Stone Age economics and land use in the Geelbek Dunes

The Archaeology of the West Coast of South Africa

Edited by

Antonieta JerardinoAntonia Malan David Braun

BAR International Series 25262013

Cambridge Monographs in African Archaeology84

Series Editors: Laurence Smith, Brian Stewart and Stephanie Wynne-Jone

Published by

ArchaeopressPublishers of British Archaeological ReportsGordon House276 Banbury RoadOxford OX2 [email protected]

BAR S2526Cambridge Monographs in African Archaeology 84


© Archaeopress and the individual authors 2013

ISBN 978 1 4073 1144 9

Printed in England by 4edge, Hockley

All BAR titles are available from:

Hadrian Books Ltd122 Banbury RoadOxfordOX2 7BPEnglandwww.hadrianbooks.co.uk

The current BAR catalogue with details of all titles in print, prices and means of payment is available free from Hadrian Books or may be downloaded from www.archaeopress.com

24

Chapter 2

Stone Age economics and land use in the Geelbek Dunes

Andrew W. Kandel & Nicholas J. Conard

Introduction In aerial and satellite photographs of South Africa, the dune field of Geelbek is an easily recognizable feature of the landscape (Fig. 1). The dunes appear as a sea of white sand emanating from the Atlantic Coast and moving inland along an elongated plume. A person on the ground heading north or south along the West Coast would have observed these hills of sand whether travelling by foot, ox cart, carriage or automobile. A recent traveller drawn to the dunes might have ventured a further look into this forbidding landscape. Once inside these sand formations, our traveller would have a hard time overlooking numerous fossil occurrences exposed within the broad linear valleys between the dunes. Questions about the origins of these fossil bones would undoubtedly arise. Furthermore, an astute individual could not fail to notice numerous concentrations of stone artefacts, pottery and marine shell scattered amongst the fossil bones. In fact, they might have recognised these occurrences as ancient in nature, and not random, but rather structured in their distribution. To the trained eye, these remains tell a story about South Africa’s Stone Age. These initial impressions are what motivated a team from the University of Tübingen to study the landscape of the Geelbek Dunes. We named the program the Geelbek and Anyskop Archaeological Survey Project (GAASP) and conceived of it as a way to study low-density occurrences of artefacts and fossils in an open-air setting. Starting in 1998, a joint German-South African research team from the Universities of Tübingen and Cape Town and the Iziko South African Museum began a systematic study of the natural processes occurring in the dune fields of Geelbek. The focus of GAASP was to explain the deposition of the archaeological and paleontological remains in this open-air, near-coastal setting about 5 km from the Atlantic Ocean. The research strategy aimed to sample the largest possible area. A further goal was to place archaeological and paleontological remains in an ecological context that would help explain the evolution of the West Coast landscape from the Late Pleistocene through the Holocene. Experts from a variety of scientific disciplines participated, coming mainly from German and South African institutions. Together, this interdisciplinary team implemented an integrative approach to understand the palaeo-landscape in which humans lived and its response to the climatic changes of the last 150 000 years (e.g. Felix-Henningsen et al. 2003, Fuchs et al. 2008).

For the next 10 years, field and laboratory research focused on cataloguing and analyzing the collections, documenting the results in more than 30 scientific publications and reports, and presenting the information at dozens of lectures at universities, museums, conferences and public meetings. In this chapter, we summarise the most salient results from our studies of the Geelbek Dunes to offer the reader a case study of how synthetic landscape studies can be used to understand settlement dynamics on a regional scale. Many of the questions we address relate to questions of geographic and chronological patterning of archaeological materials. For example, how can information about the horizontal distribution of finds on a surface become transformed into discrete chronological units? Which characteristics of artefacts and faunal material inform us about time scales to help distinguish between different Middle Stone Age (MSA) and Later Stone Age (LSA) assemblages? What role did the Pleistocene climate have on the relationship between the marine and terrestrial environments? And perhaps of greatest importance, can we create a framework that helps us model patterns that show how past people used this near-coastal environment? Background The Geelbek Dunes are situated 85 km north of Cape Town in the West Coast National Park at the south-eastern end of Langebaan Lagoon (Fig. 2). The park straddles the boundary between the marine and terrestrial environments that surrounds the lagoon, offering sanctuary to hundreds of species of migrating birds, as well as supporting a productive ecosystem. In keeping with previous studies (Conard & Kandel 2006), we define the coastal zone as the area along the seashore bounded by coastal dunes that are relics of the changing shoreline, while the near-coastal zone extends up to 10 km inland from the coastal dunes. This definition is based on the distance that foraging humans might cover in one day. Areas further than 10 km from the coastline are considered inland. As the coastline responded to changing sea levels of the past, these zones would have shifted accordingly Environment The near-coastal setting of the West Coast National Park is characterised by its strandveld vegetation, a type of small-leafed, chaparral-like, plant community adapted

Chapter 2: Kandel et al., Stone Age economics and land use in the Geelbek Dunes

25

Figure 1: Reference map of the West Coast of South Africa (left) and space image of Langebaan Lagoon region (right) showing the

location of nearby archaeological and paleontological sites (map after S.J. Walker; space image courtesy of the Image Science & Analysis Laboratory, NASA Johnson Space Center, http://eol.jsc.nasa.gov, ISS006-E-37012, 2003).

Figure 2: Map showing the geographic setting and distribution of the 23 investigated localities in the dune field (maps S.J. Walker).


