Introduction

Open-air sites are increasingly recognised as an essential component of the archaeological record for Middle Stone Age hunter-gatherer lifeways. Recent fieldwork has aimed to establish the pattern of landscape use for past humans occupying the understudied Doring River catchment zone, to the east of the Cederberg Mountains across the Tankwa Karoo. Today, this region receives some of the lowest annual rainfall levels in South Africa (<100mm per annum) and is classified as semi-arid desert. Consequently, the landscape is sparsely vegetated, featuring succulents endemic to the Karoo biome. Surveys in August 2014 were centred on a 30km-long segment of the Tankwa River, recording over 7000 artefacts at 45 different localities (sites). One of these sites contained the largest Middle Stone Age unifacial point assemblage reported from either open-air or rockshelter sites in the Western and Northern Cape regions. The specific preferential Levallois strategy used for point production, together with the unusually high use of silcrete, marks this as a site of importance for our understanding of Middle Stone Age adaptations to an arid, marginal environment.

Tweefontein site and context

The site, Tweefontein, is located on a low, flat-topped ridge (c. 330 × 180m) on the Tankwa River floodplain, lying between two channels of the river that were dry at the time of survey (Figures 1 & 2a). The ridge is situated on Dwyka Formation geology of the Karoo Super-Group, and is covered with angular clasts of rocks such as quartzite and sandstone, and outcroppings of Dwyka diamictite. This forms a single deflated ‘desert pavement’ surface of artefacts and rocks overlying sand (Figure 2b & d).

Figure 1.

a) Map showing the location of Tweefontein; b) aerial view of the site (image from the National Geo-spatial Information (Department of Rural Development and Land Reform, South Africa).

Figure 2.

Views of the Tweefontein site: a) the ridge (circled in white) viewed from the north; b) recording artefacts along a linear transect; c) Later Stone Age circular stone wind-break (white scale: 2m); d) stone artefacts on the deflated sand surface.

Artefacts deriving from the Later Stone Age and Middle Stone Age were identified across the ridge in spatially discrete clusters, distinguishable by typology and technology that correlated with the relative patination of artefacts made of hornfels. The Later Stone Age was confined to two clusters and associated with roughly built circular stone structures, interpreted as wind-breaks (Figure 2c). The composition and context of these assemblages were typical of other Later Stone Age occurrences observed during survey in the Tankwa Karoo area. The main Middle Stone Age assemblage covered the flat centre of the ridge, with a particular concentration of points and point cores towards the north-western end. In contrast to the Later Stone Age, the Middle Stone Age at Tweefontein represents a completely new type of assemblage for this period in South Africa.

Artefacts were recorded along two perpendicular transects of approximately 50m, centred on the area of densest archaeological material. The recorded sample included all points and point cores, diagnostic cores, other technologically distinctive material such as blades, and all artefacts made of silcrete. Additionally, all of the points and point cores in the concentration mentioned above were recorded. For each artefact, attributes (e.g. raw material, platform type, cortex percentage, retouch) were noted, the artefact photographed and a GPS point taken.

Lithic assemblage

The most notable feature of the assemblage at Tweefontein is the emphasis on pointed forms, in both flake and core morphology (Table 1). Over 150 points and point fragments were recorded, ranging from unretouched debitage points to invasively retouched unifacial points and some unusual bifacial points (Table 1; Figures 3, 4e & f). Most points were produced using a preferential Levallois strategy, with at least 40 preferential point cores recorded (Figures 4 & 5). Typically, these are triangular or cordiform in shape, with bilateral or radial preparation to create a steep distal ridge for a preferential point removal. The points struck from these cores have finely faceted platforms and nearly all have marginal informal retouch or edge-damage. Other flaking debitage at the site does not show such edge-damage, which would suggest it is not post-depositional.

Table 1.Cores and points recorded in the sampled area.

Figure 3.

Unifacial points displaying a range of morphologies and technologies; raw materials: a–c) hornfels; d–e) dolerite; f–i) silcrete; j & k) chert.

Figure 4.

Bifacial pieces; raw material: silcrete.

Figure 5.

Preferential point cores; raw materials: a–e) hornfels; f) quartz breccia; g) chert.

Hornfels was the most frequently used raw material in the overall site assemblage and was available as cobbles in the Tankwa River channel and from outcrops along the dolerite dyke around 5km away. Other locally occurring raw materials that were used for both points and cores were dolerite, chert, fine-grained quartzite and quartz breccia. Most significant is the high incidence of high-quality silcrete in the point and core sample, with much of it retaining both outcrop and river-cobble cortex. Silcrete is known to outcrop in three localities (21, 16 & 16.5km from the site), but, geologically, the closest outcrop would be no less than 10km away where Table Mountain quartzitic sandstone geology occurs. In spite of the range of knappable raw materials available in the immediate vicinity of the site, silcrete was systematically transported to this location, with in situ knapping indicated by very small debitage fragments and primary cortical flakes.

Discussion

Based on published reports, no other Middle Stone Age assemblages in southern Africa document this form of preferential point technology. The closest industries are those of the Middle Stone Age2 ‘Mossel Bay’ variant, which includes point or convergent flake production through a unipolar Levallois strategy (Wurz 2000), and the post-Howiesons Poort, in which unifacial points are emphasised (Conard et al. 2012). In general, the assemblage shows characteristics typical of the later part of the Middle Stone Age: the emphasis on silcrete and other fine-grained raw materials, a high level of standardisation and highly refined flaking. The absence of bifacial foliate points and backed artefacts argues against a Still Bay or Howiesons Poort affiliation. The numerous finely and invasively flaked unifacial points are characteristic of post-Howiesons Poort-age assemblages (50–60 kya) (Reference MacKayMackay 2011; Mackay et al. 2014); the lack of chronometric dates, however, means that this association remains tentative.

Preliminary technological analysis indicates that the Tweefontein point cores bear close resemblance to the specific form of Levallois preferential point production described for the Nubian Complex, a Middle Stone Age regional variant of North Africa and southern Arabia (e.g. Olszewski et al. 2010; Usik et al. 2013). In particular, the Tweefontein technology is characterised by the bilateral preparation of Nubian Type 2 cores with a smaller number of cores showing distal preparation of the Nubian Type 1 method. The three criteria that define the Nubian Levallois strategy are: preparation to create a steeply angled median distal ridge, a pointed or cordiform shape and a prepared striking platform (Usik et al. 2013), all of which are present in the Tweefontein Middle Stone Age assemblage.

The Nubian Levallois Complex is geographically far removed from the southern tip of Africa and therefore we suggest no cultural relationship. Furthermore, current dates place the Nubian Complex in MIS 5 (c. 130–74 kya) (Olszewski et al. 2010; Usik et al. 2013)—far earlier than the provisional post-Howiesons Poort association suggested here. The Nubian Complex, however, has been interpreted as a specific adaptation to marginal desert conditions (Usik et al. 2013), which is of interest given the comparable situation of the Tweefontein site. Our current understanding of the industries of the later Middle Stone Age is patchy, with the assemblage at Tweefontein signalling potential technological variability that is still unrecognised. Further work at the site and across the surrounding region will aim to define the industry so that it can be better understood in relation to other Middle Stone Age assemblages.