wear per complete flake is similar: 1.84/flake for FS1-A as opposed to 1.61/flake for FS2-A. This indicates that as far as the utilization of debitage goes, there seems to be a great deal of continuity between the two periods. My analysis of the distribution of kind of wear is still in progress, and should clarify this issue still further. However, it does indicate that there is a great deal of functional variation in these assemblages apart from those pieces generally regarded as formal tools.

3.3 Geological analyses and results.

The raw materials that I collected were subjected only to visual inspection in anticipation of a more refined analysis in 1995. The purpose of this was obtain some idea of source locations and the proportions in which different materials occur. The results of these inspections are as follows.

First, from an initial analysis of the debitage, it appears that the Epipaleolithic assemblages contain a much more eclectic blend of raw materials than the Neolithic. Both of these assemblages contain a wide variety of cherts and are not easily classified into different types purely on visual inspection. However, when trying to sort some of the debitage samples into different material types, I found the Neolithic samples to contain several coherent groups. That is, it was easier to find groups of flakes that came from a single material type in the Neolithic assemblages. The Epipaleolithic assemblages rarely contained more than one or two coherent types of materials, each containing no more than a few flakes.

Second, the Neolithic materials also seemed to be more uniform in raw material characteristics. That is, all of the flakes seemed to be from relatively fine-grained, homogeneous cherts. The Epipaleolithic materials were much more variable in this regard.

Third, the Epipaleolithic people made extensive use of at least one raw material that is virtually absent from the Neolithic assemblages. This material has a distinctive chalky-white cortex, and was found to be abundant in all FS-2 assemblages analyzed. None was found in the Neolithic assemblages analyzed thus far. This material may be locally available on-site; I have yet to confirm this with an analysis of the geological specimens.

Fourth, there may be more use of local material in FS-2 assemblages. Geologically, the FS-2 side of the beach ridge contains an abundance of gravel, much of which is cherty in nature and of sufficient size for the creation of stone tools. There is virtually no knappable material on the FS-2 side. It appears that the Epipaleolithic people may have made more use of this local material; indeed, this material may not have been available to the Neolithic people, since the lake at that time would have covered up the material on that side of the ridge. I have been unable to confirm or refute this as yet, though the amount of the chalky-white material in FS-2 assemblages leads me to believe that it was of local origin.

4.0 Conclusions.

As I indicated in the introduction, very little attention has been paid to the debitage in Epipaleolithic and Neolithic assemblages; work has usually focused on the larger, obviously worked "tools" such as projectile points. In fact, until fairly recently, debitage was not even collected as artifactual. As my preliminary results indicate, however, there is a great deal of information that can be gained from an analysis of this material.

First, the use of raw materials seems to be much more varied in the Epipaleolithic. This could result from several causes. First, the Epipaleolithic seems to be characteristic of small mobile bands of hunter-gatherers, while the Neolithic seems to contain evidence of rather more sedentary populations. The more mobile Epipaleolithic people could have encountered and used many more sources of lithic raw materials in their yearly wanderings. If the Neolithic inhabitants were in fact less mobile than their Epipaleolithic counterparts, they would have had a more restricted range to choose from. This would not preclude them from making special trips to non-local areas to obtain raw material. However, the fact that the Neolithic assemblages can be grouped rather more readily into raw material types does suggest that they were obtaining their material from several restricted sources.

The presence of the chalky-white cortex material only at FS-2 may indicate two possibilities. If this material is, in fact, of local origin, and was not utilized by the Neolithic artisans, it could indicate that some mechanical aspect of this material made it unsuitable for Neolithic tool-making, rather than some aspect of accessibility. If it is non-local, it would indicate either that its mechanical properties made the acquisition of this material too costly, or simply not within the range of Neolithic movement.

Second, the debitage indicates that reduction in the Neolithic was much more complete and complex than in the Epipaleolithic. This would be consistent with the notion of Neolithic artisans making special trips to more distant (and more difficult to obtain) sources for raw materials. These more costly materials would be used more fully than less costly materials, resulting in a more complete and complex reduction sequence. Decortification, to reduce carrying weight and/or to insure quality, may also take place at the source rather than on-site resulting in less cortical flakes in the FS1-A debitage.

Third, there is a great deal of functional information in the debitage. There do not appear to be significant differences in the percentage of utilized flakes in these assemblages, nor in the number of instances of wear per flake. This suggests a certain amount of functional continuity between the Epipaleolithic and Neolithic, though the type of wear in each assemblage has yet to be fully examined. Even so, the very presence of numbers this high is surprising in light of past research into stone tools from this period.

Given these interpretations however, the locations at which tool manufacture took place needs to be considered. From the debitage analysis, it seems as though more of the entire reduction process occurred at the Neolithic site of FS-1. This would mean that most of the waste material from the production of tools was, in fact, located on this site and was recovered in my samples. However, if the Epipaleolithic peoples were obtaining their materials from the same source as the Neolithic artisans, but carrying out the reduction process at various locations, most of which were not located within my sampling space, the raw material characteristics could look different. It is possible that within their yearly round, tool production was carried out throughout the entire range of movement, FS-2 being only one site in this range. To what degree this may affect the distribution of raw materials is unclear. A more detailed analysis will surely illuminate this problem.

I have tried to show that there is a great deal of information in the non-tool portion of late prehistoric lithic assemblages. The tools themselves are just the end product of a process of raw material collection, manufacture, use, re-manufacture, and discard. Each of these steps is bound together with other aspects of the settlement and subsistence system of the prehistoric artisans, and in each step there is the potential for a great deal of variation that can affect the final form that the tools eventually take. Thus, a detailed study of each aspect of the manufacturing process is essential to a complete study of this tool variability. I have demonstrated that there is a significant amount of functional variation in the debitage, indicating that this material was not just incidental to the manufacturing process, but was an integral part of the prehistoric tool kit. I have also shown that there are significant differences in the reduction process between the two periods. A more complete analysis of these and other aspects of lithic variation will, I believe, give a more complete picture of the range of changes that took place in this important period in Egyptian history.