26

to the nutrient-poor, loose, sandy soils along the coast (Acocks 1988). The creation of the park in 1991 restricted general access to the area, helping to preserve the integrity of the archaeological and paleontological localities, while strengthening their protection. The park’s cultural heritage extends from the MSA through the LSA and even into the European contact period. The Geelbek Dunes consist of a 2 km by 1.5 km island of modern sand surrounded by the indigenous strandveld vegetation. The modern dune field can be divided into two main areas, a large northern lobe and a small southern lobe connected by a narrow neck of sand. The sand is predominantly fine grained and very well sorted. Two distinctions in colour are apparent. Most of the sand is yellowish white and contains a significant amount (23-36%) of carbonate derived from marine shell. A darker, reddish-brown hue is visible in parts of the dunes and contains noticeably less carbonate (8–13%) (Felix-Henningsen et al. 2003). The movement of the modern sand dunes and its accretion into transverse dunes creates deep valleys between the crests. Erosion is significant in these valleys, which are known as deflation hollows or bays. The modern dune sand erodes away easily. Underlying, older sand layers are also quickly eroded by the abrasive action of the wind. Even massive calcrete horizons slowly wear away. As this process occurs, artefacts and fossils erode from the strata and deflate onto a single surface. These deflation surfaces expose archaeological and paleontological materials. Survey The starting point of the research program in 1998 was a pedestrian survey of the Geelbek Dunes. We measured the perimeter of the dune system at 100 m intervals and recorded the centre point of each of 114 deflation hollows using a Garmin GPS 12XL to create a detailed local map. We ranked the localities based on the presence of archaeological finds, paleontological remains and geological features. Our aim was to document sites with varying find densities and to examine the relationship between the archaeological and paleontological assemblages and the different geological contexts in which they were found (Conard et al. 1999). Each bay was ascribed a consecutive number, but when detailed investigations occurred at a locality, we assigned it a name based either on characteristic finds (e.g. named Pottery, Rhino), or an event (Check and Toaster) or person (Alice, Frikkie) that had relevance at the time we began work. Over the course of the next five field seasons, we investigated 23 of the 114 localities in detail and continued collecting finds until 2007 (Fig. 2). Climate The climate at Geelbek is arid to semi-arid, and the transverse dunes are a response to intense summer winds coupled with a low moisture regime. Most of the annual rainfall comes during the austral winter between June and

September, with an average 250–300 mm precipitation per year. Substantial moisture arrives in the form of a coastal fog that blankets the near-coastal environment on an almost daily basis. The austral summer from December to March is dominated by strong, dry, southerly winds that entrain and saltate loose sand grains in the Geelbek dune system. The sand accumulates in a slight depression in the bedrock along the Ysterfontein-Hopefield axis in which the dune plume lies (Rogers 1980). Changes in moisture, as well as large daily fluctuations in temperature, play a significant role in the preservation of archaeological and paleontological finds in the Geelbek Dunes, and other similar arid environmental settings. The repeated cycles of heating and cooling and wetting and drying lead to unexpected patterns of fragmentation in which solid, dense bones experience more intense weathering than more porous, less dense bones. This important observation turns out to be contrary to the results expected by faunal experts and is being replicated in an ongoing study (Conard et al. 2008). Presently, the Geelbek Dunes are migrating northwards, pushed forward by the dry, southerly summer winds. The dune crests appear crisp and peaked during this season. Then in winter, when the winds are predominantly from the north, the dune crests become rounded in response to winds from the opposite direction. However, winter’s moisture allows vegetation to grow, which stabilises the dunes so that the net southerly migration is minimal. As the southerly winds of summer pick up speed and dry the surface, the northward movement of the dunes begins anew. We precisely measured the annual movement of the localities within the dunes by recording their exact position with a total station and comparing the results over five years of study. This dataset offers a unique glimpse at the evolution of the natural environment in the West Coast (Kandel et al. 2003, Fuchs et al. 2008). Fire We studied the effect of a natural brush fire on the accumulation of faunal remains in the open environment near Elandsfontein. A 45 km long burn that covered 18 400 hectares between Ysterfontein and Hopefield in January 2000 provided us with an opportunity to assess the impact of fire on background accumulations of fauna. Since we were particularly interested in knowing how a fire affected mortality rates, we determined the distribution of fauna that perished in the fire. Our observations showed that tortoise were most susceptible to the bush fire, probably due to their limited ability to escape (Avery et al. 2004). Whole tortoises that died in the quickly moving brush fire did not get charred; however, the older, disarticulated skeletal remains that already lay on the surface were blackened. Based on our data, we estimated that at least 90 000 tortoises died in the fire. This study provided us with a model for how background fauna accumulate at Geelbek (Avery et al. 2004).


27

After the fire, we monitored the recovery of vegetation and observed whether the burning of vegetation accelerated the formation of sand dunes (Avery et al. 2004). We also conducted burning experiments on common finds such as ostrich eggshell and marine shell. We altered heat and oxygen levels and added soil, fat and salt to understand how heating changes the structure and colour of these materials. These studies aimed at differentiating natural accumulation processes from anthropogenic ones are important for understanding the role of taphonomic processes at Geelbek. Field Methods To study the spatial distribution of these materials, we needed a versatile system to quickly record each find with millimetre accuracy in three dimensions (x, y, z). We sought to link the geographical data to a detailed description of the find and plot this information on a map. Using the Electronic Distance Measurement program written by Dibble & McPherron (1996) together with a total station, the GAASP team measured each find larger than 2 cm with a Leica T1100 theodolite and logged more than 30 000 datasets on a Husky 16 field computer. For each find, the field computer recorded locality name, find number, coordinates, area, scatter, find description, date, time, collector, geological context and reflector height. Using this efficient method, we plotted a site record of 679 finds on March 30, 2000 in the locality named Shelly. The deployment of this equipment in February 1998 enabled us to efficiently collect large numbers of finds from vast surfaces. In fact, we collected all finds from an area of almost 120 000 m2. After transferring the field data to a Microsoft Access database, we linked the datasets with Golden Software’s Surfer and ESRI’s ArcMap to plot the finds. Finds in the field were divided into nine main categories based on the different types of materials collected:

Lithic artefact–chipped, ground stone tool or manuport

Faunal remain–mammal, micromammal, bird, fish, reptile or amphibian

Bone tool–worked faunal remain Ostrich eggshell–unmodified or modified Shellfish–gastropod, mollusc or crustacean Ornament–ostrich eggshell or marine shell Pottery–prehistoric Modern find (post-contact)–glazed or unglazed

ceramic, metal or glass Bucket–collected finds excavated from either 1

m2 or 0.25 m2. For each single find, we entered three tiers of information, including artefact category, material type or species, and artefact descriptor (e.g. lithic, silcrete, scraper; fauna, eland, metatarsus; shell, white mussel, retouched). In addition to single finds, bucket finds represented the dry-screened remains collected from excavated transects to sample the fine fraction picked

from 5 and 2 mm sieves. Above all, screening increased the recovery of microlithic tools, ostrich eggshell beads and small molluscs, as well as enhancing the recovery of microfauna. Besides measuring the tangible artefacts, we recorded seven other kinds of spatial data:

Features–non-portable remains resulting from human activity, such as hearths and stone structures, with each piece of the feature described, measured and entered into the database

Topography–three dimensional plotting of the surface of the bays and dunes

Borders–to demarcate the boundary between deflation bays and mobile sand dunes, in other words, the window of visibility

Geology and profiles–for differentiating strata and horizons

Datums–to set up the measuring grid anywhere in the entire dune system

Samples–for radiometric dating or other analyses

Other–for miscellaneous isotopic, botanical and ecological studies.

The fieldwork was conducted in five seasons between 1998 and 2002 and was followed by the analysis of all finds at the Iziko South African Museum. Furthermore, we continued to monitor the dunes on an annual basis until 2007 to gather additional data on partially collected find scatters and freshly exposed occurrences. Dune Movement We calculated the rate of dune movement by measuring the boundary between the deflation bays and the surrounding mobile sand dunes. From year to year, we compared this window of visibility and determined the distance that the dunes had moved since the previous year (Kandel et al. 2003, Fuchs et al. 2008). Over the five years of field study, we measured an average dune movement in all localities of 7.23 m per year, with rates ranging from 0–12 m per year. To study the accumulation of finds in such a volatile environment, we conducted the Geelbek Object Movement Experiment (GOME). The goal of GOME was to observe what happens to materials in the dunes on different geological surfaces as deflation takes place. Thus, GOME tested the effect of wind over time on the movement of lithics, bone, ostrich eggshell and marine shell of different sizes and shapes placed on loose sand, compact sand and a rocky calcrete substrate. GOME demonstrated that assemblages of artefacts remain together over time and despite the process of deflation. This core observation about spatial integrity plays an essential part in our understanding of how the various assemblages at Geelbek formed over the millennia (Kandel et al. 2003).


28

Laboratory Analysis We developed criteria for identifying archaeological remains in the laboratory and used these protocols to systematise the analysis and increase the reproducibility of the data. In this section, we present details to assist the reader in understanding the data presented. All artefacts were weighed, measured, described, analysed and entered into a Microsoft Access database. Lithics To evaluate the lithics, we used a classification system similar to that commonly used in South Africa (e.g. Deacon 1984). To identify cores from multiple time periods we developed and implemented a new classification system based on technological features of cores (Conard et al. 2004). We defined blades and bladelets to be twice as long as wide, with bladelets smaller than 12 mm in width. Tools are defined as modified debitage that show specific retouch patterns. Angular debris represents chunks larger than 10 mm, small debris ranges from 5–10 mm, and microdebitage is smaller than 5 mm. Based on their macroscopic characteristics, raw materials were grouped into main categories of silcrete, quartz, metamorphic, igneous, silicified calcrete, cryptocrystalline siliceous and indeterminate. Raw materials located directly onsite include silicified calcrete. All other raw materials come from further away and include local (5–10 km) raw materials, such as quartz and igneous. Regional raw materials (15–25 km) included silcrete, metamorphic and cryptocrystalline siliceous. Fauna Using the extensive comparative collections at the Iziko South African Museum, we were able to identify many of the mammals, birds, fish, reptiles and shellfish. We supplemented Iziko’s collection with a comparative collection from Geelbek and its surroundings for badger, steenbok, hare, Cape dune mole rat, tortoise and marine shell. We classified the fauna to the most precise taxonomic unit and skeletal element possible, depending on its completeness and preservation. If a faunal remain could not be identified to the level of family, genus or species, we ranked it based on estimated living body weight (Apps 2000) using seven relative size classes (SC):

SC0 (0–5 kg): very small animals: hare, Cape dune mole rat, microfauna, bird

SC1 (5–20 kg): small mammals: duiker, steenbok, fox, wild cat

SC2 (20–100 kg): small-medium mammals: springbok, caracal, hyena

SC3 (100–300 kg): large-medium mammals: bontebok, kudu, wildebeest, lion

SC4 (300–1000 kg): large mammals: eland, Cape buffalo, Cape zebra

SC5 (>1000 kg): very large mammals: rhinoceros, hippopotamus, giant Cape buffalo

SC6 (>1000 kg): very large mammals: elephant, giraffe

Our system incorporates analytical elements of faunal analysis from both Brain (1974) and Klein et al. (1991). If a faunal remain of SC4 could be identified as an antelope or bovid, then it was classified as “Bovid-4”. For animals such as carnivores and birds, where the range of body weight is not as broad, we modified the size class system to help separate out size differences. Ostrich Eggshell, Marine Shell, Potsherds and Historic Finds Ostrich eggshell and beads were analysed according to the procedures laid out in Kandel (2004) and Kandel & Conard (2005). Marine shell was classified by comparing the finds with known specimens collected from West Coast beaches, and consulting field guides (Branch et al. 2005) and local experts. We noted basic attributes of prehistoric potsherds and historic finds. Antonia Malan and Jane Klose of the University of Cape Town (UCT) further studied the historic finds (pers. comm.), while Martin Kügler (2002) identified the sources of nine Dutch clay pipes. Burning Burnt and partially burnt artefacts were recorded to enable the plotting of burnt zones and correlate these data with the location of stone hearths. Regardless of the material, the basic characteristics of burning included black charring, white calcination, differential coloration, potlid fracturing and craquelure texture. We also conducted burning experiments on ostrich eggshell and marine shell to quantify the effects of heat on these materials. Refitting Finally, an intensive program to refit lithics, bone, ostrich eggshell, marine shell, prehistoric pottery and modern artefacts helped reconstruct many finds. Without this effort, many objects would have remained unidentifiable (Conard et al. 2008). Refitting also helped us understand how far artefacts moved in a given locality. Chronological Framework Establishing a chronological framework was a crucial element for interpreting the finds from Geelbek. The task was accomplished in five different ways:

Using typology and technology to define cultural entities

Looking at traits such as the mineralization of bone or the patination of stone artefacts to establish a relative chronology


29

Conducting a program of radiometric dating that incorporated radiocarbon, luminescence and uranium series (U-series) analyses

Studying the geological relationships among finds by digging trenches to better understand the stratigraphy

Conducting chemical and physical analyses to test isotopic signatures of bones or chemical residues in prehistoric pottery.

