Save

Late Pleistocene and Holocene Lithic Variability at Goda Buticha (Southeastern Ethiopia): Implications for the Understanding of the Middle and Late Stone Age of the Horn of Africa

In: Journal of African Archaeology
View More View Less
  • 1 McDonald Institute for Archaeological Research, Downing Street, CB2 3ER Cambridge, UKUMR CNRS 7194, Département Homme et Environnement, Muséum national d’Histoire naturelle – Université de Perpignan Via Domitia – Sorbonne Universités, Institut de Paléontologie Humaine, 1 rue René Panhard, 75013 Paris, FranceCorresponding author
  • | 2 UMR CNRS 7194, Département Homme et Environnement, Muséum national d’Histoire naturelle – Université de Perpignan Via Domitia – Sorbonne Universités, Institut de Paléontologie Humaine, 1 rue René Panhard, 75013 Paris, France
  • | 3 Institute of Archaeology, The Hebrew University of Jerusalem, Mt Scopus, Jerusalem 91905, IsraelInstitute of Human Origins, Arizona State University, USA
Full Access

Abstracts

The Late Pleistocene is a key period to understand the shift from the Middle (msa) to the Late Stone Age (lsa) in Africa. More generally, it is also a crucial time for elucidation of changes in the technological behaviours of human populations in Africa after the main Out of Africa event of modern humans ca. 60-50 thousand years ago. However, the archaeological record for this period is relatively poor, particularly for the Horn of Africa. Here we present a detailed analysis of the lithic assemblages from Goda Buticha (gb), a cave in southeastern Ethiopia, which has yielded a long stratigraphic sequence including Late Pleistocene and Holocene levels. This study (1) contributes to a better knowledge of the late msa in the Horn of Africa; (2) documents a late Holocene lsa level (gb – Complex i); (3) highlights the presence of msa characteristics associated with lsa features in the Holocene (gb – Layer iic). This adds to the emerging record of great lithic technological variability during the Late Pleistocene and Holocene in this region.

Abstracts

The Late Pleistocene is a key period to understand the shift from the Middle (msa) to the Late Stone Age (lsa) in Africa. More generally, it is also a crucial time for elucidation of changes in the technological behaviours of human populations in Africa after the main Out of Africa event of modern humans ca. 60-50 thousand years ago. However, the archaeological record for this period is relatively poor, particularly for the Horn of Africa. Here we present a detailed analysis of the lithic assemblages from Goda Buticha (gb), a cave in southeastern Ethiopia, which has yielded a long stratigraphic sequence including Late Pleistocene and Holocene levels. This study (1) contributes to a better knowledge of the late msa in the Horn of Africa; (2) documents a late Holocene lsa level (gb – Complex i); (3) highlights the presence of msa characteristics associated with lsa features in the Holocene (gb – Layer iic). This adds to the emerging record of great lithic technological variability during the Late Pleistocene and Holocene in this region.

1 Introduction

Understanding the mechanisms behind the “transition(s)” between major techno-cultural complexes has been the focus of recent archaeological research. There are two facets to such efforts. One is theoretical, and focuses on clarifying what is meant by the term “transition” (e.g. Hovers & Kuhn 2006; Camps & Chauhan 2009). The other is practical, attempting to apply the criteria derived from theoretical considerations to the archaeological record (e.g. “transitional industries” of the Late Pleistocene in Europe and Africa, Early Upper Palaeolithic or early Late Stone Age (lsa) industries/cultures (see Tostevin 2000, 2003; Hovers & Kuhn 2006; Kuhn 2013; Tostevin 2013).

figure 1
figure 1

Location of sites mentioned in the text. (Created using Natural Earth Data in qgis).

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

In this context, the shift from the Middle Stone Age (msa) to the Late Stone Age (lsa) in Africa has been the topic of major discussion because of its numerous specificities, including the fact that the same hominin species, Homo sapiens, is regarded as the maker of most of the msa and of the lsa material cultures (e.g. Basell 2008). Moreover, this shift seems to have taken place asynchronously across geographic regions. Some of the sites document very early lsa (>50 ka bp), e.g. Border Cave in South Africa (Villa et al. 2012), Mumba rockshelter in Tanzania (Prendergast et al. 2007; Diez-Martin et al. 2009; Gliganic et al. 2012), Enkapune Ya Muto in Kenya (Ambrose 1998), while other localities present relatively late msa, e.g. Rose Cottage Cave in South Africa (~28 ka uncal bp, Clark 1997; Wadley 1997); sites within the Wasiriya Beds on Rusinga Island (>30-45 ka, eastern Lake Victoria Basin, Tryon et al. 2010); B1s3 in the Ziway-Shala Basin in Ethiopia (~33-34 ka cal bp, Ménard et al. 2014) (see Fig. 1). Such instances demonstrate the great variability of the material cultures and lithic industries dated to the Late Pleistocene and indicate a complex pattern for the shift from the msa to the lsa.

