Inter-Assemblage Variability and Characteristics of Upper Paleolithic Flake Technologies in the Central Negev Highlands, Israel

 

Text of paper presented at the 65th annual meeting of the Society for American Archaeology in Philadelphia, April 2000

 

John K. Williams

 

Introduction

 

This study presents a preliminary analysis of flake-based assemblages in southern Israel.  Core reduction strategies, blank selection, and tool properties were analyzed, which in short, revealed that what is driving the technological and typological signature of most of these sites is a strong carinated component.  In addition, the nature of these assemblages also appears to be largely influenced by their position on the landscape, namely their elevation and proximity to permanent water sources.  Finally, these assemblages will be examined in a framework that considers their relationship to the current Upper Paleolithic classification scheme.

 

Research Objectives

 

Beyond sharing the same name and broad techno-typological similarities, the scale of affinity among assemblages currently labeled Levantine Aurignacian is largely unknown, particularly between the core and marginal zones, but also within each area.  This study presents an analysis of seven assemblages from the marginal zone, specifically from the central and western Negev highlands (Fig. 1). Five of the assemblages considered for this study are located in the Avdat/Aqev area of the Central Negev Highlands.  Ein Aqev (D31) is situated in the Nahal Aqev, and consists of 60 cm of stratified cultural deposits, which date to the end of the Upper Paleolithic (ca. 17,500 B.P.) (Marks 1976a).  Flanking the rim of the Nahal Zin, Sde Divshon (D27B) is a surface and in situ concentration of lithic artifacts covering ca. 800 m² in area.  While the lithic assemblage was classified as Ahmarian by its excavator, Gilead (1981) considered it to be characteristic of the Levantine Aurignacian (compare Gilead 1981: 340, Table 1; Marks 1981: 347, Table 2).  Excavations and surface collections at four other assemblages, Arkov (D22) A&B, D27A, and D18 revealed abundant surface and shallow subsurface lithic artifacts.  These assemblages were incorporated into the two-tradition model as marginal-zone representatives of the Levantine Aurignacian (Gilead 1981; Marks 1981).

 

In addition, two surface assemblages used in this analysis, K9 and G11, were found in the Har Harif, a high plateau in the Western Negev Highlands.  These sites possess a flake technology, and a toolkit with high percentages of Aurignacian elements (carinated scrapers and burins) (Larson and Marks 1977).

 

Figure 1.  Flake-dominated assemblages, and assemblages classified as Levantine Aurignacian

within the southern and central Levant  (sites marked with an asterisk indicate those used in this analysis).

            

Methodology

 

At one time or another, each of these assemblages has been broadly classified as "Levantine Aurignacian" by various researchers.  The methodology that was used to study these assemblages examines how they might compare among themselves, and to the broader Levantine Upper Paleolithic.  To test the degree of inter-assemblage affinity, attributes and criteria were chosen for study which detail core reduction strategies and tool use, manufacture and discard, within a framework that considers the properties of the Levantine Aurignacian, as it has been defined. The methodology proceeds from a basic division into class types, to more detailed information about blank types and scar patterns, and finally to metric measurements.  Of particular interest is a subcategory of ‘carinated pieces’, which is shared by both cores and tools.  This was done because the division between carinated cores, scrapers and burins is somewhat arbitrary, which can result in major tool class discrepancies that are more apparent than real.  To avoid this problem, all carinated pieces were treated equally (Fig. 2).

Sampling procedures were tailored to the overall goals of the sampling program and the idiosyncrasies of the sites.  In an attempt to control for intra-assemblage patterning, horizontally random samples were employed for large assemblages, such as Sde Divshon.  When dealing with stratified assemblages at a single site, such as Ein Aqev, each level was sampled horizontally.

 

 

Figure 2. Process of dividing each assemblage into analyzable units.

 

 

Results

 

The flake technologies considered in this analysis can be divided into two broad groups, which have been identified as single- and multiple-reduction strategies by Reid Ferring (1979), who performed a detailed technological analysis on many of the same assemblages.  Put simply, multiple strategies consist of a large and small component.  This is best seen at Ein Aqev, where the dichotomy between the large, retouched blades and tiny lamelles is strikingly apparent.

