Navorsinge VAN DIE NASIONALE MUSEUM BLOEMFONTEIN

VOLUME 25, PART 2 MARCH 2009

THE ARCHAEOLOGY ASSEMBLAGE FROM THE 1958 EXCAVATION OF HOFFMAN’S/ROBBERG CAVE AND A COMPARISON WITH NELSON BAY CAVE

by

Katharine Kyriacou & Judith Sealy

NAVORSINGE VAN DIE NASIONALE MUSEUM, BLOEMFONTEIN is an accredited journal which publishes original research results. Manuscripts on topics related to the approved research disciplines of the Museum, and/or those based on study collections of the Museum, and/or studies undertaken in the Free State, will be considered. Submission of a manuscript will be taken to imply that the material is original and that no similar paper is being or will be submitted for publication elsewhere. Authors will bear full responsibility for the factual content of their publications and opinions expressed are those of the authors and not necessarily those of the National Museum. All contributions will be critically reviewed by at least two appropriate external referees. Contributions should be forwarded to: The Editor, Navorsinge, National Museum, P.O. Box 266, Bloemfontein, 9300, South Africa. Instructions to authors appear at the end of each volume or are available from the editor.

------

NAVORSINGE VAN DIE NASIONALE MUSEUM, BLOEMFONTEIN is 'n geakkrediteerde joernaal wat oorspronklike navorsing publiseer. Manuskripte wat erkende studierigtings van die Museum omsluit en/of wat op die studieversamelings van die Museum gebaseer is en/of wat handel oor studies wat in die Vrystaat onderneem is, sal oorweeg word. Voorlegging van 'n manuskrip impliseer dat die materiaal oorspronklik is en geen soortgelyke manuskrip elders voorgelê is of voorgelê sal word nie. Outeurs dra die volle verantwoordelikheid vir die feitelike inhoud van hulle publikasies en menings wat uitgespreek word, is dié van die outeurs en word nie noodwendig deur die Nasionale Museum onderskryf nie. Bydraes sal vir kritiese oorweging na ten minste twee geskikte buite- beoordelaars verwys word. Manuskripte vir publikasie moet voorgelê word aan: Die Redakteur, Navorsinge, Nasionale Museum, Posbus 266, Bloemfontein, 9300, Suid-Afrika. Voorskrifte aan outeurs verskyn afsonderlik aan die einde van 'n volume of kan van die redakteur verkry word.

ISSN 0067-928

NATURAL SCIENCES

VOLUME 25, PART 2 MARCH 2009

THE ARCHAEOLOGICAL ASSEMBLAGE FROM THE 1958 EXCAVATION OF HOFFMAN’S/ROBBERG CAVE AND A COMPARISON WITH NELSON BAY CAVE

by

KATHARINE KYRIACOU & JUDITH SEALY Department of Archaeology, University of Cape Town, Private Bag X3, Rondebosch, 7701 E-mail: [email protected] [email protected]

ABSTRACT

Kyriacou, K. & Sealy, J. 2009. The archaeological assemblage from the 1958 excavation of Hoffman's/Robberg cave and a comparison with Nelson Bay cave. Navors. nas. Mus., Bloemfontein 25(2): 49-72. This paper documents the archaeological finds excavated from the site of Hoffman’s/Robberg Cave by Dr A.C. Hoffman in 1958. Although this material was excavated almost fifty years ago, it has not previously been catalogued or described. The collection curated in the National Museum, Bloemfontein contains informal ‘Post- Wilton’ stone artefacts, mostly made from locally-available quartzite, as well as a range of bone artefacts. There are surprisingly few items manufactured from marine shell. Food remains include bone from seals, marine birds, fish, rock hyrax, bovids and other , as well as marine mollusc shells. We go on to compare this assemblage with that from contemporary (‘Post-Wilton’) levels from the nearby site of Nelson Bay Cave. Assessment of the Hoffman’s/Robberg Cave material is constrained by the fact that we are unsure of the extent to which the collection curated in the National Museum is truly representative of the site, given the unsystematic excavation practices employed by Hoffman and the possibility that he retained certain items from the site and discarded others. A solution to this problem would be to conduct further excavations at Hoffman’s/Robberg Cave in order to obtain an unselected sample. (Hoffman's /Robberg Cave, Nelson Bay Cave)

OPSOMMING

Die argeologiese versameling uit die 1958-opgrawings van die Hoffman/Robber-grot en 'n vergelyking met die Nelsonbaaigrot. Hierdie artikel beskryf die argeologiese voorwerpe wat by die Hoffman/Robberg-grot deur dr. A. C. Hoffman in 1958 opgegrawe is. Alhoewel die materiaal 50 jaar gelede opgegrawe is, is dit nog nie voorheen gekatalogiseer of beskryf nie. Die versameling is onder kuratorskap van die Nasionale Museum, Bloemfontein, en bestaan uit informele post-Wiltonklipwerktuie, meestal gemaak van kwartsiet uit die omgewing, sowel as ‘n reeks van been-artifakte. Daar is verrassend min items vervaardig van marieneskulpe. Voedseloorblyfsels word verteenwoordig deur die bene van robbe, seevoëls, visse, dassies, wildsbokke en ander diere, sowel as marieneskulpe. Verder het ons die versameling vergelyk met die post-Wiltonlae van die nabygeleë Nelsonbaaigrot. Evaluering van die Hoffman/Robberg-grot versameling is beperk weens die feit dat ons nie weet in watter mate die versameling wat in die Nasionale Museum is, verteenwoordigend is van die materiaal wat Hofmann deur onsistematiese opgrawingstegnieke bekom het nie. Sekerheid oor wat hy gehou en wat hy weggegooi het, is dus onduidelik. Om dus ‘n beter verteenwoordigende versameling te bewerkstellig, beteken dat daar in die toekoms verdere opgrawings by die Hofmann/Robberg-grot gedoen sal moet word. (Hoffman/Robberg-grot, Nelsonbaaigrot)

ISBN 1 86847 127 6 50 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

CONTENTS

INTRODUCTION ...... 50

Hoffman's/Robbeberg Cave ...... 50

Previous Research along the southern Cape coast ...... 51

DESCRIPTION OF THE ASSEMBLAGE FROM HOFFMAN’S/ROBBERG CAVE ...... 54

Method of Analysis ...... 54

The Hoffman’s/Robberg Cave Collection ...... 54

DISCUSSION ...... 57

INTERPRETATIONS ...... 66

Material culture and ethnic identity in archaeology: debates and approaches ...... 66

Material culture and identity at Hoffman’s/Robberg Cave ...... 68

CONCLUSIONS ...... 70

ACKNOWLEDGEMENTS ...... 70

REFERENCES ...... 70

INTRODUCTION

Hoffman’s/Robberg Cave

The site of Hoffman’s/Robberg Cave, also known as East Ghwanogat and Cave F, is located on the Robberg Peninsula west of Plettenberg Bay on the southern coast of South Africa. It is close to two intensively studied and well documented southern Cape coastal cave sites, namely Nelson Bay Cave and Matjes River Rock Shelter (Fig.1). Nelson Bay Cave provides one of the best records of human behaviour spanning the Middle and Later Stone Ages anywhere in the world. In 2006, the authors set out to examine and catalogue the collection of Late Holocene material from Hoffman’s/Robberg Cave curated in the National Museum, Bloemfontein. Derived from an excavation undertaken fifty years ago by the then director of the Museum, this material had been stored without having been documented or analyzed. In our study of this collection, we aimed to compare it with contemporary Late Holocene assemblages, notably that from Nelson Bay Cave, and in so doing, integrate the little studied site of Hoffman’s/Robberg Cave into the broader context of developments within forager societies living on the southern Cape coast during the Late Holocene.

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 51

Figure 1: Map showing location of Hoffman’s (Robberg) Cave and other important archaeological sites.