This state of affairs has led some researchers to avoid the terms Middle and Late Stone Age in their descriptions of the assemblages (e.g. at Mochena Borago, Brandt et al. 2012). The difficulty to relate lithic assemblages dated to Marine Isotopic Stages (mis) 4-3 to the general schema of the msa and lsa highlights the need for detailed studies on local and regional scales in order to best understand cultural change. This is especially important for the end of the Pleistocene where a growing body of genetic studies document complex patterns of dispersals within, out of, and back into Africa (e.g. Campbell & Tishkoff 2010; Hodgson et al. 2014).

In the Horn of Africa particularly few sites document the period of mis 4 and mis 3, and even fewer are securely dated. This paper focuses on a cave site located in southeastern Ethiopia, Goda Buticha, which yielded a long stratigraphic sequence dated from 62 ka to 1 ka by means of radiocarbon and optically-stimulated luminescence (osl) methods (Tribolo et al. 2017; Pleurdeau et al. 2014). This sequence documents a depositional hiatus and chronological gap from ca. 24 ka to 8 ka. The evidence from Goda Buticha is consistent with stratigraphic records from many other sites in the Horn of Africa, from which archaeological deposits dated to mis 2 are missing. This gap may correspond to either a cessation of human occupation in the region or results from research bias (e.g. Bon et al. 2013; Ménard et al. 2014; Pleurdeau et al. 2014; Ménard & Bon 2015).

With such a long chronological gap in the sequence of Goda Buticha, one might expect an abrupt change in material culture characteristics at the site before and after the hiatus in deposition. Yet, preliminary results of the analysis conducted on part of the assemblage indicated that this was not the case (Pleurdeau et al. 2014). This paper confirms and analyses in detail the unexpected typo-technological similarities between the assemblages from Goda Buticha, placing this sequence in the context of local scale variability of Late Pleistocene and Holocene lithic variability in the Horn of Africa.

2 Late Pleistocene and Holocene Record of the Horn of Africa

Lithic assemblages from the Late msa in the Horn of Africa (see Fig. 2) are few and characterised by high variability. In addition to the “classical” characteristics of the msa (Goodwin & Van Riet Lowe 1929; Goodwin 1946), which include flake production by means of Levallois flaking systems and high percentages of retouched points amongst the tool types, many late msa assemblages from the Horn of Africa show high percentages of elongated blank production or of pointed blank production (K’one, locality 5 extension: Kurashina 1978; Aduma, locality A5 and later: Yellen et al. 2005; Porc-Epic: Pleurdeau 2005a, 2005b; Deka Wede 1: Bon et al. 2013; Ménard et al. 2014).

Lithic assemblages attributed to the lsa in the Horn of Africa are characterised by “general” lsa technological features (e.g. Goodwin 1946; Phillipson 1982), such as prismatic blade or bladelet production, microliths as the dominant tool type, and absence of characteristic msa Levallois production or retouched points. The earliest lsa of the region is late compared to adjacent areas in East Africa: at FeJx4 around Lake Besaka, three radiocarbon dates obtained from Layer 2 on fragments of ostrich eggshells gave dates between 19-22 ka (Brandt 1982: 60-61), while the lsa sequence at Ziway-Shalla only starts during the terminal Pleistocene (Bon et al. 2013; Ménard et al. 2014). The lsa encompasses a great variability in both technology and typology and is sometimes considered as a “catch-all” category (Bon & Fauvelle-Aymar 2014).

Within the Late Pleistocene and Holocene archaeological record of the Horn of Africa a number of sites contain assemblages with features of both the msa and the lsa, and therefore have not been attributed to either one of these cultural taxonomies (see Fig. 2). Such instances include the Late Pleistocene levels at Mochena Borago with rare Levallois cores, flake and blade(let) production and retouched tools dominated by scrapers, retouched points and backed pieces (Brandt et al. 2012); at Midhishi 2 in Somalia, Unit CSUb, dated around 18 ka, has yielded an assemblage with both Levallois and blade production, backed pieces and retouched points (Gresham 1984); the lower units of Shelter 7 of Laas Geel (Somaliland), considered to be of a Terminal Pleistocene age despite conflicting dates, have yielded material with both Levallois and prismatic blade production, retouched points and microliths. These assemblages have been attributed to a revisited “Hargeisan” (Gutherz et al. 2014). Similarly, the first study of the Goda Buticha material highlighted the presence of both msa and lsa features throughout the sequence (Pleurdeau et al. 2014).

