Introduction

Ongoing debate about the characterisation of the Middle Palaeolithic to Upper Palaeolithic transition often involves opposing claims about the level of continuity or discontinuity displayed in artefact assemblages. Perhaps the most common position is that the transition is marked by increased diversity and standardisation of formal implements (see Mellars 1989:365; Binford 1989:36). Since the demonstration by Dibble (1984, 1987) that a significant portion of assemblage variability in the Middle Palaeolithic is explicable as morphological transformations of one implement "type" into another as reduction proceeds, many analysts have used this as a key contrast to the inferred technological structure of Upper Palaeolithic assemblages. Upper Palaeolithic formal implement types have often been seen as being unambiguously defined, functionally specific, and representing the end product of a sequence of reduction. For example, describing what he calls an ‘unstructured feature’ of Mousterian behaviour, Marcel Otte (1990:443) claimed that

Perception of dramatically different technological patterns is the two periods has been used in the formulation of models that posit a rapid development of more complex cultural activity. Depiction of Upper Palaeolithic implements as rigidly defined end products has often been seen as a reflection of modern cognitive patterns. The implications of perceived contrasts in assemblage structure are variously phrased; while Binford (1989:19) talks in terms of "planning depth", others relate the variety of specifically designed implements made and used to the presence and complexity of mental typologies. For example Mellars (1989:365) characterizes the situation as follows:

This depiction of the transition between Middle Palaeolithic and Upper Palaeolithic has not gone unchallenged. Dissenting opinion holds the transition begins within the Middle Palaeolithic and that the initiation of different organisational structures can be perceived in Middle Palaeolithic assemblages (eg. Lindly and Clark 1990:61; Marshack 1990:469; Reynolds 1990:273). It appears that this perception is often based on technological analyses, which are more sensitive to change than are implement typologies. Differing depictions of assemblage structure, based on whether a technological or typological perspective is applied to the analysis, suggest another challenge. While the Middle Palaeolithic has been redescribed from a technological viewpoint, by Dibble and others, Upper Palaeolithic assemblages often retain a more strictly typological description. Hence, Clark and Willermet (1995:155) have recently suggested that the morphological standardisation of Upper Palaeolithic implements is a typological illusion (see also Holdaway 1991). Intriguingly the application of technological perspectives to implement typologies in regions of the world first colonised by modern humans, that is North America and Australia, has often revealed manufacturing/recycling processes in which one distinctive morphology is transformed into another (eg. Flenniken 1985; Flenniken and Raymond 1986; Hiscock 1993, 1994; Hiscock and Veth 1991; Holmes 1890; Wheat 1976). And in Epipalaeolithic backed blades from southwest Asia similar morphological transformations have also been perceived (Neeley and Barton 1994). It may therefore be asked whether the perceived difference in technological structure between Middle Palaeolithic and Upper Palaeolithic owes more to the typological approach to assemblage description than it does to the prehistoric patterns of artefact manufacture. As noted by Lindly and Clark (1990:61), specific technological examinations of Upper Palaeolithic industries are necessary to answer this question.

In the following analysis of an assemblage from north Africa I will show that some scrapers have been reworked into burins, while some burins were modified to form scrapers. The assemblage selected for this study is from Haua Fteah, a large limestone cave on the north coast of Libya. Excavated in the early 1950’s by Charles McBurney, the 12m deep excavation yielded several hundred thousand artefacts spanning both the Middle and Upper Palaeolithic phases. McBurney’s (1967) extensive site report concentrates on descriptions of formal implement types, and his classification of specimens as burins or scrapers follows normal conventions. He argued that while the assemblages displayed regional characteristics they were broadly comparable with Palaeolithic assemblages elsewhere around the Mediterranean. In particular, McBurney (1967:135) described the Dabba phase at Haua Fteah as ‘basically similar to the Upper Palaeolithic and Mesolithic of south-west Asia, Europe and the Maghreb’. Consequently, the demonstration of morphological transitions between implement types at Haua Fteah can contribute to discussions of the structure of Upper Palaeolithic technology in general.

