Middle Stone Age bone tools from the Howiesons Poort layers, Sibudu Cave, South Africa

Recent research has shown that the use and manufacture of pointed bone tools is not the preserve of MSA or Upper Palaeolithic cultures. The earliest use of pointed bone tools comes from the Plio cave deposits at Sterkfontein, Swartkrans and Drimolen in South Africa (Robinson, 1959, Brain and Shipman, 1993, Brain and Shipman, 2004, Keyser et al., 2000, Backwell and d'Errico, 2001, Backwell and d'Errico, 2004a, Backwell and d'Errico, 2005, d'Errico et al., 2001, d'Errico et al., 2003, d'Errico and Backwell, 2003). The wear pattern on the tips of these implements has been interpreted as the result of digging, either for extracting tubers from the ground (Brain and Shipman, 1993, Brain and Shipman, 2004) or termites from their nests (Backwell and d'Errico, 2001, Backwell and d'Errico, 2004a, Backwell and d'Errico, 2005, d'Errico et al., 2001, d'Errico et al., 2003 d'Errico and Backwell, 2003). Evidence of shaping by means of grinding on some of the more robust horncores suggests that early hominins had the cognitive ability to modify bone intentionally for optimal efficiency as part of a non-modern behavioural repertoire, albeit for expedient use (d'Errico and Backwell, 2003, Backwell and d'Errico, 2004b, Backwell and d'Errico, 2005). Bone objects from Broken Hill (Kabwe), Zambia, attributed to the early MSA and thought to be associated with Homo heidelbergensis (elsewhere named Homo rhodesiensis), are interpreted by Barham et al. (2002) as two gouges and a point. A point from Mumbwa Cave (Pinto Llona et al., 2000) is considered doubtful (d'Errico, 2003). Other evidence for bone working in the MSA is provided by barbed and unbarbed bone points from the Katanda sites in the Semliki Valley, Democratic Republic of the Congo (formerly Zaire), dated ∼90 ka (Yellen et al., 1995, Brooks et al., 1995, McBrearty and Brooks, 2000, Feathers and Migliorini, 2001). A point tip, a mesial fragment, an almost complete spear point, a tanged bone point and twenty-six awls are reported from M1 and M2 MSA layers at Blombos Cave, with ages ∼84–72 ka (Henshilwood and Sealy, 1997, Henshilwood et al., 2001, d'Errico and Henshilwood, 2007, Jacobs et al., 2006a, Jacobs et al., 2006b). A single massive point, different from those found in the MSA and LSA layers at Blombos Cave, was recovered in the dune sand layer, with an age of ∼70 ka (Jacobs et al., 2003, Jacobs et al., 2006a) sealing the MSA sequence. An awl and a possible flaked shaft fragment come from the Blombos M3 phase, with an age of 98.9 ± 4.5 ka (Jacobs et al., 2006a). A bone point from Peers (Skildergatkop) Cave was retrieved from either the HP or Still Bay layers at the site (Peers, 1929). A recent study of carbon-nitrogen ratios in the Peers point, and a sample of LSA and MSA faunal remains from this site, demonstrates that the point originates from MSA layers (d'Errico and Henshilwood, 2007). A single bone point (SAM-42160) was discovered at Klasies River in layer 19 of Shelter 1a at the base of the HP (Singer and Wymer, 1982). A date of approximately 80–60 ka, centred on 70 ka, was suggested for the HP at Klasies River (Deacon, 1989, Thackeray, 1989, Wurz, 2002, Deacon and Wurz, 2005). Uranium–Thorium (U-Th) analysis of limestone at the site produced an age range of 70–60 ka (Vogel, 2001), while Thermoluminescence (TL) and OSL ages of sediments suggest an age of 60–55 ka (Feathers, 2002). Thermoluminescence dating of burnt lithics provided a weighted mean age of 56 ± 3 ka for the HP at the site (Tribolo et al., 2005). The only other pointed bone implement known from the MSA is a polished bone pin (Cain, 2004) from layer Co, with an age of ∼35 ka attributed to the late MSA at Sibudu Cave (Wadley and Jacobs, 2006). Bone points similar to those known ethnographically occur at many LSA sites, including Oakhurst (Sampson, 1974, Deacon and Deacon, 1999), Rose Cottage (Sampson, 1974, Wadley, 1993, Wadley, 2000) and Nelson Bay Cave (Inskeep, 1987, Deacon, 1982, Deacon, 1984), and Jubilee (Wadley, 1989, Wadley, 1993), Bushman Rock (Plug, 1982) and Giant's Castle Rock Shelters (Sampson, 1974). Bone points are associated with the Border Cave Early LSA assemblage, with an age of ∼36 ka (Beaumont et al., 1978, Grün and Beaumont, 2001). Bone points used as awls, arrow points and linkshafts occur with regularity in Iron Age occupations, for example at Mapungubwe, Robberg Cave and the Vredefort Dome site. Bone points serving as arrow heads, linkshafts, barbs and hooks were in common use among Bushman hunter-gatherers in the nineteenth century, and even though metal is used nowadays for most armatures, the properties of bone for arrow point production are still favoured by some hunter-gatherers in southern Africa.