Overall, the tempo and pattern of technological changes during the Late Pleistocene in the Horn of Africa remain unclear, leaving open questions about the persistence, reintroduction, or convergent, independent appearance of msa features in lithic assemblages throughout the end of the Pleistocene and the Holocene, despite the widely-recognized gap corresponding to mis 2 in many of the sites. Analyses of technological changes throughout this crucial period, focusing on elongated blank (blade / bladelet) production, Levallois production, as well as the manufacture of retouched points and microliths, may help in a more nuanced understanding of the processes leading to the observed record. The analysis of Goda Buticha, with its long and well-dated Late Pleistocene and Holocene sequence and rich lithic assemblages, contributes to these issues.

The main objective of this paper is therefore to conduct an in-depth study of the lithic assemblages from Goda Buticha in order to further investigate diachronic technological change (or persistence) at the site and how these characteristics fit with the local and regional archaeological record and more generally with the current definitions of the msa and the lsa.

figure 2
figure 2

Characteristics of lithic assemblages from main sites in the Horn of Africa dated between 100 ka and 8 ka.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

3 Material and Methods

3.1 Goda Buticha

Goda Buticha (gb), located some 30 km from Dire-Dawa, in southeastern Ethiopia, was discovered1 in 2007 by the South East Ethiopia Cave Survey Project (seecsp) (Assefa et al. 2014). The seecsp noted the presence of rock art as well as a high density of lithic material on the floor of one of the chambers. Two field seasons at the site took place in 2008 and 2011. The long (>2 m) stratigraphic sequence was divided on the basis of sedimentological observations into two main complexes (ii and i, from bottom to top) and three main layers (Layers iid-iif, Layer iic and Complex i). The sequence is dated from the end of mis 4/mis 3 to mis 1, with a major depositional and chronological gap corresponding to mis 2 (Tribolo et al. 2017; Pleurdeau et al. 2014) (see Fig. 3). This site represents one of the rare well-dated long stratigraphic sequences in the Horn of Africa. Sedimentological analyses (Tribolo et al. 2017) confirmed initial field observations that did not identify any major disturbances or mixing of sediments.

figure 3
figure 3

Goda Buticha. A. general view of the cave. B. Floor plan indicating the excavated areas from which the current sample derives. C. A schematic stratigraphic section showing the main units and dates D. Barchart: count of lithic material according to the main stratigraphic units. The Y-axis shows elevation (in cm) in relation to the arbitary excavation datum (*osl dates after Tribolo et al. 2017; ** radiocarbon dates, after Pleurdeau et al. 2014).

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

The sequence of Goda Buticha yielded human remains, ostrich eggshell beads, engraved ostrich eggshell, macro- and micro-faunal remains, and abundant lithic artefacts (Tribolo et al. 2017; Assefa et al. 2014, 2018; Leplongeon 2014; Pleurdeau et al. 2014). The material analysed in this paper derives from the two excavated squares and includes the as yet unpublished lithic material excavated in 2011. The lithic assemblages discussed in the current paper originated from Layers iid-iif (ca. 63-25 ka), Layer iic (ca. 8-6 ka) and Complex i (ca. 2-1 ka). Lithic artefacts show peaks of density corresponding to these layers (see Fig. 3 and Table 1). Material derived from the boundaries of the layers (i.e. iic/iid or i/iia-iib) was not considered in the analysis.

table 1

Goda Buticha. General composition of the material from 2008 and 2011 excavations (Squares A1 and B1) (elong. = elongated blanks, conv. = convergent blanks, ret. = retouched)

table 1

3.2 Methods: chaîne opératoire and Attribute Analyses

In order to best highlight changes or continuities in the techno-typological characteristics of the assemblages and place them in a comparative context, they were subjected to an in-depth analysis of four main lithic categories corresponding to key-elements that define the msa or the lsa in the Horn of Africa (see section 2, Fig. 2 and suppl. data): the Levallois production, the production of elongated blanks, the manufacture of retouched points and the microliths. The latter has been fully presented in a previous paper (Leplongeon 2014) and will not be developed here.