Previous Research along the southern Cape coast

Hoffman’s/Robberg Cave As part of a series of caves on the Robberg Peninsula, Hoffman’s/Robberg Cave was probably explored by a number of local residents, guano diggers, skeleton hunters and amateur collectors prior to the first systematic archaeological excavations. The first known investigation of the site was undertaken by the Rev W.G. Sharples and Mr. W. van Rooyen who, at the request of Dr Peringuey of the South African Museum, began excavating the caves on the Robberg Peninsula in 1917. They focused on the recovery of human remains and highly decorated or elaborate items of material culture including painted stones and bone artefacts. Excavations at the site referred to as Cave F, a large cave containing stalactites along the right wall, were begun in March of that year. A trench was dug near the stalactites, where the skeleton of a child was retrieved, and also a quartzite pebble decorated on one side with painted images of two black animals pursued by a human figure. The substantial shell midden deposit located beneath the stalactites yielded additional finds

52 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2 including a large number of bone points and some pieces of red ochre (Rudner & Rudner 1973).

The site was next excavated by Dr A.C. Hoffman in 1958. Very little information is available on these excavations, and no records pertaining to them have been found at the National Museum. The museum’s Annual Reports refer to the collection of samples from the site in 1952, and to site excursions in 1963 and 1964. Hoffman applied for and was granted a permit to excavate the site in 1958. The excavation was probably carried out that year. A trench measuring 1.5 by 5m was dug through approximately 2m of Later Stone Age shell midden deposit, and large quantities of material including human and faunal remains and a variety of artefacts were removed (Fairhall, Young & Erickson 1976). Certain components of the excavated finds were subsequently curated at the National Museum in Bloemfontein. In 1970, samples of shell were taken by Hilary Deacon and Richard Klein from the top and bottom levels of the midden (as exposed by Hoffman’s trench). These were submitted for radiocarbon dating, and yielded dates of 3190  110 BP (UW 204) and 3770  100 BP (UW 205) respectively (Fairhall, Young & Erickson 1976). When calibrated using the Pretoria radiocarbon calibration programme (Marine ’93), the most likely dates are 1011 BC and 1729 BC, with two-sigma ranges of 1327-787 and 1989-1491 BC respectively. The material cultural and other remains derived from the site thus represent a period of occupation by hunter-gatherer people that probably lasted less than a thousand years. Apart from the radiocarbon dates, nothing has been written or published about the material recovered from Hoffman’s/Robberg Cave.

Nelson Bay Cave and Matjes River Rock Shelter Nelson Bay Cave, situated approximately 300m to the west of Hoffman’s/Robberg Cave, was first excavated in 1964 by R.R. Inskeep. Inskeep focused on the younger deposits at the front of the cave, virtually blocking its entrance. He identified 148 separate stratigraphic units containing a wealth of archaeological material, the details of which are reported in his 1987 monograph on the site. These represent a period of more or less regular occupation of this cave by Later Stone Age hunter-gatherers during the second half of the Holocene. He also excavated a test trench further back in the cave, referred to as “Inskeep’s deep sounding”, which exposed additional deposits dating to the early Holocene and the Middle Stone Age. These were further explored in excavations led by Richard Klein (Klein 1972a,b). As a result of the efforts of both of these researchers, the Middle and Later Stone Age sequence at Nelson Bay Cave has been extensively studied and documented, making it one of the most informative cave sites in South Africa (Deacon 1984).

Matjes River Rock Shelter lies approximately 14 km east of Nelson Bay Cave. This extensive shell midden deposit was first excavated in 1928 and 1929 by the entomologist T.F. Dreyer, who dug a large trench along the back wall of the cave and removed substantial amounts of material, including numerous human remains. Excavations at Matjes River were resumed in the early 1950s under the leadership of Hoffman and A.J.D. Meiring (Deacon 1984). The methods employed in the course of these excavations and in subsequent interpretation and documentation were crude, and the monograph published a few years later by J.T. Louw was also of a poor standard (Inskeep 1960). Further excavations carried out in 1993 by Hilary Deacon and Willemien Döckel provided additional information on the previously poorly resolved stratigraphy of the site (Döckel 1998).

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 53

In the 1980s, one of us (Sealy) embarked on an extensive series of studies, based on the isotopic analysis of archaeological human skeletons, to reconstruct the diets and subsistence strategies of Later Stone Age hunter-gatherers in the Western Cape. Her aim has been to integrate biological and dietary information, accessed through analysis of the stable carbon and nitrogen isotopic composition of bone, with more traditional lines of archaeological and ethnographic evidence to provide a clearer view of the lifeways of prehistoric people. Her recent work has focused on Matjes River Rock shelter, Plettenberg Bay and the Robberg Peninsula, including Hoffman’s/Robberg Cave. Carbon and nitrogen isotope analysis of human remains dated to 4500 – 2000 BP shows that people buried in cave and open sites on the Robberg Peninsula and Plettenberg Bay ate diets rich in high trophic level marine protein. These foods probably included predatory fish, seabirds and Cape fur seals from the nearby Robberg seal colony, a “special foraging opportunity” for hunter-gatherers living in the vicinity (Sealy 2006: 578). Their contemporaries at the nearby site of Matjes River Rock Shelter consumed more mixed diets, relying to a greater extent on lower trophic level marine foods such as shellfish and on terrestrial resources. This amounts to a significant economic and therefore also social separation between these two groups of foragers. The most likely geographical boundary would have been the estuary of the Keurbooms/Bietou River. We can infer that, rather than forming part of widespread, mobile hunter-gatherer communities such as those in the arid Northern Cape and the Kalahari, these southern Cape hunter-gatherers occupied clearly demarcated geographical territories, sometimes with significant differences between the lifeways of neighbouring groups (Sealy 2006).

In 2003 and 2004, Ben Ludwig examined the excavated archaeological collections from the sites of Nelson Bay Cave and Matjes River Rock Shelter curated at the Iziko/South African Museum in Cape Town and the National Museum in Bloemfontein, respectively. His aim was to ascertain whether or not the economic and social differentiation inferred from the isotopic data is reflected in differences in material culture between the two sites (Ludwig 2005). If so, this would support the assertion that these groups had separate identities. Ludwig’s approach assumes that stylistic variation in material cultural objects is a product of deliberate design and manipulation by prehistoric people, who would have used such artefacts as communicable expressions of their own personal or group identity. The objects most likely to have served as so-called identity markers are those used in personal adornment or decoration. Variation in less highly visible features such as the presence or absence of backing on scrapers is better ascribed to adherence to different tool-making traditions. Ludwig observed a number of potentially significant differences between the two assemblages and attributed them to the existence of separate technological traditions and the active negotiation and expression of particular group identities (Ludwig 2005).

This argument would be considerably strengthened if consistent patterns in material culture and technological tradition could be demonstrated in additional assemblages from each side of the proposed geographical and cultural boundary. The material from Hoffman’s/Robberg Cave offers one such opportunity. Given the close proximity of this site to Nelson Bay Cave, it seems reasonable to expect that the hunter-gatherers living at these locations would have had very similar lifeways and material cultural traditions. In this paper, we describe the archaeological collection from Hoffman’s/Robberg Cave excavated in 1958 and curated in the National Museum, Bloemfontein. We then go on to compare it with the assemblage from Nelson Bay Cave, and assess the similarities and differences between the two.

54 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

DESCRIPTION OF THE ASSEMBLAGE FROM HOFFMAN’S/ROBBERG CAVE

Method of Analysis

We examined the entire assemblage excavated in 1958 from Hoffman’s/Robberg Cave. Our particular interest was in items of material culture that might have been used by the inhabitants of the site as expressions of identity and cultural affinity. To this end, we performed only basic faunal identifications (we did not have comparative specimens at hand), and none of the faunal or artefactual remains were examined under a microscope. In our documentation of the Hoffman’s/Robberg Cave collection, we have followed Deacon’s (1984) terminology in our description of the lithic assemblage, and Schweitzer’s (1979) and Inskeep’s (1987) definitions of bone and shell artefacts.