Figure 3: Diagnostic lithic artefacts selected from Equus,

Homo, Mathilda Rose and Rhino used to identify general period of occupation: (1) MSA parallel core, G-HO-1782; (2) MSA

inclined core, G-EQ-688; (3) MSA bifacial point, G-MR-1042; (4) Howiesons Poort large segment, G-EQ-497; (5) LSA small segment, G-RH-1937; (6) LSA backed point, G-RH-2083. All artefacts are made on silcrete except (3), which is made on

quartzite (drawings S. Feine).

Figure 4: Schematised stratigraphic sequence of ancient dunes (AD) at Geelbek, calcretes and paleosols with ages based on OSL, IRSL (Table 2) and U-series dating (Table 3) after Felix-Henningsen et al. (2003).

Figure 5: Average degree of mineralisation (oval) showing range of one standard deviation for each locality. The analysis excludes burnt or indeterminate bones and is capped at the maximum value of MC5. The number of faunal remains is in parentheses next to the locality name. Ovis had the least mineralised fauna, while Equus yielded the most. The Earlier Stone Age locality of Anyskop

Blowout at the West Coast Fossil Park in Langebaanweg is included for comparison.


30

Classification System One of the first steps in dealing with the lithic artefacts was to develop a system that could embrace all time periods. A unified lithic taxonomy was developed, which aimed at organizing cores from multiple time periods into a single classification scheme based on technological attributes (Conard et al. 2004). We also employed existing, typologically oriented systems to order the basic artefact types into a temporal framework. For example, handaxes signify Acheulean occupation, bifacial leaf points were equated with Still Bay, and large backed silcrete segments denoted Howiesons Poort. Levallois-like technology evidenced by parallel and inclined cores signalled the MSA (Conard et al. 2004), whereas small backed bladelets, points and segments correlated with later phases of the LSA (Fig. 3) (Deacon 1984). The study of pottery forms and styles was used to indicate occupation after 2000 BP (Sampson & Sadr 1999). Finally, manufacture dates for Dutch clay pipes (Kügler 2002), glazed ceramics and other modern finds helped provenience these objects within the historic period. Mineralization, Patination, Weathering The state of mineralization of the faunal assemblages offered a significant means of categorizing finds chronologically. Based on our experience in the dunes, we recognised six different states of fossilization, which we correlate with the geological sequence (Conard et al. 2008). We ranked these states from zero to five and viewed the mineralization category (MC) as a proxy for increasing antiquity:

MC0–indeterminate bones including most dental remains and burnt or calcined bones

MC1–fresh appearing, white to yellowish bone MC2–bones with brown staining on the surface

and interior MC3–denser bone, white and chalky, with dark

flecks of mineralization MC4–dense bone, beige to brown in colour MC5–very dense bone, dark brown to dark red

in colour We used this system to rank assemblages in each locality based on their average state of mineralization, excluding bones of MC0 from the calculation. These data provide a reasonable impression of the overall antiquity of a given locality (Fig. 4), especially in conjunction with additional lines of evidence: for example, MC4 and MC5 fauna in the presence of MSA artefacts, or MC1 and MC2 fauna co-occurring with marine shell, ostrich eggshell beads or prehistoric pottery. The patination and the surface preservation of stone artefacts played a small but significant role in giving an impression about the age of an assemblage. Artefacts with a higher degree of surface patination were assumed to have undergone longer periods of weathering and were therefore assumed to be older. For the same reason, artefacts with poorer surface preservation were

interpreted to have experienced longer periods of weathering in the soil and on the surface. While these criteria alone were not used to determine the age of an assemblage, when evaluated alongside with other attributes, they helped us make informed decisions about the overall antiquity of an assemblage (Dietl et al. 2005). Bone Modification The faunal assemblages were subjected to a macroscopic study of modifications to identify the main accumulators of the assemblages (Behrensmeyer 1978). We classified modifications into three main categories:

Anthropogenic–burnt, fresh or green breaks, cut marks, percussion or impact marks, and in a few cases, chop or saw marks

Biogenic–carnivore biting, porcupine chewing, rodent gnawing, gastric etching, insect borrowing

Geogenic–calcrete crust, sand crust, chemical etching.

Radiometric Dating Direct dating of archaeological remains and soil horizons added structure to the chronostratigraphy of the Geelbek Dunes (Fig. 5). By radiocarbon dating organic materials such as bone, ostrich eggshell and marine shell, we constructed a chronology of occupation especially for periods younger than 2000 cal BP (Fig. 6) (Conard & Kandel 2006). The radiocarbon dating of land snails and fossil plant remains helped determine the timing of ecological events associated with periods of greater moisture. We calibrated radiocarbon dates using the OxCal 4.1 online version (Bronk Ramsey 2009). We present all radiocarbon dates as calibrated ranges (1σ) in the format suggested by the journal Radiocarbon (Table 1). In the text, laboratory identification number follows the calibrated date range in parentheses. For terrestrial samples, we applied the ShCal04 calibration curve that includes an offset for the southern hemisphere (McCormac et al. 2002). However, we included an additional correction of 180 ± 120 BP to account for the offset from dating ostrich eggshell (Vogel et al. 2001). For marine samples, we applied the Marine09 calibration curve (Reimer et al. 2009) and used a ΔR value of 152 ± 60 BP to account for the local marine reservoir effect (Stuiver & Braziunas 1993). This ΔR value represents the average of six West Coast datasets located between Table Bay and Lamberts Bay that are available online from the 14CHRONO marine reservoir correction database of Queen’s University, Belfast (Reimer et al. 2009). This approach differs from that proposed by Stynder et al. (2009), Dewar (2010) and Dewar & Pfeiffer (2010). However, these authors make use of a single dataset collected at the Cape of Good Hope. Our use of six datasets from the immediate West Coast region surrounding Geelbek offers a more realistic approach for