‘Elongated blank’ is a generic term used here for blanks that are more than twice as long as they are wide. Elongated blanks are divided into blades (large elongated blanks) and bladelets (small elongated blanks). Here we use an assemblage-specific method to separate blades from bladelets by observing the distribution of the length of all elongated products. In case of a bimodal distribution, the group with the larger dimensions are considered blades and the other bladelets. Although subject to sample-size bias, this method alleviates the need to rely on a fixed, arbitrary value for defining blades and bladelets. The elongated blanks and associated cores are analyzed in a second step in order to characterize the different technologies behind their production.

In this paper, the lithic assemblages of Goda Buticha are first generally described, including the frequency and modalities of the Levallois production, in order to have a general view of the composition of the assemblages. This will also ensure that the assemblages studied are comparable with previous studies of the material (2008 material of gb, Pleurdeau et al. 2014). An in-depth description of the characteristics of the elongated blank production(s) and of the retouched tools focusing on the retouched points (and with a summary of the data on microliths) will then be presented. Relying on these data, comparative analyses between the levels of Goda Buticha will be conducted.

All selected artefacts were analysed relying on the chaîne opératoire concept. This approach, borrowed from ethnologists, was first applied to prehistoric artefacts by Andre Leroi-Gourhan (Leroi-Gourhan 1964) and later developed by other prehistorians (e.g. Geneste 1985; Pelegrin et al. 1988; Karlin et al. 1991; Soressi & Geneste 2011). It allows a reconstruction of the reduction strategies and their analysis in terms of “knowledge” and “know-how”, which can then be interpreted to distinguish between groups with different technical practices. This qualitative approach is combined with an attribute analysis. The same attributes are described for each lithic artefact (e.g. Hovers 2009: 18). This allows the quantification of technological characteristics. Non-parametric statistical tests were deemed favorable for this study (the chi-squared test with simulate p-value, the Fisher test, the Kruskal-Wallis test followed by a Wilcoxon pair-wise test) because they allow treatment of samples that are not normally distributed.

4 Lithic Assemblages from Goda Buticha

4.1 Goda Buticha: Previous Studies

Pleurdeau et al. (2014) describe the trends in lithic technology throughout the sequence based on the material from the 2008 season. Raw materials are dominated by chert (51%), obsidian (30%), basalt (9%) and quartzite (6%). Flake production predominates (55%). Levallois cores are numerous and show both centripetal preparation for the removal of a preferential flake and the unipolar recurrent method. Other cores show several flaking surfaces. Levallois products, including mainly flakes but also a few blades (n=2) and points (n=5) are present, associated with debordant flakes. Volumetric blade production also occurs. Blades account for 16% of debitage; single platform cores (prismatic / pyramidal) are also present (7% of cores). A few crested blades and overshot products attest to the presence of a “true” blade production. Retouched tools account for 14% of the 2008 assemblage (excluding debris) and consist of retouched points (21%), backed pieces (15%), side-scrapers (35%) and various retouched pieces. There are no abrupt typo-technological changes throughout the sequence (Pleurdeau et al. 2014: 126). The frequency of obsidian increases from the bottom to the top of the sequence, as does the frequency of backed microliths. Overall, the sequence shows the association of typical “lsa” elements such as the backed pieces with typical “msa” elements such as the Levallois flaking and retouched points.

Overall, the lithic material retrieved from both the 2008 and 2011 excavations present similar characteristics (see Table 1).

A local origin is postulated for the chert, basalt and quartzite used at Goda Buticha. Of note is the relative high frequency of artefacts made on obsidian. In the near-by site of Porc-Epic cave, some 30 km northeast of gb, where obsidian represents about 6% of the assemblage (e.g. Pleurdeau 2005b), a distant origin is suggested. No obsidian source is known in the vicinity of the site although Clark & Williamson (1984: 49) mention that “a source of poor obsidian is known about 40 km to the east of Dire Dawa but this would not have been suitable for artifacts (W.H. Morton, pers. comm.)”. Geochemical analysis of obsidian artefacts from Porc-Epic Cave, show the use of sources located 140 km or even 300 km away but also identify a number of unknown sources (Negash & Shackley 2006; Vogel et al. 2006; Negash et al. 2011). Thus we cannot rule out the possibility that some obsidian source(s) may have been closer to gb.