We also attempted, with the help of Museum staff, to locate Hoffman’s field notes or any other documentation relating to the site or the excavation. We found only the last field notebook (probably the second of two), containing a couple of lines saying that the team had packed up after the excavation and returned home. No other documents of any relevance could be found. Some details concerning the grid lay-out and stratigraphy of the site can be inferred from the labels on the bags of excavated materials. These were marked with the name of the site, followed by the “blok” or square of the grid in which the artefacts or other remains were found, followed by the depth measured in feet and inches. The labels therefore serve as a partial record of the progress of the excavation. Hoffman began by removing the top six inches of the deposit, referred to as “Boonste Laag” or “Oppervlak”. The only squares explicitly mentioned at this level are designated by the letters K, L, i and l (lower-case l, may mean the same as capital L). There is some inconsistency in the use of lowercase and uppercase letters. We cannot tell whether this is random or whether it was intended to be meaningful. Bags from the next six inches are labelled as coming from squares a,c,d,e,f and g. The first material curated from squares b, h and j is from a depth of 1½ feet, square M from a depth of 3½ feet, square N from four feet, O and P from 5½ feet, Q and R from six feet, and S and T from 6½ feet. Most of the squares referred to in the upper levels of the excavation are no longer mentioned once we reach the deeper levels. Squares labelled with letters from the later part of the alphabet continue to depths of between nine and a half and ten and a half feet, after which it appears that no more material was recovered. Two bags in Box 135 are labelled “6 vt, R (oppervlakte)” and “6 vt, Q (oppervlak)”. This is consistent with the possibility that Hoffman laid out his excavation grid over a downhill slope, but measured all of the depths from a single datum point. The outline of Hoffman’s trench is still visible at the site today, where it cross-cuts a steeply sloping area of deposit. Thus squares labelled a,b,c etc. probably lie towards the back of the cave, where the surface of the deposit is higher, and those labelled R,S,T etc. are likely to be closer to the mouth.

The Hoffman’s/Robberg Cave Collection

The collection of material from Hoffman’s/Robberg Cave includes large quantities of mostly informal lithic artefacts (Table 1) characteristic of Post-Wilton assemblages from sites along the southern Cape coast. It also contains a variety of beautifully manufactured bone artefacts (Table 2). Items manufactured out of marine shell and ostrich eggshell (Table 3) make up a very small component of the collection. Traces of ochre-staining are present on a wide range of both lithic and non-lithic materials (Table 4). The collection also

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 55 contains shells of a number of marine mollusc common along the southern Cape coast. These include Perna perna, Donax serra, argenvillei, Scutellastra barbara, Scutellastra cochlear, Scutellastra longicosta, Scutellastra tabularis, Cymbula oculus, Turbo sarmaticus, Oxystele sp. and spadicea. Faunal remains, including bones from numerous fish and bovid species, as well as seals, hyrax and freshwater turtle (Pelomedusa), to name a few, are also present but have not been tabulated. There are many compacted wads of the estuarine grass Zostera, thought to have been brought into the site as bedding or cushioning material. For a detailed catalogue of all finds, see Kyriacou (2006).

Table 1. Lithics Raw Material Category Total no. Quartzite Manuports Slabs 4 Pebbles 48 Cobbles 32 Waste Material Chunks 46 Cores 15 Unretouched flakes 78 Unretouched blades 15 Unretouched flakes/blades 3 Utilized Pieces Upper grindstones 51 Hammerstones 24 Upper grindstones/hammerstones 6 Milled-edge pebbles 3 Pebbles/cobbles with evidence of grinding 12 Pebbles/cobbles with hammerstone damage 29 Conjoining fragments of a bored stone 3 Perforated stones 1 Formal Tools “Fauresmith” hand-axes 1 Quartz Manuports Pebbles 1 Formal Tools Scraper 1 Ccs Waste Material Chunks 6 Unretouched flakes 1 Other Manuports Ochre 9 Sandstone 15 Sandstone Ground sandstone 11 Grooved stones 2 Shale Palettes 2 Sandstone Fragments of palettes 2

56 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

Table 2. Worked Bone Category Total No. Awls 21 Points 10 Hollow-tipped points 3 Spatulae 4 Beads/tubes (complete) 4 Beads/tubes (partial) 2 Ringed/snapped bone 7 Decorated/incised bone 4 Ringed/snapped and incised bone 1 Section of hippo tusk, split and smoothed 1 Flaked bone 1 Ground and smoothed bone 3 Ground freshwater turtle carapace 3 Bones with impact fractures 3 and 2 fragments Fish gorge 1 possible

Table 3. Worked Shell Category Total No. Perforated marine shell 2 Marine shell crescents 2 Incised ostrich eggshell 1

Table 4. Incidence of ochre-staining Category Total No. Lithics Quartzite spalls 2 Quartzite slabs 2 Quartzite fragments 1 Quartzite pebbles 4 Quartzite cobbles 2 Quartzite chunks 3 Quartzite cores 4 Quartzite flakes 8 Quartzite blades 1 Quartzite upper grindstones 17 Quartzite hammerstones 2 Quartzite upper grindstones/hammerstones 3 Quartzite milled-edge pebbles 1 Quartzite pebbles/cobbles with grinding facets 2 Quartzite pebbles/cobbles with hammerstone damage 1 Bone Bone points 1 Shell Limpets 3 Total ochre-stained items: 57

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 57

DISCUSSION

Assessment of this material can only be tentative, as we do not know to what extent the assemblage housed in the National Museum represents what was originally present at the site. In the absence of field notes it is difficult to reconstruct exactly what Hoffman and his team did, but it appears that field procedures were crude. The presence of many large stone artefacts, bones, etc. but few small ones clearly indicates that only a selection of the items recovered was kept, and we do not know the criteria used in this selection. One might expect that larger and more aesthetically pleasing objects were preferentially retained, as well as those thought to have been of special significance. These factors have probably skewed the representation of different categories of remains in the collection. Nevertheless, some broad patterns can still be discerned, and compared with those at the nearby site of Nelson Bay Cave. The levels of Nelson Bay Cave that are chronologically comparable to the occupation at Hoffman’s/Robberg Cave span units 31 – 64, with radiocarbon dates on charcoal of 2950 ± 80 (Pta-1485) and 3350 ± 60 BP (Pta-2910) respectively. When calibrated using the Pretoria radiocarbon calibration programme (Southern Hemisphere ’93), the most likely date for the younger radiocarbon determination is 1065 BC with a two- sigma range of 1363-1361 and 1314-889 BC. For the older determination,, the most likely date is 1590 or 1578 or 1541 BC, and the two sigma range is 1754-1415 BC. These figures are very similar to those reported above for Hoffman’s/Robberg Cave.

Lithic Artefacts Untrimmed quartzite flakes constitute the most numerous lithic artefacts throughout the sequence at Hoffman’s/Robberg Cave (Table 1). Stone artefacts are somewhat less frequent in the lower part of the deposit, below a depth of six feet. The assemblage includes a couple of old, heavily patinated flakes and blades probably dating to the Middle Stone Age, and one very heavily patinated, possibly Early Stone Age flake as well as a “Fauresmith” hand- axe. These may have been collected from the large open-air Early/Middle Stone Age site about 300 metres from the cave, close to the “island”. Cores are fairly frequent at Hoffman’s/Robberg Cave, and appear to be somewhat more numerous between three and four and a half feet than in the uppermost and lowermost levels of the excavation. In common with the majority of south coast Later Stone Age assemblages dating to the Post- Wilton, the lithic industry at Hoffman’s/Robberg Cave is highly informal: the single formal tool identified was a microlithic quartz scraper found at a depth of four and a half feet. In the comparable post-Wilton levels of Nelson Bay Cave, formal tools are rare and secondary retouch is much less common than in older units belonging to the Wilton cultural industry (Inskeep 1987). At Hoffman’s/Robberg Cave, as in the Post-Wilton layers of Nelson Bay Cave, Matjes River Rock Shelter and other Cape coastal sites, quartzite was clearly the preferred raw material. A single crypto-crystalline silicate (ccs) flake was found at a depth of seven and a half feet, and six ccs chunks were recovered nearby at a depth of five and a half feet.