31

calculating ΔR until more detailed results are published. The sand horizons were dated using optically stimulated (OSL) and infrared stimulated luminescence (IRSL) to determine when these geological units were deposited, or more precisely, the last time the sand grains were

exposed to light (Table 2) (Fuchs et al. 2008). The calcrete layers were dated using U-series to ascertain their age of formation, taking into consideration that subsequent water movement through the calcrete can alter the concentrations of radioactive elements (Table 3) (Felix-Henningsen et al. 2003).

Figure 6: Summary of calibrated radiocarbon ages for archaeological finds younger than 2000 cal BC. Negative numbers indicate years cal BC, while positive numbers indicate years cal AD. Laboratory numbers can be cross-referenced with Table 1.

Lab ID Geelbek ID Dating method

Material dated Find description Uncal

age BP ±STD Calibration curve Offset ±STD min Cal

BC/AD max Cal BC/AD

Beta-240257 AL-480,1 AMS Bone Cow, humerus 130 40 ShCal04 – – 1700 1930

Beta-240261 BB-948,1 AMS Shell White mussel, scraper 2860 40 Marine09 152 60 -570 -370

Pta-8707 CH-1214,1 14C Shell Black mussel, bulk sample 1730 45 Marine09 152 60 720 900

Beta-240259 CH-1344,1 AMS Bone Cow, lumbar vertebra 1780 40 ShCal04 – – 250 380

KIA-17745 CH-1460,1 AMS Bone Bone tool, point 710 25 ShCal04 – – 1290 1380Beta-240262 CR-111,1 AMS Shell Turbo, operculum 2260 40 Marine09 152 60 160 350

KIA-17746 CR-914,1 AMS Bone Bone tool, linkshaft fragment 560 25 ShCal04 – – 1410 1430

GrA-19666 EQ-525,1 AMS OES OES, modified 37 050 310 ShCal04 180 120 -35 400 -34 800Beta-240265 EQ-2655,1 AMS Shell Argenvillei limpet, burnt 890 40 Marine09 152 60 1510 1650

Pta-8382 EQ-2943 14C OES OES, bulk sample 44 600 +3500-2400 ShCal04 180 120 -49 000 -39 100

GrA-13530 HO-103,1 AMS Bone Human, femur 1100 50 ShCal04 – – 900 1030Beta-240258 HO-715,1 AMS Bone Eland, pelvis 2020 40 ShCal04 – – -40 80

Beta-240260 HO-1463,1 AMS Shell White mussel 1990 40 Marine09 152 60 490 650

GrA-17558 HT-101,1 AMS Bone Human, tibia 405 40 ShCal04 – – 1460 1620KIA-17756 LO-298,1 AMS OES OES bead, large 3670 30 ShCal04 180 120 -1930 -1670GrA-17565 LO-1600,1 AMS Bone Human, radius 2040 35 ShCal04 – – -40 60KIA-17750 NO-951,1 AMS Shell White mussel, scraper 2970 25 Marine09 152 60 -730 -530

Pta-8695 NO-1223,32 14C Shell Black mussel, bulksample 2980 20 Marine09 152 60 -730 -540

KIA-17755 NO-1321,11 AMS OES OES bead, blank 2580 25 ShCal04 180 120 -630 -350

Pta-8691 PO-527,3 14C Shell Black mussel, bulk sample 2950 60 Marine09 152 60 -720 -480

KIA-17761 PO-1006,1 AMS OES OES bead, small 2500 25 ShCal04 180 120 -530 -220GrA-28388 PO-3050,1 AMS Shell Whale barnacle 2900 35 Marine09 152 60 -620 -400KIA-17751 PO-3730,1 AMS Shell White mussel, scraper 2715 25 Marine09 152 60 -360 -200Beta-240264 RH-120,1 AMS Shell Keyhole limpet 3790 40 Marine09 152 60 -1670 -1500

GrA-13527 RH-673 14C Geo Trigonephrus, bulk sample 40 890 +1060

-950 ShCal04 – – -40 100 -37 900

Pta-7928 RH-962 14C Geo Wood, carbonised 210 15 ShCal04 – – 1670 1800Pta-7929 RH-973 14C Geo Wood, carbonised 130 45 ShCal04 – – 1700 1940Pta-7949 RH-1246 14C Geo Trigonephrus, bulk 1325 15 ShCal04 – – 690 770


32

Lab ID Geelbek ID Dating method

Material dated Find description Uncal

age BP ±STD Calibration curve Offset ±STD min Cal

BC/AD max Cal BC/AD

sample Pta-8770 RH-2419,1 14C Bone Bovid, metacarpus 6800 140 ShCal04 – – -5770 -5520KIA-17748 SH-1251,1 AMS Shell White mussel, scraper 2960 30 Marine09 152 60 -720 -510

Pta-8689 SH-2514,52 14C Shell Black mussel, bulk sample 2870 20 Marine09 152 60 -570 -370

KIA-17757 SH-2935,14 AMS OES OES bead, small 2465 25 ShCal04 180 120 -480 -170Pta-8688 SH-3406,1 14C Shell Granite limpet 2905 20 Marine09 152 60 -620 -400KIA-17762 SN-493,1 AMS OES OES, modified 890 20 ShCal04 180 120 1230 1460KIA-17759 SN-500,1 AMS OES OES, modified 775 40 ShCal04 180 120 1310 1530KIA-17758 SN-1327,1 AMS OES OES bead, large 380 20 ShCal04 180 120 1650 1910Pta-8406 SN-1525,1 14C Bone Eland vertebra 460 45 ShCal04 – – 1430 1610Pta-8768 SN-1600,1 14C Bone Large bovid, humerus 410 45 ShCal04 – – 1450 1620Pta-7760 ST-1915,1 14C Geo Calcrete, 2nd 8810 90 ShCal04 – – -7950 -7610Pta-7759 ST-1916,1 14C Geo Rhizolith 1890 50 ShCal04 – – 90 250Beta-240263 ST-2505,1 AMS Shell Turbo, operculum 3350 40 Marine09 152 60 -1160 -950