The bone implements described here come from the HP layers at Sibudu Cave, situated on the KwaZulu-Natal north coast of South Africa (Fig. 1). The site has revealed a long sequence of MSA occupations, with stone tool assemblages attributed, from the bottom to the top, to a pre-Still Bay phase, the Still Bay technocomplex, the Howiesons Poort technocomplex, a post-Howiesons Poort phase, and late and final MSA phases directly overlain by Iron Age occupation (Wadley and Jacobs, 2004, Wadley and Jacobs, 2006, Villa et al., 2005a, Wadley, 2005a, Wadley, 2007, Lombard, 2008). A combination of single aliquot and single grain Optically Stimulated Luminescence (OSL) ages for post-Howiesons Poort assemblages at Sibudu Cave provide a weighted mean age of 57.5 ± 1.4 ka (Jacobs et al., in press). Ages for the older HP industries fit within Oxygen Isotope Stage (OIS) 4, i.e. earlier than ∼61 ka (Wadley and Jacobs, 2006). The HP has been excavated from a 2 m² trial trench, and is found in layers PGS, GS2, GS, DRG2, DRG, WA, GR2 and GR (Table 1, Fig. 2). The polished spatula-shaped bone implement derives from near the top of the HP sequence in layer GR2. The refitted bone point comes from the underlying layer GS, and the slender incomplete point from layer PGS at the bottom of the sequence (Table 1).

The lithic assemblage in these HP layers is dominated by segments and other backed tools manufactured mostly from dolerite and hornfels (Delagnes et al., 2006). There is a relatively high proportion of blades and large sandstone flakes (Wadley, 2006). The extensive use of ochre at the site is evident in the high number of modified ochre nodules recovered (Wadley, 2005a) and lithics that have deeply ingrained ochre on their backed edges or bases. The evidence suggests that in addition to serving a potentially symbolic role, ochre was used as a loading agent in mastic recipes (Lombard, 2005a, Lombard, 2006, Lombard, 2008; Wadley, 2005b, Wadley, 2005c). Environmental data indicate a humid climate for the area during the HP, with surrounding evergreen forests (Allott, 2006, Glenny, 2006, Schiegl and Conard, 2006, Sievers, 2006, Renaut and Bamford, 2006, Wadley, 2006). Preliminary faunal analysis shows a wide range of mammal species represented, with blue duiker (Philantomba monticola) predominating (Wadley, 2006, Clark and Plug, in press, Lombard, 2008).

The Howiesons Poort is a MSA technocomplex characterised by the production of backed tools. Blades, backed segments and trapeze-shaped tools, of varying lengths, predominate. This archaeological entity is recorded at a number of well-dated sites in southern Africa, with long stratigraphic sequences that pre- and post-date the HP industry, including Klasies River, Border Cave, Rose Cottage, Peers Cave, Apollo 11, Diepkloof and Sibudu Cave. The recent discovery at the last two sites of Still Bay assemblages below the HP layers (Rigaud et al., 2006, Wadley, 2007) has finally settled the debate over the stratigraphic succession of these two MSA facies. Based on OSL ages of overlying layers, the HP at Sibudu Cave predates ∼61 ka and falls within the estimated age range for the HP, radiometrically dated at six sites to between ∼76 and 45 ka. Klasies River spans ∼66–45 ka (Vogel, 2001, Feathers, 2002, Tribolo et al., 2005). Rose Cottage Cave has single aliquot OSL ages of between 67.7 and 54.4 ka (Pienaar, 2006). An age of 62.4 ± 2.0 ka was obtained for Boomplaas based on the Uranium–Thorium method (Vogel, 2001). Electron Spin Resonance (ESR) ages of 76 ± 4 and 58 ± 2 ka (Grün and Beaumont, 2001, Grün et al., 2003) are reported for the HP at Border Cave, and Diepkloof has TL ages of between 65 and 55 ka (Rigaud et al., 2006). At Sibudu Cave, as at many other South African sites, the HP is replaced by about 60 ka (Villa et al., 2005b, Cochrane, 2006, Wadley and Jacobs, 2006, Jacobs et al., in press) with a flake technology characterised by a low standardisation of the end-products, many of which are retouched to obtain regular edges.

It is generally accepted that, by the time of the HP, and probably well before, people in southern Africa were anatomically modern (McBrearty and Brooks, 2000, Grün and Beaumont, 2001, Deacon and Wurz, 2005, Marean and Assefa, 2005, Brooks et al., 2006, Wadley, 2006, Lombard, 2008). This idea is based mostly on human remains from Klasies River (Rightmire, 1984, Rightmire and Deacon, 1991, Bräuer et al., 1992) and Border Cave (Beaumont et al., 1978, Rightmire, 1979, Rightmire, 1984, Beaumont, 1980) and to a lesser extent on the identification of modern features in human remains from Ethiopia, such as Herto (White et al., 2003) and Omo Kibish I (McDougall et al., 2005).