The only single reduction strategy is Sde Divshon, which is characterized by initial large removals with complex scar patterns, then a continuous shift to almost exclusively single platform cores for interior blade removal, which were often used as blanks for scrapers, or retouched into quite large el-Wad points.

 

A predominant carinated component is driving the nature of the remaining assemblages (Fig. 3).  The index was calculated by dividing the number of carinated pieces by the total number of burins and scrapers, since these are the two classes which merge into the carinated component in traditional type lists.  Sde Divshon has almost no carination index to speak of, while it is quite common in all of the other assemblages.  While D18 can be considered to have a significant carinated component, it is noticeably lower than assemblages like Arkov and Ein Aqev.  D18 is different is other aspects as well, which will be dealt with later in the paper.

 

Figure 3. Carination Indices for sampled assemblages.

 

 

Carination presents a real problem when trying to compare assemblages with basic technological and typological characteristics. While carination is a technique, traditional typologies pull it out of an assemblage's technology and put it into the type list. Further, there is a continuum from cores to scrapers to burins with carination, which might only reflect the arbitrary boundaries imposed by individual archaeologists.

 

To avoid this problem, everything with a series of elongated, twisted removals was separated into a single category: carinated artifacts.  While the percentages of scrapers and burins dropped in most toolkits, especially at Arkov and G11, these tools remain predominant across the board, despite having the carinated pieces removed (Fig. 4).  There were so few true carinated pieces from Sde Divshon, that the percentage of scrapers and burins did not change, falling just under 50% of the total tool kit

 

Figure 4. Scraper and burin percentages, before and after removing carinated pieces.

 

Because the scars on most carinated artifacts suggest that they could have been cores for twisted bladelets and lamelles Dufour, carinated artifacts may well display different size distributions than scrapers and burins.  The underlying reasoning is that if carinated pieces have non-overlapping distributions with scrapers and burins, then they should not be considered together in type lists.  As Figure 5 shows, carinated pieces are clearly different metrically than scrapers, but they cluster near the burins.  Each plot represents the mean removal surface length and width for a single assemblage. The result is three distinct groups: non-carinated cores, scrapers, and carinated pieces with burins.  Rather than suggesting that carinated pieces are burins, it is probable that most of these burins are initial stages of carinated core manufacture.  The dihedral burins in the sampled assemblages were produced with the same technique as the carinated artifacts, the only difference being that more spalls were removed on the carinated pieces, to the point where they started twisting around the blank.  It is also conceivable that a single technological sequence could have been used for two purposes.  If this were true, then carination is an efficient technique that produced more than one tool with a single method.  The separation between non-carinated cores and carinated pieces reflects multiple reduction strategies. Most of the non-carinated cores are bulky flake cores, which served to produce thick blanks that were required for use as secondary carinated cores.

 

Figure 5. Mean dimensions of artifact classes.

 

 

Considering the number of carinated pieces that were produced, it is curious why there are almost no twisted bladelets to speak of at most of these sites.  The only site with significant numbers of twisted bladelets is Ein Aqev, which is the source for the predominance of blades and bladelets at the site. Twisted bladelets should also be present at the other assemblages, since their negative impression is witnessed on the carinated pieces, but they are rare to absent.  To demonstrate this phenomenon, the minimum number of twisted bladelets that were produced at each assemblage was calculated.  Because there was a good sample of these bladelets at Ein Aqev, they were used to calculate the average twisted bladelet dimensions (Table 1).

 

Table 1. lamelles Dufour and twisted bladelets with carinated scar patterns from Ein Aqev, outliers removed.

 

Of 114 pieces at Ein Aqev, the average width was 5.6 mm.  For each assemblage, the total removal surface width was calculated for carinated pieces, then divided by 5.6.  This provided an estimate of the minimum number of bladelets that were produced from the existing cores.  If this number remained positive after subtracting the number of existing twisted bladelets, then an approximation of missing bladelets was provided. 

 

Most assemblages had large percentages of missing bladelets (Fig. 6).  Blades were included in

Figure 6 in this chart to show how the number of missing bladelets might affect the overall blade-flake dichotomy.  These percentages reflect both debitage and tool blanks since the bladelets produced from these carinated pieces would have probably ended up as both debitage and tools.  Except for Ein Aqev, where the twisted bladelets were present, and Sde Divshon, where there is very little carination, the missing bladelets provide quite a boost to the blade indices for the remaining sites.  In fact, both Arkov and D27A become dominated by blades and bladelets when this minimal estimate is applied.  This is important to consider, since the percentage of blades versus flakes is one of the major definitional criteria for Upper Paleolithic traditions.