4.2 Goda Buticha IId-IIf, 63-25ka

4.2.1 General Characteristics

osl and ams radiocarbon age estimates place layers iid-iif between 63±7 ka and 24.8±2.6 ka (Tribolo et al. 2017; Pleurdeau et al. 2014). These layers yielded in total 3068 lithic artefacts, representing an average density of around 1500 artefacts per m3. A peak in density is observed from spit 90-80cm to spit 40-30cm above datum (with ages between 45 and 25 ka, Tribolo et al. 2017, see Fig. 3). There are no major changes in the frequencies of main categories of stone artefacts (see Table 1). Given its homogeneity, the lithic assemblage from the lower part of complex ii (including layers iid, iie and iif) was analysed as one assemblage.

table 2
Goda Buticha. General overview of the Levallois production
table 2

The assemblage is oriented towards flake production, which represents two-third of the artefacts (excluding chips and chunks). Cores are also mainly oriented towards flake production (57%, n=24/40). Point production is minor (less than 1% of the blanks). None of the cores are for point production. Levallois products and elongated blanks are frequent.

The Levallois production is well represented, with 14 cores (one third of all cores) attributed to the Levallois concept (see Table 2). All the cores but two are for flake production, including centripetal preparation for the removal of a preferential flake (n=4), unipolar recurrent flaking (n=5) and centripetal recurrent flaking (n=3). In addition, 73 Levallois blanks (representing ca. 5% of the debitage items) were recovered, as well as 8 retouched Levallois blanks (9% of all retouched tools), consisting of four unifacial points, two retouched flakes and one notched piece).

Nearly a third of the Levallois blanks (32%) are made on basalt, compared to all the debitage (9%). The frequency of Levallois cores made on basalt is much lower (ca. 7%; Table 2). This may suggest an off-site Levallois production on basalt. Levallois blanks are mainly flakes (n=48), and the use of the Levallois flaking system for elongated blank production is minor (n=5 Levallois blades, n=2 Levallois cores for recurrent blade production). The majority of the products have a plain or faceted striking platform and show centripetal (n=27), or bidirectional (n=14) scar patterns, consistent with the observations on the cores.

Figure 4 presents a synthetic view of the technological characteristics of the lithic artefacts from layers IId-IIf.

4.2.2 Production of Elongated Products

Elongated Blanks

Layers iid-iif yielded 279 elongated blanks. Chert is the dominant raw material (52%), followed by obsidian (21%), basalt and quartzite (12% each). These raw material frequencies are similar to those observed in all the debitage, indicating that there was no particular raw material selection for producing the elongated blank in iid-iif (see Table 3). Dimensions of complete elongated blanks (N=130) are given in Table 4 and Fig. 5. There are no statistically significant raw material-related differences in artefact dimensions (see Table 4).

figure 4
figure 4
Main characteristics of the lithic assemblage from layers iid-iif, Goda Buticha.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

Although very small elongated products are present (length below 25mm), the distributions of length and width are unimodal and suggest a single production system of both the large and small items. The elongation index is relatively low as well as the thickness index.

table 3
Goda Buticha. Percentage of artefacts made on obsidian, per category
table 3
table 4
Goda Buticha. Dimensions of elongated blanks
table 4
table 4

* p-value Kruskal-Wallis <.05 (comparisons of thickness of obsidian elongated blanks between layers, post-hoc Wilcoxon test with p-value <.05 only between gb i and gb iic.

** comparisons of dimensions of elongated products within the levels according to their raw material. Obsidian elongated blanks significantly smaller than chert elongated blanks, themselves significantly smaller than blanks on other raw materials (p-value <.01, except for thickness between chert and other (p-value <.05)).

Only seven cores show elongated removal scars, including four in obsidian. Two of them are Levallois (one unipolar recurrent, the other one bidirectional recurrent). Another two are related to a non-Levallois planimetric conception of debitage (one with a unipolar debitage, the other one with a bidirectional debitage). Two cores show a volumetric conception of debitage and one an intermediate one (see Fig. 6).2 Striking platforms are mainly faceted (n=3,) or plain (n=3), with one dihedral striking platform.

The elongated products themselves present plain platforms (56% of determinable platforms), a rather high frequency of linear and punctiform platforms (17%), dihedral (13%) and faceted (9%) platforms (see Table 5). The high percentage of salient to very salient bulbs (66%) associated with rather large platforms (mean=3.24mm, sd=2.2) suggests the use of hard hammer percussion. Core trimming elements related to blade production are consistent with production based on a planimetric conception of debitage with flakes bearing elongated removal scars (n=7) and non-Levallois planimetric debordant products (n=7). One overshot item and one ridge blade indicate the use of a volumetric conception of debitage (see Fig. 6 h).