In the utilized pieces category (Table 1), upper grindstones are by far the best represented items. Fifty-one upper grindstones were recovered from the site, plus six upper grindstones/hammerstones. A further twelve pebbles and cobbles showed slight signs of grinding. Upper grindstones were distributed relatively evenly throughout the deposit. Hammerstones are less numerous (n = 24, plus the six upper grindstones/hammerstones already mentioned, plus 29 pebbles or cobbles with minimal hammerstone damage), and were found scattered at irregular intervals in the different levels of the site. The assemblage

58 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2 also includes three “milled edged pebbles” (Deacon 1984: 377). The post-Wilton assemblage at Nelson Bay Cave is similarly characterized by an abundance of grindstones and what Inskeep refers to as “rubbing stones”. Rubber/hammerstones are apparently better represented in the older layers of the deposit, although “the total of grinding equipment is biased in favour of the upper group of units” (Inskeep 1987:105), that is, units 22-62, which partially correspond with the timespan for Hoffman’s/Robberg Cave. Inskeep tentatively suggested that larger numbers of rubber/hammerstones could be indicative of changes in the types of foods being processed.

Stone objects believed to have had symbolic significance and sometimes interred as “grave goods” include those referred to in the literature as palettes. These are typically flat pieces of stone, most commonly shale, which are ground at the edges to give them a roughly rounded or oval shape (Ludwig 2005: 39). Two shale palettes, one elongated, complete and undecorated, the other broken and highly decorated (Figure 2), are included in the Hoffman’s/Robberg Cave collection. They were found in the upper and upper-middle levels of the deposit. Ten specimens are listed in Inskeep’s inventory for Nelson Bay Cave. Four of these are of uncertain provenience; three derive from the post-Wilton units of the deposit; and the remaining three were recovered from two older layers within the sequence. A number of the palettes from Nelson Bay Cave as well as Matjes River Rock Shelter are perforated, but this is not the case for either of those from Hoffman’s/Robberg Cave. Both the Nelson Bay and Hoffman’s Cave assemblages also contain thin and often irregular tabular fragments of sandstone, and in the case of the former site, quartzite (Inskeep 1987: 103). At Nelson Bay Cave, these appear to have been confined to the mid-Holocene stratigraphic units pre-dating the Hoffman’s/Robberg Cave occupation by several thousands of years. At Hoffman’s/Robberg Cave, eleven were found scattered throughout the deposit. Several showed evidence of having been ground. Stone “sinkers” (small, rounded stones bearing visible grooves around their circumferences, possibly used as weights on fishing lines) were common at Nelson Bay Cave (138 specimens, mostly from the early post-Wilton). There are, however, none among the curated material from Hoffman’s/Robberg Cave. This is unlikely to indicate less emphasis upon fishing, and may therefore be the result of rough excavation practices.

Evidence of Ochre Usage Traces of ochre are evident on the surfaces of many items in the lithic assemblage (Table 4), including flakes, cores, chunks, and particularly upper grindstones. Of a total of 51 securely identified upper grindstones, 17 show evidence of often substantial ochre-staining. In many cases, ochre stains occur on one or both of the grinding facets, so they could have been used to grind ochre. Upper grindstones with evidence of ochre-staining derive from the upper and middle levels of the deposit, up to a depth of seven and a half feet. Ochre- stained flakes occur in the upper-middle, middle and lower-middle levels. These are considerably rarer than ochre-stained upper grindstones, with only eight out of 78 quartzite flakes bearing traces of ochre. Four flakes show ochre-staining on their ventral surfaces, while two are stained on the dorsal and ventral surfaces. Ochre-staining on flakes does not necessarily imply that ochre was deliberately applied: traces of ochre could have been transferred from the hands of people handling this substance onto the surfaces of these items. In the case of thickly smeared ochre such as that shown in Fig. 3, however, it is likely that the pigment was purposely applied. One large Scutellastra tabularis shell from the uppermost six inches of the deposit was clearly used as a container for ochre-based paint (Figure 4). Three-dimensional crusts of ochre are preserved inside the shell. Two

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 59 additional Patella shells with much lighter ochre-staining were found slightly lower in the deposit. Some ochre-staining on lithic artefacts was recorded by Inskeep at Nelson Bay Cave, but most of these items occurred in the lower stratigraphic units which are well beyond the temporal range of the Hoffman’s/Robberg Cave occupation. In units 22-62, only eight items with evidence of ochre-staining are listed, which seems to suggest that the use of ochre was considerably less common than at Hoffman’s/Robberg Cave. Inskeep also recovered a number of ochre “pencils” - a somewhat misleading term for elongated, heavily-ground pieces of ochre. These were more abundant in the middle levels of the deposit, from units 64 – 134, than in the upper and lower layers. Fragments of unworked ochre were regarded as more ambiguous, as the walls of Nelson Bay Cave were found to be naturally ochre-bearing. At Hoffman’s/Robberg Cave, a number of fragments of ochre were found in the upper, middle and lower-middle sections of the deposit, along with the numerous ochre-stained artefacts. No ochre “pencils” of the kind observed by Inskeep were found, and only one fragment had an obviously ground edge.

Bone artefacts The presence of a range of bone artefacts, some elaborately and extensively worked, has been identified as a characteristic of archaeological assemblages from Holocene south coast sites for a number of decades (Schweitzer 1976: 129). Sites that have yielded large assemblages of worked bone include Die Kelders, Nelson Bay Cave and Matjes River Rock Shelter. Hoffman’s/Robberg Cave is no different, with a collection that features a number of bone artefacts including awls, points and linkshafts, spatulae, beads and tubes, as well as miscellaneous engraved and decorated pieces (Table 2). The Hoffman’s/Robberg Cave bone artefact collection is similar to the Nelson Bay Cave assemblage dating between 3300 – 2000 BP. At Hoffman’s/Robberg Cave, awls are the most numerous bone artefacts (n=21), and appear to be distributed fairly evenly throughout the deposit. Awls are defined by Inskeep as asymmetrical artefacts that retain some unmodified surfaces, but are worked to point at one or both ends. They are believed to have been used in skin-working, for puncturing holes into hides which were then made into items including clothes and skin bags. Both Hoffman’s/Robberg Cave and Nelson Bay Cave yielded complete as well as broken specimens. At Nelson Bay Cave, Inskeep noted that awls were more common in the upper layers of the deposit, in stratigraphic units 22 – 62, with a wider range of different types of awls and a preference for more “robust cannon-bone awls” and those manufactured out of bird bones. Ludwig (2005) has suggested that the larger number of awls in the upper layers at Nelson Bay Cave may reflect spatial differentiation of activities in the site, if the majority of skin-working activities were carried out near the mouth of the cave, where it would have been warmer and the light would have been better. An alternative explanation may be that greater volumes of deposit were excavated from this part of the cave. Ludwig also proposed that the abundance of awls in the recent stratigraphic units at Nelson Bay Cave may also indicate more intensive “production and maintenance of clothing and other accessories” (Ludwig 2005: 43) during this time.

Bone points are slender, symmetrical, generally well-finished artefacts pointed at one or both ends and identified as projectile parts on the basis of analogy with surviving ethnographic specimens from the Cape and Kalahari (Inskeep 1987). These are less numerous in the Hoffman’s/Robberg Cave assemblage than are awls. A total of thirteen points were found on the surface and in the middle and lower-middle layers of deposit above a depth of seven and a half feet. Two specimens are shown in Figure 5. Three of the points recovered from the middle and lower-middle layers of the site have flat butts

60 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

Figure 2: Shale palette with incised grid design on both surfaces.

Figure 3: Snapped quartzite blade with ochre-staining on the ventral surface.

Figure 4: Outer and inner surfaces of large Scutellastra tabularis which contained ochre.

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 61

Figure 5: Double-ended bone points.

Figure 6: Two bone points with flat butts.

Figure 7: Bone point with narrow tang-like base.

Figure 8: Flaked bone artefact.

62 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

Figure 9: Bird bone with row of deep, regular incisions. Photographed from four sides.

Figure 10: Fragments of ground Pelomedusa carapace. Outer (left) and inner (right) surfaces, ground edges to the left.

Figure 11: Perforated valve of Lutraria lutraria. Inner and outer surfaces, and close-up of hole.