KIA-17760 TO-1548,1 AMS OES OES bead, large 1260 25 ShCal04 180 120 870 1140KIA-17749 TO-2627,1 AMS Shell White mussel, scraper 2385 25 Marine09 152 60 30 200

Table 1. Geelbek Dunes. Summary of radiocarbon dating results, ordered alphabetically by Geelbek ID, presenting uncalibrated and

calibrated ages with one standard deviation. For calibrated ages, negative numbers indicate years cal BC, while positive numbers indicate years cal AD. See text for details on calibration, offset on ostrich eggshell dates and ΔR on marine shell dates. See also

Figure 6 for a plot of dates younger than 2000 cal. BC. OES stands for ostrich eggshell.

Laboratory ID Geelbek ID Dating method Material dated Geological horizon Age ka BP ± STDMGF3-BT10 AL-552 OSL Sand AD2 4.87 0.25MGF3-BT11 AL-553 OSL Sand AD2 4.62 0.27MGF3-BT12 AL-554 OSL Sand AD1 9.91 0.54MGF4-BT9 AL-555 OSL Sand Pavement 16.39 0.78MGF1-BT2 HO-1905 OSL Sand Pavement 26.83 2.48MGF1-BT4 HO-1906 OSL Sand Pavement 24.74 1.26MGF2-BT5 HO-1907 OSL Sand Pavement 19.20 1.03MGF2-BT6 HO-1908 OSL Sand Pavement 21.18 1.13GBPO3713 PO-3713 IRSL Sand AD2 6 1GBRH1170 RH-1170 IRSL Sand AD0 61 5GBRH1776 RH-1776 IRSL Sand AD1 10 1MGF5-BT7 SN-1646 OSL Sand Pavement 1.15 0.1GBSR993 SR-993 IRSL Sand AD0 57 4GBSC1 ST-1910 IRSL Sand AD2 5 1GBSB1 = GBK1 ST-1911 IRSL Sand AD1 11 1GBSB2 = GBK2 ST-1912 IRSL Sand AD0 147 12

Table 2. Geelbek Dunes. Summary of OSL and IRSL dating results, ordered alphabetically by Geelbek ID (Felix-Henningsen et al.

2003, Fuchs et al. 2008). See Fig. 4 for geological cross-section.

Laboratory ID Geelbek ID Dating

method Material dated Geological horizon

Linear uptake (f0) ka BP ± St Dev Early uptake

(f1) ka BP ± STD

C-7108 EQ-2939 U/Th Calcrete CAL4 111 5 105 5.60C-7109 EQ-2941 U/Th Calcrete CAL2 150 8.30 138 10.30Pta-7465 HO-1897 U/Th Calcrete CAL4 114.64 7.15 102.43 7.05Pta-7464 HO-1898 U/Th Calcrete CAL1 196.67 8.82 174.32 9.24C-6556 RH-1245 U/Th Calcrete CAL5 83 3.20 67.10 3.20C-7111 SR-997 U/Th Calcrete CAL4 119 5.50 109 7.10C-7112 SR-1011 U/Th Calcrete CAL1 251 29 241 33C-6432 ST-1913 U/Th Calcrete CAL2 161.80 10 146.40 10.70C-6431 ST-1916 U/Th Rhizolith Pavement 25.30 0.70 21.30 0.70Pta-7466 TO-2924 U/Th Calcrete CAL1 227.44 27.45 221.10 43.51Pta-7464 TO-2924,1 U/Th Calcrete CAL3 132.63 10.16 122.02 4.98Pta-7467 TO-2925 U/Th Calcrete CAL1 226.82 41.61 211.21 35.73

Table 3. Geelbek Dunes. Summary of U-series dating results with results for linear and early uptake models, ordered alphabetically

by Geelbek ID (Felix-Henningsen et al. 2003, Fuchs et al. 2008). See Fig. 4 for geological cross-section.


33

Geology Understanding the stratigraphic sequence of geological units in the dunes (Fig. 5) turned out to be the best way to determine the relative age of an archaeological assemblage. The exposed aeolian sediments and the underlying geological strata each represent distinct layers of windblown sand that have undergone different soil formation processes. Each layer of sand represents an ancient dune (AD) that accreted during a dry period and then underwent soil development during moister, warmer periods (Felix-Henningsen et al. 2003). Layers of windblown sand that were subsequently cemented underground due to water percolation and perching form the primary and secondary calcrete banks (Netterberg 1969). Due to deflation these sand and calcrete horizons are now exposed differentially in each locality at Geelbek and provide the basis for understanding climatic changes that affected the region during the Pleistocene. Furthermore, it follows that the cultural remains found upon the surface of a geological unit must be younger than the date of that unit’s deposition. Ancient Dunes Two ancient dune horizons are distinct and widespread so that they could easily be recognised based on macroscopic properties. The older unit, AD1, is semi-consolidated, reddish-brown sand with frequent black flecks of carbonised, but not burnt, remains of wood and roots. This cambic Arenosol is a soil that developed during a moist period after the sand was deposited about 10–11 ka (IRSL dates, Felix-Henningsen et al. 2003). However, the frequent carbonised plants remains observed in this unit are not associated with its deposition, as shown by radiocarbon dates of 1670–1800 cal AD (Pta-7928) and 1700–1940 cal AD (Pta-7929) (Table 1). The observation of modern dunes covering bushes and trees also suggests that these carbonised plant remains are recent. The younger unit, AD2, is less consolidated than AD1 and unconformably overlies it. AD2 is yellowish sand noted for its thick stands of rhizoliths. These dense stands and mats represent vertically and horizontally oriented, mineralised, fossil root casts that once grew in AD2. The development in AD2 of a weak soil horizon, a calcaric Arenosol, occurred during a period of increased moisture after the sand was deposited about 5–6 ka (Felix-Henningsen et al. 2003). Deeper AD layers were recognised, but not temporally distinct enough to warrant a specific designation. These include a series of deflated surfaces with OSL ages ranging from 1–27 ka that resemble wind-blown desert pavements. Deeper sand units interstratified between underlying calcrete layers yield IRSL dates of 57 ± 4 ka, 61 ± 5 ka and 147 ± 12 ka that are referred to collectively as AD0 (Table 2) (Felix-Henningsen et al. 2003, Fuchs et al. 2008).