 

Figure 6.  Blade indices, before and after calculating missing bladelets.

 

Two possibilities might explain where these bladelets went: they were carried away by the people who made them, or they were washed away by nature. Taphonomy is a possible candidate, since all of the sites with missing bladelets were found deflating from eolian deposits.  It is also very possible that they were carried out.  In a recent study by Chico Almeida (2000), twisted bladelets were shown to have been curated from a carinated assemblage at Anacreál.  Although this was a Terminal Gravettian site in Portugal, it is relevant here because of the similar technology.  Ninety-two percent of the weight of the assemblage was refitted.  This revealed that 50% of the twisted bladelets were missing from carinated reductions. Because this was a cave site, they were almost definitely removed by their makers.  It is also worth mentioning that a use-wear study of this assemblage found no signs of wear on the carinated pieces, unlike the scraping wear observed on flat, uncarinated scrapers.

 

Regardless of how they were removed, these missing bladelets bring the remaining carinated assemblages into the same range of blade-bladelet percentages as Ein Aqev, making them similar in that respect.  But not all carination is equal.  These assemblages can be further divided into two groups based on the method of carination.  Most of these sites show a predominance of lateral carination, but K9A and D18 are both dominated by typical carination.  Although this may seem trivial, it is actually quite a different concept for removing twisted bladelets, and accounts for much of the variability in other aspects of the technology.  Most of the carinated pieces at K9A and D18 were produced on core blanks, whereas the assemblages dominated by lateral carination used almost exclusively secondary blanks for carination.  Thus, instead of producing the bladelets through a separate reduction strategy involving secondary core blanks, the approach at K9A and D18 largely involved the sequential intensive reduction of larger cores into carinated cores.

On the basis of these preliminary results, the assemblages fall into three groups (Fig. 7).  As Reid Ferring (1979) observed before using some of the same assemblages, Sde Divshon most closely resembles the single reduction strategy of the Early Ahmarian, bearing a high degree of affinity with Boker BE, level III, which has a series of radiocarbon dates averaging 27,000 B.P.  The multiple reduction strategies form another group, characterized by a technology dominated by carination.  I have further separated these assemblages into two groups, on the basis of their carinated technique.  Since Ein Aqev is the only dated assemblage of these groups, at around 17,500 B.P., at least the lateral carinated assemblages can be tentatively placed within the Late Upper Paleolithic.

 

 

Figure 7. Basic division of sampled assemblages.

 

 

Discussion and Conclusions

 

How do these multiple strategy, carinated technologies fit into the broader Levantine Upper Paleolithic?  If we expand our scale only to the marginal zone, then the assemblages used in this study show a pattern of land use distinct from that of the Ahmarian.  While it is true that most Upper Paleolithic sites tend to be situated near highland areas, such as the central Negev Highlands, the mountains running from central-western to southern Jordan, and southern Sinai, the blade-dominated Ahmarian assemblages are almost always situated adjacent to the highest areas, typically in the bottom of the wadis, and specifically near permanent water sources.  In contrast, the sites used in this study are nearly all situated at high elevations, away from permanent water.  Figure 8 shows how various sites are situated on the landscape.  Although Ahmarian assemblages tend to be situated at various elevations, they are found within 1 kilometer of a perennial spring.  In contrast, all of the assemblages used in this study, except for Ein Aqev, are over 1 kilometer from the nearest spring.

 

Figure 8. Distribution of Upper Paleolithic sites in the marginal zone.

 

 

Technology also appears to be related to a given site's proximity to water.  Figure 9 shows that as a site's distance from a spring increases, it's blade-flake ratio decreases.  In other words, sites with predominant blade technologies tend to be sitting near a spring.  It is interesting to note that while Ein Aqev is sitting near a spring, with a blade-dominated technology, the remaining carinated technologies used in this study are all on the plateau, away from springs, with flake-dominated technologies.  This undoubtedly relates to the missing bladelet issue discussed earlier, but whether it is caused by curation or taphonomy remains to be seen.  However, since Sde Divshon also conforms to this pattern, perhaps we are seeing a real human response to resource distribution, rather than post-occupational disturbance.  Sde Divshon has an undeniable Ahmarian flavor, yet its broad artifact indices show a predominance of flakes, and a toolkit dominated by scrapers and burins.  Perhaps this relates to its location high on the Divshon plain, in contrast to the wadi floor, where the typical Ahmarian sites are located.  Judging from its size, Sde Divshon was probably a base camp, where a different set of activities was carried-out than the hunting camps sitting near water sources.  In this respect, the carinated technologies on the high plateaus may represent part of a broader subsistence strategy that ties-in with the bladelet-rich Ein Aqev.