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 63

(Figure 6), and there is one stemmed specimen (Figure 7). Bone points were recovered in three of the four groups of stratigraphic units excavated by Inskeep at Nelson Bay Cave, in significantly smaller numbers than awls: stratigraphic units 22-62 yielded a total of fifty bone awls and only eleven bone points. The Hoffman’s/Robberg Cave collection contains slightly greater numbers of awls than points. Both the Hoffman’s/Robberg Cave assemblage and the comparable assemblage from Nelson Bay Cave include so-called hollow-tipped points manufactured out of the shafts of slender, hollow bird bones and given their characteristic shape by being “cut or snapped and ground obliquely at the tips to create a teardrop shape in [the] cavity at [the] pointed end” (Ludwig 2005: 44). These were found in small numbers at both sites and in the case of Hoffman’s/Robberg Cave, seem to have been confined to the upper and lower-middle sections of the deposit. Bone points and linkshafts have been interpreted as evidence for use of the bow and arrow, a hunting practice that has been widely inferred for the technologically sophisticated hunter-gatherers of the Holocene. The category “bone point”, however, probably encompasses artefacts with multiple uses, as an asymmetrical specimen such as that shown at the top of Figure 5 would not have functioned well as part of an arrow.

Another type of bone artefact included in both the Nelson Bay Cave and Hoffman’s/Robberg Cave assemblages is “spatulae”. This definition is applied to (usually) elongated, flat pieces with an asymmetrical, polished, wedge-shaped working edge, generally manufactured out of fragments of large limb bone or sections of rib. Their function remains somewhat speculative, although they are thought, like awls, to have been used in skin working (Inskeep 1987; Schweitzer 1979: 205) because of their smooth, polished working edges. The Hoffman’s/Robberg Cave collection includes only four of these artefacts, two found near the surface and two from the middle levels of the deposit. A much larger number of spatulae was recovered at Nelson Bay Cave, with twenty out of a total of twenty two specimens being derived from the later Holocene levels between units 22-62. If these items were indeed employed as skin-working tools, this would support suggestions of greater emphasis on hide-working after 3300 BP as hinted by the large number of awls. The rarity of spatulae in the Hoffman’s/Robberg Cave assemblage could be the result of failure to recognise them as artefacts due to hasty excavation procedures. An artefact designated by Hoffman as a “been saag” or bone saw, a fragment of bone flaked along one edge and with a burin-like tip (Figure 8), was also recovered from the surface of the deposit.

Eight bone tubes, manufactured from long bone shafts that were deliberately cut and snapped at one or both ends (Ludwig 2005), were recovered from units 22-62 at Nelson Bay Cave, within the last 3300 years. This artefact type was particularly characteristic of the later stages of occupation at this site. As with the greater concentration of awls in the upper layers of the deposit, this could provide clues to the layout of particular activity areas by the Later Stone Age inhabitants of the site. At the nearby site of Matjes River Rock Shelter, a much larger number of bone tubes were recovered, all of them from considerably earlier in the Holocene. At Hoffman’s/Robberg Cave, a bone tube was recovered from the lower levels of the deposit. Five additional specimens termed bone “beads” (in fact, short tubes) were loosely clustered in the upper-middle, middle and lower-middle sections of the deposit. Other pieces of ringed and/or snapped bone were recovered on the surface and in the upper, middle and lower levels of the deposit. Furthermore, eight pieces of worked bone identified as having been ground, scraped, incised, or otherwise modified, were recovered from the middle and lower levels of Hoffman’s/Robberg Cave. One of these, with

64 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2 numerous deeply incised grooves at short, regular intervals along the shaft of the bone (Figure 9), is comparable to a similarly notched specimen recovered from Bonteberg Shelter on the Cape Peninsula. Although the purpose of this artefact remains unknown, it has been proposed that it may have been drawn across the string to produce vibrations in a musical bow (Maggs and Speed 1967). The “teeth” of the notched artefact included in the Hoffman’s/Robberg Cave collection are rounded, which is consistent with its having been drawn back and forth across something. At Nelson Bay Cave, a number of decorated and incised pieces of bone were found in the Post-Wilton layers. These appear to have been more variable than those recovered by Hoffman. Bone rings, several examples of which were found in the stratigraphic units above 66 at Nelson Bay Cave, are completely absent from the Hoffman’s/Robberg Cave collection.

At Nelson Bay Cave, there is substantial evidence of pendants and bowls made from what Inskeep referred to as “tortoise” carapace. He furthermore suggested that since terrestrial tortoises are seldom to be found in the vicinity of the site, these items were imported as finished objects from regions to the north or east of Nelson Bay. It has subsequently been pointed out (by Royden Yates, pers. comm. to JS) that the bone is thinner and flatter than tortoise carapace, and probably comes from the fresh-water turtle Pelomedusa subrufa from the Piesang River or the Keurbooms/Bietou estuary. At Nelson Bay Cave, turtle shell was present only in unit 58 and above, post-dating 3300 BP. Ludwig noted that only two fragments of perforated turtle carapace were present in the collection from Matjes River, which was significant because it contradicted the general trend of greater variety in most categories of decorative artefacts from Matjes River, compared with Nelson Bay Cave. He suggested that the anomalous scarcity of turtle shell could result from poor excavation practice at Matjes River Rock Shelter. A similar explanation could be invoked to account for the rarity of Pelomedusa in the collection from Hoffman’s/Robberg Cave. Several fragments of turtle shell were found in the upper, middle and lower levels of the deposit. Only three from the upper levels show evidence of having been worked or ground. Two of these are shown in Figure 10).

Marine shell and ostrich egg-shell artefacts Artefacts made from marine shell and ostrich egg-shell are rare in the Hoffman’s Cave collection. There are two perforated marine shells: a large Patella tabularis roughly perforated at the , and a valve of Lutraria lutraria or otter shell, with a very smoothly drilled hole (Figure 11). Both were found unbroken in the surface layer of the deposit. On the basis of the curated material, it appears that Hoffman’s Cave yielded no shaped shell pendants or Glycymeris shells (perforated or unperforated). This is in sharp contrast to Nelson Bay Cave, where post-Wilton levels yielded almost all of the 202 Glycymeris shells (Ludwig 2005), as well as 18 Type 1 shaped shell pendants. These are made from large shells (probably Turbo sarmaticus) and are nearly all edge-nicked (Inskeep 1987 Plate 18). A further 32 examples were recovered from a burial in unit 65. Shell pendants are, therefore, something of a feature of the post-Wilton at Nelson Bay Cave. Their extreme rarity in the Hoffman’s/Robberg Cave collection raises the question of whether they were destroyed, or simply not recognized and therefore discarded, as a consequence of the excavation techniques employed by Hoffman and his co-workers. If this was not the case, their absence is indicative of a significant divergence between Hoffman’s/Robberg Cave and Nelson Bay Cave, which requires further investigation.

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 65

Shell crescents, that is, segment shaped artefacts of unknown function manufactured out of Perna perna shells, have been recovered from the mid- to late Holocene levels of a number of sites along the southern Cape coast including Nelson Bay Cave, Matjes River Rock Shelter and Die Kelders. Some researchers have doubted the artefactual status of shell crescents, suspecting that they are produced by natural breakage (Schweitzer 1979). Inskeep does not discuss them in his monograph on Nelson Bay Cave. Ludwig (2005) argued that they are as standardized in size and shape as mid-Holocene stone segments, and therefore unlikely to have been the result of accidental natural breakage. He regarded them as the products of intentional manufacture and design by the inhabitants of these sites. Some problems remain: we do not know which of the edges represents the working edge of the presumed tool, while the lack of any detectable mode of fixation suggests that they would have been impractical as items of personal adornment or ornamentation (Ludwig 2005). At Nelson Bay Cave, shell crescents are most numerous in the levels postdating 3300 BP, with the majority dating between 3020 – 2560 BP. The inverse of this pattern was observed in the Matjes River material, where the majority of shell crescents were recovered from Layer C, which pre-dates 3300 BP. Only two complete specimens and no incomplete or broken ones are present in the Hoffman’s/Robberg Cave collection. It is uncertain whether these were the only shell crescents recovered by Hoffman and his team, or if they are the only specimens which happen to be included in the collection.