Calcrete The formation of calcrete is a pedogenic process that occurs under semi-arid climatic regimes where the seasonal variation in rainfall is pronounced (Netterberg 1969, Eitel 1994). Several different calcrete layers can be recognised in the dunes based on their different dating signatures. Because calcrete weathers variably and can be stained by iron and manganese oxides, it was not possible to use macroscopic attributes to reliably differentiate the many calcrete exposures. Thus, we viewed geographically close calcretes that yielded similar ages as probable extensions of the same depositional unit. The linear uptake (f0) model for the U-series age of calcrete formation cluster in five main groups, 83 ka, 111–119 ka, 133–162 ka, 197 ka and 227–251 ka. The early uptake model (f1) yields slightly younger ages of 67 ka, 102–109 ka, 122–146 ka, 174 ka and 211–241 ka for the same units (Table 3) (Felix-Henningsen et al. 2003, Fuchs et al. 2008). However, the similarity of both models suggests that the results are good estimations for the date of calcrete formation, though greater precision could not be achieved. Other Analyses Other types of chemical and physical analyses led to a better understanding of certain ecological and behavioural parameters. Judith Sealy of UCT established the ratios of stable carbon and nitrogen isotopes in human bone from the localities Homo (GrA-13530), Hetero (GrA-17558) and Loop (GrA-17565) dated between 40 cal BC and 1620 cal AD (Table 1) to assess where these individuals obtained most of their food. According to Sealy (pers. comm.) the three individuals gained a significant portion of their diet from marine resources. Frans Radloff of Stellenbosch University analysed the carbon isotope ratios from elephant bone to yield information about climate and the presence of C3/C4 grasses in the region. The elephant from the locality Toaster, apparently killed by lead shot, is one of the last known from the West Coast. According to Radloff (pers. comm.) its diet consisted of ca. 10% C4 grasses, obtaining most of its nutrition from C3 grasses or from browsing. Mark Copley of the University of Bristol tested lipids from residues scraped off the interior of potsherds to provide data about the sources of fat in the diets of Geelbek’s inhabitants. According to Copley (pers. comm.) absorbed lipid residues were detected in seven of nine sherds from the locality Pottery. Two of them provided indication of marine derived fats, but the other lipid sources could not be identified further. Landscape Approach One of the questions of our research was to determine whether the finds on the landscape at Geelbek were discarded at specific focal points to which people returned. Alternatively, the finds could represent a thin blanket of materials scattered across the entire region. Archaeologists tend to assume that prehistoric people used the entire landscape and left materials behind nearly


34

everywhere they went. Caves and rock shelters drew people to a specific feature in the landscape, and in these, multiple finds layers with dense artefact accumulations testify to long and repeated occupations. Coastal shell middens are a type of site that attracted people because of their proximity to marine resources. At shell middens, thick horizons attest to intense occupational phases. The same could be said for other sites at water sources such as springs and seasonal pans. People would have repeatedly been drawn to these types of sites, but because these features can be transient and often cover larger areas, accumulations are not as dense. On the other hand, the open-air sites of Geelbek are ephemeral in nature. They were occupied briefly by people who engaged in different activities at various points on the landscape. The shifting dunes open windows into this ancient landscape. Thus, the sites at Geelbek offer glimpses of past activities that

occurred at diverse points on the landscape. We assume that these sites were utilised similarly to areas outside the modern dunes, and if we were to survey a broader area, we would expect to see a similar background scatter covering the entire landscape. To examine these issues, our research strategy deliberately documented large archaeological exposures. As described in the section entitled Field Methods, we carefully piece-plotted all finds from a total area of almost 120 000 m2 with much of it collected on multiple occasions to monitor the deflation and exposure of new finds. Only through this major commitment to work on a large scale and study numerous localities is it possible to document the wealth and diversity of these kinds of archaeological settings, which had previously been ignored by researchers (Conard et al. 1999).

LOCALITY GENERAL GEOLOGY LITHICS CHRONO-TECHNOLOGY

Tot

al a

rea

coll

ecte

d

(m²)

Rat

e of

du

ne

mov

emen

t (m

/yr)

Ave

rage

min

eral

izat

ion

co

de

Cal

cret

e

Rew

ork

ed p

avem

ent

AD

1

AD

2

Lit

hic

arte

fact

s (n

)

Lit

hic

arte

fact

s (g

)