The Dabba sample

Burins and scrapers employed in this study were recovered from eight spits excavated during McBurney’s 1955 field season (see Table 1). These spits span much of the Dabba phase represented at Haua Fteah, beginning with spit 1955/88 in layers XVII-XVIII and extending down to spits 1955/28, 1955/29, and 1955/93 in layers XXI-XXII. These excavation units were selected because they were available for inspection, contained relatively large numbers of implements, and the packaging of the collection largely retained McBurney's original categories. In addition spits excavated in 1955 typically have a more precise stratigraphic provenience than excavation units from McBurney's earlier field seasons, and for this reason samples have been drawn only from the 1955 material.

A total of 204 implements were present in these eight spits. All were manufactured on a homogeneous chert, and therefore raw material properties are not considered in the following discussion. Following McBurney's classification the dominant class are minimally retouched backed blades, which make up 47.6% of the assemblage. Backed blades are not discussed in this paper. The next most abundant class of implement are scrapers (28.9%), most of which were labelled as end scrapers by McBurney. Burins were the other common implement form found in the Dabba, and in this sample they comprise 19.1% of McBurney's implements. While chronological changes in relative proportions of implement types are documented by McBurney (1967:166-167), the relative abundance of each implement in this sample reflects the pattern for the Dabba as a whole. As listed in Table 1, the following analysis is based on a comparison of 39 burins and 59 scrapers, as identified by McBurney, drawn from a representative sample of spits in the Dabba levels.

McBurney (1967:71) has provided a series of radiocarbon age-estimates for the strata from which these artefacts derive. The specimens discussed below are effectively bracketed by radiocarbon dates of 14,120± 100 (GrN2586) and 16,670± 150 (GrN2585) in spit XVIII and 31,150± 400 (GrN2550) in spit XX. Hence the spits represented in this sample have an antiquity of roughly 14,000 years BP to more than 30,000 years BP.

Comparison of Dabba scrapers and burins

The feasibility of morphological transformations between burins and scrapers can, in the first instance, be evaluated by comparing the size and shape of the flakes on which each class of artefact is made. This procedure has been used successfully by Dibble (1984, 1987) for Middle Palaeolithic implements. For this purpose the best comparison is the dimensions of platforms on specimens where retouching has not obscured platform features. Table 2 shows that there is virtually no difference between burins and scrapers in terms of platform width (t = 0.378, df = 23, p = 0.705) and platform thickness (t = -0.335, df = 23, p = 0.744). This pattern is consistent with both implement forms being made on flakes of similar dimensions.

When the two implement forms were discarded they also had extremely similar dimensions (Table 3). Mean values of burins are consistently slightly larger and longer that the values for scrapers, but although both classes are quite variable in size they show much the same range of sizes. Simple comparison of dimensions using t-tests reveal that the differences between the two classes are not significant at even the p = 0.10 level (df = 96). While this similarity in plan shape and size is not a test of the proposition that scrapers were transformed into burins, or burins into scrapers, it is consistent with the notion that transformations could have occurred.

Retouching patterns are also extremely similar for both classes of implement. For example, most burins have one or more margins that display a series of regular flake scars indicating blows applied to the ventral surface of a flake have retouched the dorsal face. Indeed, inspection of burins shows that many have extensive retouch onto the dorsal face on the lateral margins or distal end, and at one point in their reduction these specimens must have looked very like what McBurney called ‘end scrapers’, ‘flake scrapers’, ‘denticulate scrapers’, or ‘side scrapers’. The form and extent of this scraper-like retouch is generally just like that of the specimens McBurney recognised as scrapers. One measure of the nature of this retouch is the length of these kinds of retouch scars on burins ( = 4.79mm, sd = 2.3, n = 19) and on scrapers ( = 4.10mm, sd = 1.7, n = 57). This comparison shows that the size of scraper-like retouch on both implement classes is virtually the same (t = 1.200, df = 74, p = 0.241).

Another example of similar retouching patterns is the number and size of burin facets, the negative scars of ‘burin spalls’, on specimens. As noted above, scrapers as well burins sometimes have burin facets (Table 4). Again, t-tests indicate the differences between the two classes are not significant at even the p = 0.10 level (df = 45). Hence, when burin facets are present on specimens McBurney classified as scrapers, the width of the facets and the number per specimen is almost identical to those visible on burins. This observation is also consistent with the proposition that individual specimens have been transformed between classes.