The Hoffman’s/Robberg Cave collection includes two fragments of ostrich eggshell, one a burned, un-worked fragment derived from the middle section of the deposit and the other also burned, but decorated with incised lines. Since ostriches are arid environment birds which are not found in the vicinity of Plettenberg Bay today, these items were probably traded from some distance away. There were no ostrich egg-shell beads, probably because these are too small to have been recovered with Hoffman’s field techniques.

A number of differences and similarities between the curated material assemblages from Hoffman’s/Robberg Cave and Nelson Bay Cave can tentatively be noted. The post-Wilton lithic assemblages from both sites are highly informal, consisting mainly of unretouched quartzite flakes and utilized pieces. Both sites have yielded abundant grinding equipment, but there is more ochre-staining evident in the lithic assemblage from Hoffman’s/Robberg Cave. Decorated shale palettes are present, but rare in relation to Nelson Bay Cave. Bone awls are less common, and bone points more common at Hoffman’s/Robberg Cave compared with the contemporaneous levels of Nelson Bay Cave. This may be indicative of differences in the relative importance of skin-working and hunting activities between the two sites. In addition, the Hoffman’s/Robberg Cave collection includes relatively few bone spatulae, similar numbers of tubes/beads compared with Nelson Bay Cave, but no bone rings. Marine shell pendants and perforated shells are absent or very rare at Hoffman’s/Robberg Cave, which is surprising, given the wealth of these items recovered from Nelson Bay Cave. Items made from Pelomedusa carapace, too, are much rarer at Hoffman’s/Robberg Cave. It is, unfortunately, very difficult to tell whether these differences and similarities are in fact reflective of wider sociocultural relationships among the Later Stone Age inhabitants of these sites, or whether they are merely the result of excavation bias in the Hoffman’s/Robberg Cave collection.

66 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

INTERPRETATIONS

Material culture and ethnic identity in archaeology: debates and approaches

Ideas about the relationship between material cultural remains and ethnic or group identity have been central to the discipline of archaeology almost since its inception. The link between material culture and ethnicity is perhaps most explicit in the culture-historical approach which held sway up until the 1950s, and in which complete archaeological assemblages were equated with particular groups of people. The spatial and temporal distribution of discrete archaeological cultures and “units” was taken to be a reflection of the corresponding distribution of past peoples, and as material evidence whereby the interactions among prehistoric groups and the diffusion of particular cultural traits could be reconstructed through the analysis of artefactual and other remains (Lipo et al. 1997; Jones 1997). This approach has undergone significant revision in the light of numerous critiques and methodological and theoretical shifts that have occurred within archaeology and broader subject of anthropology over the last several decades. Nevertheless, the relationship between material culture and ethnicity remains a topic of considerable interest and controversy among archaeologists and anthropologists, who have continued to view this relationship as having great potential in informing their reconstructions of prehistoric lifeways and understanding of the role of material things in complex forms of human behaviour. Of particular importance in this regard is the “meaning” of stylistic and material cultural variation and the extent to which similarities and differences in and between archaeological assemblages can be interpreted as indications of prehistoric group relations, self-conceptualizations and interactions.

Hegmon (1992) and Jones (1997) have reviewed recent debates concerning the connection between ethnicity and material culture. These consist of four closely related components: the definition and nature of style, and where exactly in different objects of material culture style resides, as well as how style is related to ethnic and group identity and in what ways the investigation of style can be applied in archaeological studies of ethnicity and inter- group relations. The simplest and most widely accepted definition of the term “style” is that it represents a particular way of doing and manufacturing things that may be peculiar to certain groups of people in time and space (Sackett 1977; Hegmon 1992). This very general consensus among archaeologists ends here, however. There is considerable and often acrimonious disagreement over issues such as the properties and behavioural basis of style – for instance, whether style is an active or passive and more or less unconscious quality inherent in the formal or functional components of material culture – which of course has important implications for the interpretation of stylistic variation in archaeology (Hegmon 1992).

As a corrective to the earlier culture-historic approach, Binford (1965) proposed that culture is not necessarily shared; rather it is participated in, and participated in differentially, by the various individuals and “social units” that make up cultural systems. In this context, culture is viewed as a multivariate phenomenon consisting of a number of differentially articulated and integrated components or subsystems which together constitute the “means of extra- somatic adaptation” whereby human beings function within their physical and social environments. He rejected the simplistic notion of artefacts as “equal and comparable traits” whereby the levels of social and cultural distance between different groups of prehistoric people can be assessed by means of the analysis of broad differences and

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 67 similarities in their material cultural products. Instead, Binford distinguished three classes of material cultural remains, “technomic”, “socio-technic” and “ideotechnic”, on the basis of the supposedly distinctive social-cultural domains in which they routinely function. He furthermore asserted that certain formal qualities which cannot be accounted for in terms of functional explanations effectively cross-cut these different functional categories. These stylistic attributes arise as a product of the enculturation process and may in some instances “become standardized symbols” appropriate to artefacts “used in specific social contexts” (Binford 1965: 206); they nevertheless play a functional role in promoting group solidarity and providing a basis for group awareness and identity (Binford 1965; Jones 1997). Among the attributes frequently analysed in this way are the often culture-, time- and place-specific decorative motifs applied to archaeological ceramic vessels.

James Sackett’s views of the interpretation of assemblage variability in archaeology differed significantly from those of Binford. He rejected the strict dichotomy between formal, stylistic or “adjunctive” variation and functional or utilitarian variation, and regarded them as essentially complementary. He coined the term isochrestism, derived from the Greek “equivalent in use”, to describe the formal and/or functional attributes of material cultural objects that arise as a result of these objects being made and manufactured in more or less consistent and socially prescribed ways by artisans who have undergone similar enculturation processes, notwithstanding the existence of alternative, equally viable options. In this case, style becomes a passive, or rather a latent property of material culture which, although it is often symbolically rich and ethnically-loaded, is not deliberately intended to signal group identity. In the main, material cultural objects tend to conform with and reflect standard patterns and practices of manufacture within the ethnically bounded contexts in which they are produced (Sackett 1982, 1986; Jones 1997). This is especially true of lithic artefacts, which can be grouped into distinct regional or local stone tool technologies on the basis of preferred knapping and secondary retouch techniques and raw material utilization. Sackett pointedly contrasted isochrestic variation with what he referred to as iconic or active style, which involves the self-conscious and intentional expression and signalling of ethnic identity. The latter type of style, according to Sackett, is by far the more unusual of the two, with the majority of formal and especially functional variation in material culture falling within the category of isochrestic variation (Sackett 1986).

The so-called “iconic approach” is evident in the work of Wiessner (1984), who, following Hodder and Wobst, perceived style as a form of assertive, nonverbal communication by means of which important information regarding ethnic and/or personal identity is visually transmitted to members within and between social groups. In this case, style, if not in all then in a significant number of instances, acquires an explicit behavioural dimension in the cognitive process whereby people seek to define themselves both as individuals and as members of particular social and ethnic groups by means of self-reflexive comparisons with other people and other groups. Rather than constituting a passive reflection of ethnicity or culture, so-called “emblemic” style is perceived as an active and effective tool for the articulation of personal and group identity and as a strategic resource for the negotiation and manipulation of social relationships and boundaries. Furthermore, the nature and intensity of this ethnic signalling may change with fluctuating socio-economic conditions, in terms of which social and ethnic boundaries are either maintained or re-negotiated (Wiessner 1984, 1985; Hegmon 1992; Jones 1997).