The direction of debitage as shown by the removal scars on the elongated products is mainly unipolar (61%), while unipolar and orthogonal (12%), bidirectional (9%) or centripetal (14%) scar patterns occur in low frequencies. Elongated products are relatively thin, with flat to slightly curved longitudinal profiles (51%), flat lateral profiles (75%) and with parallel edges (39%) or show at least one distal convergent segment (30%).

The data from the cores, technical pieces and elongated products (see Figs 6 and 7) indicate that part of the knapping at iid-iif was oriented towards the production of relatively small elongated products, with the occasional use of the Levallois recurrent method or of the prismatic blade production. The production of elongated products seems to have involved little preparation of the cores’ striking platforms or of the debitage surfaces. The latter appear to be quite flat, as indicated by the longitudinal and lateral profiles of the blanks as well as some debordant products.

figure 5
figure 5
Graphs of length and width of complete elongated blanks from Layers iid-iif, iic and i of Goda Buticha: A: scatterplot of length and width according to raw material, B: histogram of lengths; C: Goda Buticha iic. Histogram of lengths according to raw material.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

Retouched Elongated Products

Twenty-eight percent of the retouched tools (n=29/102) are made on elongated blanks, six of which are small (below 30mm of length). This percentage is higher than the frequency of elongated blanks among the debitage (18%), which may reflect a selection of this type of blanks for the retouch. The majority of retouched items on elongated blanks are retouched blades. There are also scrapers, composite tools, unifacial or bifacial tools and a microlith on a bladelet (see Table 6). The selection of obsidian for retouched elongated blanks is clear (62% of items in this category).

4.2.3 Retouched Points

Production of Pointed Flakes

There are no cores with point removal scars in the iid-iif lithic assemblages, and the assemblage contains only 25 pointed flakes (<2%, see Table 1). Five of them are Levallois points (see above and Table 2). However, in the absence of cores, the use of Levallois flaking systems for the production of pointed blanks remains hypothetical.

Retouched Points

Thirty retouched points (including ten fragments) were counted in the iid-iif assemblage (29% of all retouched tools; Table 7 and Fig. 7). The extent of the retouch often prevents the determination of the type of blank for many points, which can be determined for only nine retouched points. Six are made on pointed flakes, including one Levallois point, and three are made on flakes.

table 5
Goda Buticha. Types of platform and bulb for proximal fragments and complete elongated blanks
table 5

Thirteen of the retouched points are made on chert (43%), seven (23%) on obsidian and nine (30%) on basalt. These values are comparable with raw material distribution in the total assemblage, suggesting that there was no selection for specific raw material for retouched points.

The points are retouched mainly unifacially (n=20). Two points are parti-bifacial and eight are bifacial. The retouch is invasive to covering, especially for bifacial points (Table 7 and Fig. 7 (f-p)). Most of the retouched points show regularly convergent edges, the edges starting to converge from the first half of the length (as opposed to converging occurring only in the distal part).

The retouched points are relatively small (mean length =37.8mm, sd =6.7) and narrow (mean laminarity index = 1.8, sd=0.4) (Table 8). There are no statistically significant differences in point dimensions according to either raw material or type.

4.2.4 Microliths

Only two pieces in iid-iif, one backed microlith and one retouched bladelet, can be considered as microliths (Leplongeon 2014). In addition, it may be noted that a relatively low number of small elongated products were retouched, despite their high frequency among the unretouched blanks (see above).

Goda Buticha iic, 8-6ka

4.3.1 General Characteristics

Layer iic, dated between 6372 +/- 47 cal bp and 7771 +/- 51 cal bp (Tribolo et al. 2017, Fig. 3), yielded 2276 lithic artefacts, representing an average density of around 1625 artefacts per m3. The peak of density is at 140-120cm above datum (table 1 and fig. 3). The assemblage is oriented towards flake production; more than half of the artefacts (excluding chips and chunks, see Table 1) are flakes. Cores are also mainly oriented towards flake production, at least in their last stages of reduction (58%, n=35/60). Point production is minor while elongated blanks are frequent (29% of debitage).

figure 6
figure 6
Lithic artefacts from Goda Buticha iid-iif. (a,c): blade cores, (b): Levallois core, (d): bladelet core, (e,f ): Levallois blades, (g-m): technical pieces related to elongated blank production, including (h,l,m): technical pieces related to a volumetric conception of debitage, (n-r): elongated blanks.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