 

Figure 9. Technology and distance from springs in the Upper Paleolithic of the marginal zone.

 

That a non-Ahmarian entity is present in the Negev is undeniable.  Putting a name on this entity, however, becomes more controversial. Suffice it to say that the two-tradition framework becomes problematic when the it crosses the boundary between the marginal and Mediterranean Core zone.  This is not surprising, since it was originally used to describe assemblages that were being discovered in a small area of the central Negev.

 

The danger lies in the use of broad technological and typological criteria to classify assemblages spread over a relatively large area.  As this Figure 10 shows, the sampled assemblages in this study conform in a very general technological sense to the Levantine Aurignacian of the Mediterranean Core Zone.  But if these broad categories are used, Sde Divshon is grouped with the remaining flake-dominated assemblages used in this study, which is misleading.  Also, the carinated index for all non-sampled assemblages was calculated using carinated burins and scrapers from type lists, which may not be comparable to the method used in this study.

 

Figure 10. General technological categories for sampled assemblages and Levantine UP.

 

 

Typological indices are even more difficult to compare, due to the carination problem.  Although carinated tools were removed from the type lists to make the comparisons in Figure 11, it is impossible to know if the remaining tools are actually uncarinated without having applied the same method of analysis.  Also, relatively nondescript tools, such as burins and scrapers, can vary within a singular technology, as seen at Sde Divshon.  This tends to call into question their utility as definitional criteria.

 

Figure 11. General typological indices for the sampled assemblages and the Levantine UP.

 

The multiple reduction strategies analyzed in this study do posses some of the attributes traditionally considered to be Levantine Aurignacian (e.g., carination, nosed and shouldered scrapers, multifaceted burins), while some attributes are extremely rare or altogether missing (Aurignacian retouch, bone and antler tools, and art).  If the most important attributes were tabulated and displayed as an Aurignacian index (e.g., Belfer-Cohen 1995), the question remains, how much of an index is needed before an assemblage can be called Aurignacian?  There is no denying the presence of an Aurignacian in the central-northern Levant, since this is where it was first identified and defined.  Even Bordes recognized the similarities of Ksar Akil XIII-VI to the French Aurignacian (Bergman 1987: 8).  Since there is a definite Aurignacian presence in the central-northern Levant, the next logical step is to expand an analysis of this sort to the classic cave and rockshelter sites.  While significant strides have been made in this direction (e.g., Belfer-Cohen 1995; Belfer-Cohen and Bar-Yosef 1999), a comparison of the available data, beyond the most basic level, is difficult because of differences in the way data were tabulated.  Currently, a large scale comparison of assemblages which have been considered to be Levantine Aurignacian, between the Mediterranean core and marginal areas, must deal with the apples-oranges syndrome.  Clearly it would be useful to do an inter-regional study, holding the method of analysis constant, to see if the analytic units fall within the same ranges, and what might be conditioning any perceived differences.

 

 

Bibliography

 

Almeida, F.

     2000    The Terminal Gravettian of Portuguese Estremadura: Technological Variability of the Lithic Industries.  Unpublished Ph.D. dissertation, Dallas: Southern Methodist University.

 

Bar-Yosef, O. and A. Belfer-Cohen

1988    The Early Upper Paleolithic in Levantine Caves. In The Early Upper Paleolithic: Evidence from Europe and the Near East. J. Hoffecker and C. Wolf (eds.), pp. 23-41. British Archaeological Reports International Series 437, Oxford.

1996    Another look at the Levantine Aurignacian. Coloquium XI – The Late Aurignacian, A.P. di Cesnola and A. Montet-White (co-organizers). Section 6 – the Upper Paleolithic. XIII International Union of Prehistoric and Protohistoric Sciences, Forli. ABACO, Edizioni.