Evidence for morphological transformations

Evidence presented thus far could be indicative of a consistent classificatory ‘mistake’. For example, if McBurney had classified a number of objects that were burins as scrapers that might explain the presence of burin facets on some items labelled as scrapers, although it would not explain the other similarities in size, shape and retouch form between the two types. However, it is not possible to explain away the similarities in this way, because there is independent and undeniable evidence of specimens being transformed from one class to the other. This evidence is the physical superpositioning of negative flake scars, in which burin facets either truncate, or are truncated by, scraper-like retouch (i.e. short scars on the dorsal face caused by blows to the ventral face). The arrangement of scar superposition is direct evidence of the sequence of retouch and documents the morphological transformation of specimens. The relationship between these traditional classes of implement can therefore be better understood if we look at the sequence of retouch rather than dimensions alone.

As noted above, a substantial portion (70%) of burins have one or more rows of short, steep retouch onto the dorsal face (see Figure 1). These specimens are in fact just like ‘scrapers’, except that superimposed over the scraper retouch are longer scars, or burin facets, removing part of the lateral or distal margin. Such patterns, which indicate that scraper morphologies were transformed into burin morphologies, can be seen in McBurney’s (1967:152) illustrations.

Figure 1. Specimen from 1955/93, classified by McBurney as a burin. The flake was extensively retouched along the lateral and distal margin prior to multiple burin spalls being struck from the distal end. (Drawings Ó Joanna Richards.) Click for a photograph of the specimen.

In a similar way there are specimens in several of McBurney’s ‘Scraper’ categories that have clear burin facets on them (see Figure 2). Measurements presented earlier have already demonstrated that the form and extent of this burin-like retouch is generally identical to that of the specimens McBurney recognised as burins. These specimens are in fact just like ‘burins’, except that superimposed over the long scars removing part of the lateral or distal margin are a series of steep scars encroaching onto the dorsal surface (as per scrapers). Such patterns, which indicate that burin morphologies were transformed into scraper morphologies, can also be seen in McBurney’s (1967:153) illustrations.

Figure 2. Dorsal surface of a specimen from 1955/93, classified by McBurney as an end scraper. On the left burin facets originating from both the proximal and distal ends are overlaid by extensive scraper-like retouch. (Drawings Ó Joanna Richards.) Click for a photograph of the specimen.

Repeated transformations of this kind are documented within the sample of artefacts studied here. For example, 4.5% of specimens had a sequence of scar superpositioning that showed a transition from scraper morphology to burin morphology and then back to scraper form again. One specimen even showed a four stage sequence of retouching: scraper-burin-scraper-burin. Evidence for repeated transformations is also visible on specimens illustrated in the original Haua Fteah site report, but not included in the sample used here (eg. McBurney 1967:152, specimens 21 and 22).

While the Dabba manufacturing processes are not limited to a single manufacturing sequence, most of the transitions mentioned here follow a sequence broadly as follows:

1. The first phase consists of retouching onto the dorsal face to give a simple end and/or side scraper form.
2. In the second stage burin spalls may be removed from the margins.
3. The third stage reverts to scraper-like retouch
4. In the fourth stage further burin spalls may be removed from the margins.

Evidence for transformations between burins and scrapers is surprisingly common. Of the 98 specimens in the composite sample, 25 show at least one transformation clearly visible in the scar pattern they present. This should be considered a minimum count, since the most recent phase of retouch could conceivably remove evidence for a pre-existing morphological state. For example, some scrapers have extensive retouch on two or three edges, covering more than 100mm of edge, which could destroy definite evidence of burin spalls. Similarly, some burins, especially dihedral forms, have little of the original margins remaining, and evidence of any earlier and different morphology has been removed. Hence, the evidence of scar sequences that remains on these artefacts indicates that morphological transformations were common, bi-directional, and any single specimen might have undergone several cycles of transformation.

Consequently the available evidence does not support any suggestion that the excavator has mis-classified these specimens. As his illustrations and discussion clearly shows McBurney (1967) was able to see the features mentioned here, and documented both burins with scraper-like retouch and scrapers with burin facets. In fact, with an understanding of the reduction processes outlined here it is possible to perceive regularity and consistency in McBurney’s classification. In most instances his classification of a specimen relates to the sequence of retouching. When scraper-like scars comprised the most recent layer of retouching he classified the object as a ‘scraper’, even though it might have burin facets visible. Conversely, when the final retouching event produced burin spalls he classed the object as a ‘burin’", even though a pre-existing scraper-like morphology was visible. Hence, on most occasions, the most recent set of scars appears to have been used to classify the objects into mutually exclusive classes. In doing so he found a coherent way to classify complex objects, and ignored the implications of the typological transformations that are apparent within the Dabba assemblage at Haua Fteah.