While the planimetric conception of debitage dominates among the flake cores, only three (5% of all cores) can be attributed to the Levallois flaking system, including two with a centripetal preparation for the removal of a preferential flake and one with a bipolar recurrent flaking. In addition, 33 Levallois blanks have been recovered (3% of the debitage), including eight possible Levallois blades. Only two of these blanks are retouched. Levallois blanks show mainly centripetal or bipolar scar pattern (see Table 2).

figure 7
figure 7
Lithic artefacts from Goda Buticha iid-iif. (a-e): elongated blanks, ( f-p): retouched points, (q-s): small uni- or bifacial obsidian ovates, (t-u): microliths.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

Figure 8 presents a synthetic view of the technological characteristics of the lithic artefacts from iic.

table 6
Goda Buticha. Types of retouched tools
table 6
table 7
Goda Buticha. Characteristics of retouched points
table 7

4.3.2 Production of Elongated Blanks

Elongated Blanks

Layer iic yielded 355 elongated products. They are made mainly on chert (64%), followed by obsidian (27%) and basalt (5%). One-hundred and ninety elongated products are complete (see dimensions in Table 4). The items are relatively small (with a mean length ca. 30mm), with low values of the laminarity and thickness indices. The length distribution of complete elongated products (Figure 5), suggests that blades (n=214) are distinguished from bladelets (n=141) at the length threshold of 26mm.

Elongated products made on obsidian tend to be smaller than those made on other raw materials (p-values of a Kruskal-Wallis test and the following pairwise Wilcoxon test <0.05: Figure 5). Since bladelets are made on obsidian more often (38%) than on chert (20%), the bimodal distribution may be a reflection of the preference of using small obsidian nodules. This may also indicate a more intensive reduction of obsidian compared to other raw materials or reflect the smaller size of the nodules.

Fourteen cores in Layer iic show elongated removal scars (Figure 9 b-f). Although the sample is too small to characterize their size distributions, eight of the cores are smaller than 30mm and qualify as bladelet cores. Seven cores are made on chert, while five cores are made on obsidian. None of these cores are Levallois. Four of them are planimetric in conception, with a “direct” (i.e. opportunistic) debitage without any preparation of the debitage surface. Seven are volumetric (semi-rotating) cores, including six of the bladelet cores. Most of the cores show plain striking platforms (n=10/18 surfaces), followed by dihedral (n=5/18), cortical (n=2/18) and faceted (n=1/18) striking platform. Thirty-six core trimming elements associated with elongated blank production were found in Layer iic. Ridge blades (n=4) and overshot products (n=7) attest to the use of technologically formal blade production, at least for some of the elongated blanks.

table 8
Goda Buticha. Main dimensions (in mm) of retouched points per layer and per raw material
table 8

Eight elongated blanks are of Levallois type but in the absence of Levallois blade cores, these products may represent equifinality rather than unequivocal evidence for the use of the Levallois flaking system. A planimetric conception of debitage is indicated by “debordant” products (n=15), as well as flakes with elongated removal scars (n=9).

Consistent with core platform characteristics, the blades and bladelets from Layer iic present mainly plain platforms (43%) followed by faceted (17%) and linear / punctiform platforms (11%) (see Figures 14 and 15). Bulbs are mostly salient, even when associated with narrow (linear or punctiform) platforms (Table 5), suggesting the use of a hard hammer percussion. Given the high frequency of linear/punctiform platforms, the use of soft hammers cannot be excluded.

Most of the elongated blanks do not show any cortical surfaces. The direction of debitage as shown by the removal scars is mainly unipolar (55%), but differs between blades and bladelets. The unipolar scar pattern characterizes more than 70% of the bladelets compared to 45% for the blades. The blades present a higher frequency of the “unipolar and lateral” scar pattern (22%) compared to the bladelets (10%) and of the “bidirectional” scar pattern (12% vs. 4% for blades and bladelets, respectively).

The thickness index shows that blades are relatively thick (36%), while bladelets are thin to relatively thick. Their longitudinal profile is flat to curved, and lateral profile mainly flat. Blades and bladelets show similar characteristics for their contours, with 41% showing parallel edges and around 25% with convergent edges.

Retouched Elongated Products

Sixty two percent of the 111 retouched tools of the Layer iic assemblage are made on elongated blanks, 42 on blades and 28 on bladelets (table 6). This high frequency indicates that elongated blanks (blades and bladelets), which only amount to 29% of the debitage, were preferentially selected for retouch. Most of the retouched elongated blanks are on obsidian (52% of retouched blades and 61% of retouched bladelets. This indicates a preferential selection of obsidian for the retouched tools (Fisher Test <0.05).

figure 8
figure 8
Main characteristics of the lithic assemblage from layer iic at Goda Buticha.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

4.3.3 Retouched Points

Production of Pointed Flakes

A single very small core bearing one pointed flake removal scar was found in the IIc assemblage, along with 15 pointed flakes (including 4 fragments), of which two are of Levallois type.