1999    The Levantine Aurignacian: 60 Years of Research. In Dorothy Garrod and the Progress of the Palaeolithic, W. Davies and R. Charles (eds.), pp. 119-134. Oxbow Books.

 

Belfer-Cohen, A.

1995       Problems in Defining a Prehistoric Culture: An Example from the Southern Levant. In Nature et culture, M. Otte (ed.), pp. 245-257. Colloque de Liege (13-17 decembre 1993), Liege, E.R.A.U.L. 68.

 

Bergman, C. A.

1987    Ksar Akil Lebanon: A Technological and Typological Analysis of the Later Palaeolithic Levels of Ksar Akil, Volume II: Levels XIII-VI.  Oxford: BAR International Series 329

 

Bricker, H.M.

1976    Upper Paleolithic Archaeology. Annual Review of Anthopology 5:133-148.

 

Ferring, C. R.

1977    The Late Upper Paleolithic Site Ein Aqev East. In Prehistory and Paleoenvironments in the Central Negev, Israel. Vol 2. The Avdat/Aqev area, Part 2, A.E. Marks (ed.), pp.81-118. Southern Methodist University Press, Dallas.

1979    Technological Variability and Change in the Late Paleolithic of the Negev.  Unpublished Ph.D. dissertation.  Southern Methodist University: Dallas.

1988    Technological Change in the Upper Paleolithic of the Negev.  In Upper Pleistocene Prehistory of Western Eurasia. H. Dibble and A. Montet-White (eds.), pp. 333-348.  The University Museum, Philadelphia.

 

Garrod, D.A.E.

1953    The relations between southwest Asia and Europe in the latter Paleolithic age. Journal of World History 1: 13-38.

 

Gilead, I.

1981    Upper Paleolithic tool assemblages from the Negev and Sinai.  In Prehistoire du Levant, J. Cauvin and P. Sanlaville (eds.), pp. 331-342.  Paris, C.N.R.S.

1991    The Upper Paleolithic Period in the Levant. Journal of World Prehistory 5, 2: 105-153.

 

Goring-Morris, A.N.

1987      At the Edge: Terminal Pleistocene Hunter-Gatherers in the Negev and Sinai, 2 Volumes. Oxford: BAR International Series 361(i).

 

Hahn, J.

1972    Aurignacian signs, pendants and art objects in Central and Eastern Europe.  World Archaeology 3: 252-266.

 

Henry, D.O.

1989    From Foraging to Agriculture: The Levant at the End of the Ice Age.  University of Pennsylvania Press, Philadelphia.

 

Larson, P.A. and A.E. Marks

1977        Two Upper Paleolithic Sites in the Har Harif. In Prehistory and Paleoenvironments in the Central Negev, Israel. Vol 2. The Avdat/Aqev area, Part 2, A.E. Marks (ed.), pp.173-189. Southern Methodist University Press, Dallas.

 

Marks, A.E.

1976a  Ein Aqev: A late Levantine Upper Paleolithic site in the Nahal Aqev. In Prehistory and Paleoenvironments in the Central Negev, Israel. Vol 1. The Avdat/Aqev area, Part 1, A.E. Marks (ed.), pp. 227-292. Southern Methodist University Press, Dallas.

1976b  Terminology and chronology of the Levantine Upper Paleolithic as seen from the Central Negev, Israel. In Deuxieme Colloque sur la Terminologie de la Prehistoire du Proche-Orient, F. Wendorf (director), pp.49-75.  Union Internationale des sciences prehistoriques et protohistoriques. Sep. 15, Conseil International de la Philosophie et des Sciences Humaines de I’Unesco, Nice.

1981    The Upper Paleolithic of the Negev.  In Prehistoire du Levant, J. Cauvin and P. Sanlaville (eds.), pp. 343-352.  Paris, C.N.R.S.

 

Peyrony, D.

1933    Les industries “aurignaciennes” dans le bassin de la Vézère.  Bulletin Société Prehistorique la France. 

 

Phillips, P.

1980    The Prehistory of Europe. Indiana University Press. Bloomington and London.

 

Svoboda, J.

1994    The Pavlov Site, Czech Republic: Lithic evidence from the Upper Paleolithic. Journal of Field Archaeology, 21: 69-81.