Implications of morphological transformations

There are a number of important implications of this work for the interpretation of Upper Palaeolithic artefact assemblages. Some of these implications can be described by noting three differing interpretations of the pattern described above.

1. The first, and most conventional explanation, might be that of the two forms recognized in the traditional typology, one is a preparatory stage (i.e. preform) while the other is the end product and tool. This model would imply that many specimens in the preform stage should not automatically be regarded as implements. For example, if the final product was considered to be the burin, then the retouch that gave the scraper form might be seen to merely shape the piece and to prepare a suitable surface on which to apply the burin blow. Such an interpretation could apply even to those ‘scrapers’ with no burin facets, since they may have proved to be an inappropriate form after the preparatory retouching and were therefore discarded. This argument would be inverted if it were argued that it was the scraper morphology that was the end-product and the removal of burin spalls merely a form of preparatory shaping. Irrespective of which morphology is identified as the preform this interpretation of implement form would reduce the diversity of end products typically recognized in typological analyses.
2. Both typological forms might have functioned as tools, and specimens might therefore have multiple functions throughout their life. From this view retouching might change the form of the object to suit the function at hand, and the typological composition of the assemblage would therefore partly reflect the number of functional changes, and partly the nature of the final function since the morphological transformation related to that use would often over-write earlier morphologies. This explanation retains the notion that tool form might be manufactured to fit a functional design, but does not imply that the tool form is necessarily an end-product.
3. Changing morphologies might reflect the application of knapping actions, chosen from a repertoire of technological procedures, with the general aim of extending the use-life of each item, either by generating flakes or maintaining a useable edge. Implement form at the time it is discarded is then conditioned by the interaction between its history of use and the technical difficulties of continuing to flake the piece (such as raw material properties and the economics of obtaining replacement stone). Such a suggestion has been made for artefact form on other occasions (eg. Hayden 1977), and is one model of Middle Palaeolithic scrapers (Dibble 1987). While such a view does not necessarily deny a role for stylistic factors, it allows for regular sequences of morphological change without such factors.

While debate is likely to continue about which of these depictions is most useful, they all create difficulties for conventional typological analyses of Upper Palaeolithic industries. In all three cases the percentage of traditional implement classes in an assemblage cannot necessarily be taken to accurately reflect either site function or cultural affinity in any simple way. The transformation of one type into another may allow more sophisticated discussion of the economics and structure of artefact manufacture and function, while complicating any attempt to characterize Upper Palaeolithic formal implement types simply as the end-products of a sequence of reduction. This conclusion was anticipated by Dibble (1984:435) when he argued that

An important consequence of the presence of morphological transformations between supposedly discrete implement types is that the Dabba artefact assemblage at Haua Fteah is not structured in the way that Upper Palaeolithic industries are normally said to be structured. Indeed, the presence of morphological transformations in the Dabba clearly parallels their widely documented existence in earlier assemblages (see Dibble 1987; Cornford 1986), suggesting continuity between Middle Palaeolithic and Upper Palaeolithic assemblages, at least in this aspect of their technological structure and in North Africa.

If the patterns of transformation found in the Dabba levels of Haua Fteah are duplicated more widely in Upper Palaeolithic assemblages, this would be consistent with observations in the New World and the Antipodes and would suggest that at least in some production contexts morphological transformations frequently exist in the technologies of recent hominids. Hence the posited distinction between Middle Palaeolithic and Upper Palaeolithic technological structures may need to be re-examined.

Acknowledgments. I am indebted to Chris Chippindale, Robin Boast and the staff of the Cambridge University Museum of Archaeology and Anthropology for providing access to the Haua Fteah collections. I thank the Antiquity referees for helpful comments on the paper. This research was funded by a research grant from the Northern Territory University. The drawings are by Joanna Richards.

Author: Peter Hiscock, Dept. Archaeology and Anthropology
Feedback: peter.hiscock@anu.edu.au .
Date Last Modified: 1-July-97
URL: http://artalpha.anu.edu.au/web/arc/resources/papers/ph/hauantiq.htm