Retouched Points

Layer iic has yielded 23 retouched points (Tables 6 and 7), out of which 18 are unifacial points (including 16 unifacial bilateral points). Two additional ones are parti-bifacial points and another three are bifacial points. Most of the retouched points present covering retouch (52%), leading to a high frequency of undetermined blanks. Seventy percent of the retouched points (N=16/23) are made on obsidian (Figure 10), which indicates a strong preference for this raw material.

figure 9
figure 9
Lithic artefacts from Layer iic at Goda Buticha. (a) Levallois core, (b,c): blade cores, (d-f ): bladelet cores, (g-n): technical pieces related to blade(let) production, (o-v): blades, (w): bladelet.

Citation: Journal of African Archaeology 15, 2 (2017) ; 10.1163/21915784-12340010

Only ten points present identifiable platform types: six are faceted, three dihedral and one cortical. This would indicate the use of blanks from prepared cores.

The retouched points from iic are relatively small (mean length is 37.8mm±13.8) and elongated (mean length to width ratio is 1.9±0.4) (Table 8). Although retouched points on obsidian tend to beare smaller, the difference is not statistically significant (p-value of the Kruskal-Wallis test >.05), which is probably due to the small sample. Most of the retouched points have regularly convergent edges with a rounded proximal part.

4.3.4 Microliths

Layer iic has yielded 31 microliths (28% of all retouched tools). They are composed of 10 retouched bladelets and 21 backed microliths, including four geometric microliths. Detailed treatment of these items is presented by Leplongeon (2014).

4.4. Goda Buticha i, <4 ka

4.4.1 General Characteristics

Complex i is dated by 14C and osl from 4.07 ± 55 ka cal bp to 0.7 ±0.1 ka and may reach even later dates, given that the topmost layers remain undated (Tribolo et al. 2017). The assemblage consists of 550 artefacts. The calculated density of ca. 550 artefacts per m3 (Table 1, Figure 3) is a third of the values obtained for the layers in Complex ii.

The assemblage is oriented towards flake production (56% of the debitage), with elongated blank accounting for 22% of the debitage. Among the 15 cores from Complex i, nine present flake removal scars and six present elongated flake removal scars. Planimetric conception of debitage dominates among the cores. Two Levallois cores were shaped by bipolar recurrent flaking, for elongated blank production and for flake production (Table 2). Seven additional cores present blank reduction from the widest surface, from one or two striking platforms. There is a single pyramidal core for bladelet production.

The assemblage contains no points and only two Levallois flakes. Elongated blanks are relatively frequent (21%), although less than in the lower layers. In contrast to the lower layers, obsidian is the main raw material used in the assemblage (accounting for 56% and 75% of the assemblage (excluding and including chips, respectively) (Table 3). Levallois production is of minor importance, with only two possible cores and two flakes.

Figure 11 presents a synthetic view of the technological characteristics of the lithic artefacts from Complex i.

4.4.2 Production of Elongated Blanks

Elongated Blanks

Only 44 elongated products were counted in Complex i, of which 33 are complete (Table 2). Most of the elongated items are small (n=28 less than 30mm long). The majority (ca. 85%) are made on obsidian, indicating a preferential selection of this raw material for blade(let) production.

Six cores in Complex i present elongated removal scars, five of them are microlithic (< 30mm long). All of the cores are on obsidian. One is pyramidal while the others seem to be related to a planimetric or intermediate conception of debitage, with little preparation of the flaking surface (see Fig. 12). Six core trimming elements are related to blade(let) production. Of these two are flakes with elongated removal scars, three are non-Levallois debordant products and one is an overshot flake.

Most of the elongated products show plain (33%) or faceted (28%) platforms, this is consistent with the data from the cores’ striking platforms where both of these types are present. Bulbs of percussion are mostly salient to very salient for blades, as well as for bladelets, suggesting the use of hard hammer percussion. The majority of the elongated blanks show a unipolar direction of debitage (n=23/40). Blades and bladelets are relatively thick, with curved longitudinal profiles and flat lateral profiles (n=26/40), although twisted profiles occur (n=13/40). Most of the artefacts show parallel edges.

Retouched Elongated Products

Half of the retouched tools from Complex i are made on elongated