68 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

In order to demonstrate her assertions about the role of style in the negotiation and expression of identity, Wiessner investigated stylistic variation in two important components of the material culture of the Kalahari San of Botswana and Namibia, namely metal-tipped arrows manufactured and used by men and decorated headbands produced and exchanged by women. The latter items in particular met Wiessner’s proposed criteria for artefacts suitable as identity or ethnic markers: they combine a long use-life with high visibility in a number of different socio-economic activities and domains. Furthermore, a significant amount of time and effort is devoted to their manufacture, which provides artisans with abundant scope for stylistic expression. Different degrees of stylistic variation in beaded headbands were observed in and related to different contexts and spheres of social relations, including within and between bands, language families and regions (Wiessner 1984). Wiessner (1985) also identified instances, often associated with periods of ecological stress, where the purposive expression of ethnic or group identity was deliberately held in check in order to promote solidarity and cooperation rather than individuality and differentiation. The similar use of certain cultural products by prehistoric populations as self-conscious expressions of identity is likely to have produced discontinuities in the distribution of these items in the archaeological record. The theoretical position concerning artefacts and ethnic identity taken by Wiessner and a number of other researchers seems to encourage archaeological applications in terms of which the presence or absence of certain items of material culture, and stylistic variation within artefact types can be investigated to gain insight into intergroup relations among past peoples. This does not imply, however, that similarities and differences in material cultural assemblages should be simplistically interpreted as indexical measures of social distance or relatedness (Wiessner 1984; Hegmon 1992; Jones 1997).

The so-called “practice theory” of ethnicity and material culture articulated by Sian Jones represents something of a compromise between the active and passive views of style expressed by Wiessner and Sackett, respectively. Jones’ ideas on human agency and ethnic consciousness are grounded in Bourdieu’s concept of habitus, which consists not of abstract or normative mental rules or structuring principles, but of the “durable dispositions towards certain perceptions and practices which become part of people’s sense of self at an early age” and serve to structure their choices, actions and decisions and to shape their experiences of their particular social world (Jones 1997; Baumann 2004). In this context, material culture is seen as an “active constitutive dimension of social practice” which simultaneously structures, and is structured by, human agency. Stylistic variation in material cultural components is therefore not merely a largely unintentional product of the enculturation process as implied in Sackett’s isochrestic variation; neither is it always the result of deliberate ethnic signalling on the part of particular ethnic groups as proposed by Wiessner. Jones’ definition of material culture as actively structured and structuring appears to lend itself once again to archaeological applications and interpretations. Nevertheless, like other researchers before her, she urges caution with regard to the identification of ethnic groups and social boundaries in the archaeological record on the basis of observed continuities or discontinuities in material cultural remains (Jones 1997).

Material culture and identity at Hoffman’s/Robberg Cave

In the case of Hoffman’s/Robberg Cave, the existence of an economic difference between the Later Stone Age inhabitants of the Robberg/Plettenberg Bay area, including Nelson Bay Cave, and Matjes River Rock Shelter has already been established through the results of

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 69

Sealy’s isotopic research. We can infer that this translates into the existence of a social boundary between the two. Hoffman’s/Robberg Cave is located less than a kilometre from Nelson Bay Cave, on the same side of the proposed geographical boundary of the Keurbooms/Bietou estuary. The later Holocene occupants of Hoffman’s/Robberg Cave and Nelson Bay Cave were near neighbours and would frequently have encountered one another in the course of their subsistence and social activities. They may even have been the same group of people, camping at different times in different shelters on the Robberg Peninsula. Judging from previous works concerning material culture and ethnicity, it follows that if this had been the case, the awareness of a common ethnic identity may have been expressed at either a conscious or unconscious level in the form of certain items which may have been preserved in the archaeological record.

The relative abundance of grinding and skin-working equipment at the two sites could be indicative of varying economic or social importance of these activities to the inhabitants of Hoffman’s/Robberg Cave and Nelson Bay Cave. The more frequent occurrence of ochre- staining in the Hoffman’s/Robberg Cave collection (on 33.3 % of the upper grindstones, and to a lesser extent on other artefacts such as quartzite flakes), may suggest that this substance had considerable significance for the occupants of the site in relation to their neighbours at Nelson Bay Cave, among whom the use of ochre appears to have declined during the Post-Wilton. The association between ochre-staining and symbolism and spirituality seems to be of considerable antiquity among hunter-gatherers in South Africa, with foundations extending back into the Middle Stone Age. The occurrence of painted images on portable stones, and association of these and other items with graves, is particularly marked in the southern and south-eastern Cape during the Later Stone Age (Lewis-Williams and Pearce 2004). The rarity or absence of other utilitarian and ornamental objects, for example stone sinkers, marine shell and turtle carapace pendants, and bone rings from the Hoffman’s/Robberg Cave collection could be the result of sampling bias, or it could imply that the occupants of this site did not manufacture them or participate in the activities or behaviours with which they are associated. This would be surprising, given the quantities of these items in the Post-Wilton assemblage from Nelson Bay Cave, and may have significant implications for an analysis of group identity.

We observed some broad similarities, as well as a number of glaring discrepancies, in the lithic, bone and shell components of the Nelson Bay Cave and Hoffman’s/Robberg Cave assemblages. The lithic assemblages of the two sites are relatively similar. Both are highly informal, with an emphasis on unretouched quartzite flakes and grinding equipment. However, the Hoffman’s/Robberg Cave collection lacks stone sinkers, which are common at Nelson Bay Cave. Ochre-staining on various stone items, particularly upper grindstones and flakes, appears to be more frequent in the Hoffman’s/Robberg Cave material than at Nelson Bay Cave. With regard to lithic objects of supposed symbolic or ritual significance, two shale palettes, but no quartz crystals are included in the Hoffman’s/Robberg Cave collection. The bone assemblage from Hoffman’s/Robberg Cave contains fewer awls, and considerably more points, than the Post-Wilton assemblage from Nelson Bay Cave. The Hoffman’s/Robberg Cave collection also includes bone tubes/beads as well as decorated and incised pieces in similar proportions to those recovered from Nelson Bay Cave. However, none of the bone rings found at the latter site and described by Inskeep are present among the curated material from Hoffman’s/Robberg Cave. These differences could be the result of unsystematic fieldwork practices and/or excavator bias on the part of

70 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

Hoffman and his team. Alternatively, they could imply that specific artefacts absent from or occurring in very small numbers in the Hoffman’s/Robberg Cave collection were manufactured and used rarely, or not at all, by the occupants of the site. The marine and other shell assemblage is the cause for most of our concern regarding the completeness of the Hoffman’s/Robberg Cave collection. Perforated Glycymeris shells, and perforated and shaped shell pendants are completely absent from the curated material from the site, although they were recovered in significant numbers from the nearby site of Nelson Bay Cave. Only two shell crescents are present in the Hoffman’s/Robberg Cave collection, in comparison with the two hundred and forty specimens recovered from Nelson Bay Cave. It is logical to deduce that these items would have been easy to miss by a team of excavators using rather crude fieldwork techniques.

CONCLUSIONS

Our interpretation of the apparent similarities and differences between the archaeological assemblages from Hoffman’s/Robberg Cave and Nelson Bay Cave is constrained by our doubts as to whether the curated material from the former site is at least reasonably representative of the range of cultural items manufactured by its prehistoric inhabitants. We have identified some broad similarities in the two assemblages, as well as a number of differences. The differences may be the result of the excavation practices employed by Hoffman and his team, or alternatively may reflect significant socio-cultural differences between the occupants of the two sites, contrary to our expectations. This would, however, be surprising, given the likelihood that the late Holocene inhabitants of Hoffman’s/Robberg Cave and Nelson Bay Cave had very similar lifeways and in effect formed part of the same group. A possible solution to this dilemma would be to re-excavate the site of Hoffman’s/Robberg Cave in accordance with standard contemporary excavation practices. A comprehensive comparison of the material cultural remains from the two sites could then be confidently undertaken. Finally, we believe that the study of similarities and differences in material cultural assemblages as indications of the conscious or subconscious expression of identities has considerable potential in South African Later Stone Age studies. This is particularly so along the southern Cape coast, characterised by rich assemblages consisting of a variety of material cultural remains which lend themselves to analyses of style.

ACKNOWLEDGEMENTS

We would like to express deep appreciation to the staff of the National Museum in Bloemfontein for granting us access to the collection and facilitating our work. Members of the Archaeology Department, especially Zoë Henderson and Gerda Coetzee, provided space in which to work and assisted us in many ways. James Brink answered our questions about the faunal remains, and identified the hippo ivory. Our visit to Bloemfontein was funded by the National Research Foundation and University of Cape Town. Finally we thank the reviewers Dr Bronwen van Doornum (Department of Human Sciences, Natal Museum, Pietermaritzburg) and Dr Johan Binneman (Department of Archaeology, Albany Museum, Grahamstown) for their valuable comments.

REFERENCES

BAUMANN, T. 2004. Defining ethnicity. The SAA Archaeological Record 12-14.

BINFORD, L.R. 1962. Archaeology as anthropology. American Antiquity 28: 217-225.

BINFORD, L.R. 1965. Archaeological systematics and the study of culture process. American Antiquity 31: 203- 210.

2009 Kyriacou K. & Sealy, J. – Hoffman's/Robberg Cave 71

DEACON, J. 1984. The Later Stone Age of Southernmost Africa. Oxford: British Archaeological Reports International Series 213.

FAIRHALL, A.W., YOUNG, A.W. & ERICKSON, J.L. 1976. University of Washington Dates IV. Radiocarbon 18: 221-239.

HEGMON, M. 1992. Archaeological research on style. Annual Review of Anthropology 21: 517-536.

INSKEEP, R.R. 1960. Review of J.T. Louw’s The Prehistory of the Matjes River Rock Shelter, Part II, The archaeology. South African Archaeological Bulletin 16: 29-31.

INSKEEP, R.R. 1987. Nelson Bay Cave, Cape Province, South Africa: The Holocene Levels. Oxford: British Archaeological Reports International Series 357 (i) & (ii).

JONES, S. 1997. The archaeology of ethnicity: constructing identities in the past and present. London: Routledge.

KLEIN, R.G. 1972a. Preliminary report on the July through September 1970 excavations at Nelson Bay Cave, Plettenberg Bay (Cape Province, South Africa). In: Van Zinderen Bakker, E.M. (ed.) Palaeoecology of Africa, the surrounding islands and Antarctica. Cape Town: Balkema.

KLEIN, R.G. 1972b. The Late Quaternary mammalian fauna of Nelson Bay Cave (Cape Province, South Africa): its implications for megafaunal extinctions and environmental and cultural change. Quaternary Research 2:135-142.

KYRIACOU, K. 2006. The material cultural assemblage from Hoffman’s/Robberg Cave – a comparison with Nelson Bay Cave. Unpublished BA (Hons) thesis, Department of Archaeology, University of Cape Town.

LEWIS-WILLIAMS, J.D. & PEARCE, D.G. 2004. San spirituality: roots, expressions and social consequences. Cape Town: Double Storey Books.

LIPO, C.P., MADSEN, M.E. & DUNNELL, R.C. 1997. Population structure, cultural transmission and frequency seriation. Journal of Archaeological Anthropology 16: 301-333.

LUDWIG, B. 2004. A comparison of hunter-gatherer material culture from Matjes River Rock Shelter and Nelson Bay Cave. Unpublished M.Sc. Thesis, University of Cape Town.

MAGGS, T. & SPEED, E. 1967. Bonteberg Shelter. South African Archaeological Bulletin 22:87: 80-102.

RUDNER, J. & RUDNER, I 1973. A note on early excavations at Robberg. South African Archaeological Bulletin 28: 94-96.

SACKETT, J.R. 1982. Approaches to style in lithic archaeology. Journal of Archaeological Anthropology 1: 59- 112.

SACKETT, J.R. 1985. Style and ethnicity in the Kalahari: a reply to Wiessner. American Antiquity 50: 154-159

SACKETT, J.R. 1986. Isochrestism and style: a clarification. Journal of Anthropological Archaeology 5: 266-277.

SCHWEITZER, F.R. 1979. Excavations at Die Kelders, Cape Province, South Africa: the Holocene deposits. Annals of the South African Museum 78: 101-233.

SEALY, J. 2006. Diet, mobility and settlement pattern among Holocene hunter-gatherers in southernmost Africa. Current Anthropology 47: 569-595.

WIESSNER, P. 1984. Reconsidering the behavioural basis for style: a case study among the Kalahari San. Journal of Anthropological Archaeology 3: 190-234.

WIESSNER, P. 1985. Style or isochrestic variation: a reply to Sackett. American Antiquity 50: 160-166.

72 Navors. nas. Mus., Bloemfontein, Volume 25, Part 2

EDITORIAL COMMITTEE

Editor: J. Haasbroek, D.Phil. (UOFS); Co-editors (Natural Sciences): R.J. Nuttall, M.Sc. (Natal); P.C. Zietsman, Ph.D. (UOFS).

Consulting Editors: Prof. C. Chimimba (Department of Zoology and Entomology, University of Pretoria, South Africa); Dr J. Deacon (South African Heritage Resources Agency, Cape Town, South Africa – retired); Dr A. Dippenaar-Schoeman (ARC – Plant Protection Research Institute, Pretoria, South Africa); Dr A. Kemp (Northern Flagship Institution, Museum of Natural History, Pretoria, South Africa – retired); Dr D.T. Rowe-Rowe (Natal Parks Board, Pietermaritzburg, South Africa – retired); Prof. B.S. Rubidge (Bernard Price Institution for Palaeontological Research, University of the Witwatersrand, Johannesburg, South Africa); Prof. A.E. van Wyk (Department of Botany, University of Pretoria, South Africa); Prof. A. Wessels (Department of History, University of the Free State, Bloemfontein, South Africa).

Orders to: National Museum, P.O. Box 266, Bloemfontein 9300, Republic of South Africa. E-mail: [email protected]

VOLUME 23 2007

Part 1: Lotz, L.N. The genus Cheiracanthium (Araneae: Miturgidae) in the Afrotropical Region. 1. Revision of known species ...... 1

Deel 2: Haasbroek, J. Die gebeure wat gelei het tot die bedanking van die Bloemfonteinse stadsraad in 1907 ...... 77

Part 3: Coetzee, L. The genus Afroleius Mahunka (Acari, Oribatida, Haplozetidae) I. Redescriptions of A. deformis, A. minor and A. simplex ...... 101

Deel 4: Havenga, S. & Wessels, A. Die veranderde silhoeët van bruidsuitrustings, 1830-1930 ...... 121

Part 5: Lotz, L.N. The genus Cheiracanthium (Araneae: Miturgidae) in the Afrotropical Region. 2. Description of new species ...... 145

Part 6: Bates, M.F. First records of the Cape Girdled Lizard, Cordylus cordylus (Linnaeus, 1758), in Lesotho ...... 185

VOLUME 24 2008

Grobbelaar, J.U. A century of Botany at the University of the Free State ...... I

Part 1: Coetzee, J.A. Reminiscences of palynological research during early years in the Department of Botany, University of the Free State ...... 1

Part 2: Grobbelaar, J.U. Algal biotechnology: Four decades of research at the University of the Free State, moving from laboratory to commercial scale ...... 9

Part 3: Groenewald, E.G. Plant tissue culture research at the Department of Plant Sciences, University of the Free State ...... 21

Part 4: Mohase, L. Plant activators in disease management ...... 29

Part 5: Potgieter, G.P. & Pretorius, J.C. Control of reserve mobilization in seed cotyledons: Review of research done on Cucumis sativus L. at the University of the Free State ...... 37

Part 6: Scholtz, O.F. & Potgieter, G.P. An Overview: Phytoextraction as a Rehabilitation Strategy for Metal Polluted Soils ...... 49

Part 7: Scott, L. & Smith, V.R. Vegetation and peat development on Marion Island, Southern Ocean, during the Late Holocene ...... 61

Part 8: Van der Westhuizen, A.J. The biochemistry of resistance in wheat to the Russian wheat aphid: review of research done at Botany, Department of Plant Sciences, University of the Free State ...... 71

Part 9: Van Wyk, P.W.J., Kock, J.L.F. & Wingfield, M.J. Ascospore shape and development: a new frontier of mycology ...... 81

Part 10: Venter, A.M. of Lycium in Africa: a historical review ...... 91

Part 11: Verhoeven, R.L. Evolution from single pollen grains to tetrads and pollinia in Apocynaceae ...... 105

JOURNAL OF THE NATIONAL MUSEUM, BLOEMFONTEIN *

* WORLD LIST ABBREVIATION: Navors. nas. Mus., Bloemfontein

C U I RA A TOR D