Lit

hic

arte

fact

s (n

) /m

²

Lit

hic

arte

fact

s (g

) /m

²

Mid

dle

Sto

ne A

ge

Bif

acia

l poi

nt

How

ieso

ns

Poo

rt

Lat

er S

tone

Age

Alice 9555 10.3 2.31 X X X X 88 1959 0.01 0.21 X X

Bay 35 2117 11.3 3.59 X X X X 82 981 0.04 0.46 X X

Bleached Bone 4317 7.0 1.60 X X 9 17 – – X

Bovid 3850 No data 1.88 X X X 28 3140 0.01 0.82 X

Check 6371 0.0 1.65 X 120 924 0.02 0.14 X

Crow 7619 8.0 3.38 X X X X 460 4368 0.06 0.57 X

EDM 4455 8.8 3.25 X 82 127 0.02 0.03

Equus 10 844 9.0 4.30 X X 91 3149 0.01 0.29 X X X

Frikkie 1221 7.0 1.79 X X 2 92 – 0.08

Hetero 896 9.5 2.13 X X 1 0.01 – –

Homo 5636 7.0 3.06 X X X 435 21 031 0.08 3.73 X X X

Loop 5530 7.0 4.17 X X X 1019 4927 0.18 0.89 X

Matilda Rose 9614 9.3 2.74 X X 118 992 0.01 0.10 X X X

Nora 1792 5.0 1.12 X 884 478 0.49 0.27 X

Olifant 2016 No data 1.15 X 3 177 – –

Ovis 82 No data 1.00 X X – – – –

Pottery 7637 6.0 1.07 X 628 1400 0.08 0.18 X

Rhino 3532 7.0 2.24 X X X X 1061 11 151 0.30 3.16 X X

Shelly 5319 8.0 1.29 X 2001 7067 0.38 1.33 X

Snoek 4614 2.7 1.24 X X X X 259 448 0.06 0.10 X X

Stella 9266 8.8 2.68 X X X X 52 2912 0.01 0.31 X X X

Stonering 8331 4.5 3.02 X X X X 759 6835 0.09 0.82 X

Toaster 3947 8.5 1.31 X X 184 3534 0.05 0.90 X X

TOTAL 118 562 16 14 10 17 8366 75 710 8 2 3 18

AVERAGE 7.23 2.26 0.08 0.63

Table 4. Geelbek Dunes. Summary of data for each locality, including surface area collected, rate of dune movement, average

mineralization code (see Fig. 5), geology, lithic artefacts and a summary of chrono-technological units. Presence is indicated with an X.


35

Table 5. Geelbek Dunes. Summary of further data for each locality, including terrestrial faunal remains, ostrich eggshell, marine

fauna, pottery and features, continued from Table 4. Results In this section, we present a summary of the results (Tables 4 & 5) of the systematic study of the Geelbek Dunes in a succinct format on a locality-by-locality basis starting in the south of the project area moving north (Fig. 2). It is important to recognise that over the course of field work, the area of visibility within a locality moved northwards and changed shape as the dunes migrated. The locality names always refer to the same deflation hollow as its window of visibility changed. The results of radiocarbon dating refer to Table 1. Following the presentation of the results, we summarise the results grouped into chrono-cultural entities. Locality EDM The locality EDM (Bay 32) was named after the crew spent the larger part of a day battling the total station in the field. We collected materials from EDM mainly in 1998 and over the course of study investigated a combined area of more than 6100 m2. EDM was a large, east-west oriented elongate bay, approximately 70 by 20 m that migrated north at an average rate of 8.75 m/yr (Fig. 2). Its exposed position just south of the narrow neck between the southern and northern dune lobes of the

Geelbek Dunes explains this rate of movement that is higher than the dune-wide average of 7.23 m/yr. The overall archaeological character of EDM cannot be determined reliably since few diagnostic artefacts were recovered. The mineralised faunal remains are not clearly archaeological in nature based on a lack of anthropogenic modification. Some of the non-mineralised fauna are burnt, suggesting either an anthropogenic component or natural burning as the result of a brush fire. Locality Bleached Bone Situated just 50 m west of EDM over a high dune ridge (Fig. 2), the locality Bleached Bone (Bay 12) was named after the characteristic bleached preservation of the fauna in this bay. Collections in Bleached Bone began in 1998 and over the course of study a combined area of more than 4300 m2 was investigated. Bleached Bone is situated on the north-western edge of the southern dune lobe. A north-south oriented, oval bay, measuring approximately 70 by 100 m, it moved north at an average rate of 7 m/yr. Despite its position just south of the narrow neck between the southern and northern dune lobes, this bay experienced a rate of movement that was slightly below the dune-wide average.

LOCALITY TERRESTRIAL FAUNA OSTRICH EGGSHELL MARINE FAUNA POTTERY

All

fau

na

(n)

Fos

sil f

aun

a

(MC

4 +

MC

5) (

n)

All

fau

na

(g)

Fos

sil f

aun

a

(MC

4 +

MC

5) (

g)

Fos

sil /

All

fau

na

(W

t %

)

Fau

nal

den

sity

(n

) /m

²

Fau

nal

den

sity

(g

) /m

²

Wor

ked

bon

e

OE

S (

n)

OE

S (

g)

Per

fora

ted

OE

S (

n)

OE

S b

ead

s (n

)

Mar

ine

shel

l (n

)

Mar

ine

shel

l (g)

Ret

ouch

ed s

hel

l (n

)

Sh

ell o

rnam

ents

(n

)

Fis

sure

llid

lim

pet

s (n

)

Fis

h r

emai

ns

(NIS

P)

Pot

tery

(n

)

Pot

tery

(g)

Cal

cret

e fe

atu

res

(n)

Alice 292 86 18

786 5694 30% 0.03 0.01 – 1 2 – – 2 5 – – – 2 – – – Bay 35 573 300 6400 5235 82% 0.27 0.14 – 52 54 1 – 109 77 3 – 11 – – – – Bleached Bone 797 16 7204 84 1% 0.18 0.00 – 7 309 – – 11 22 – – – – – – – Bovid 374 5 3043 275 9% 0.10 0.00 – 53 67 1 – 5 3 – – 1 1 – – 2 Check 412 58 6485 251 4% 0.06 0.01 1 153 133 – – 383 193 – 3 2 -- 3 25 2 Crow 561 325 3895 3012 77% 0.07 0.04 9 298 754 10 – 16 12 – – 3 5 2 2 – EDM 167 101 3545 3372 95% 0.04 0.02 – 4 8 – – 4 20 – – – 1 – – –

Equus 836 697 14

628 13

896 95% 0.08 0.06 – 1304 1609 76 – 28 32 2 – 2 2 1 3 3 Frikkie 26 4 318 234 74% 0.02 0.00 – 6 6 – – 6 2 – – – – – – – Hetero 38 11 360 229 63% 0.04 0.01 – 1 1 – – 1 0.2 – – – – – – –

Homo 783 314 19

087 13

469 71% 0.14 0.06 2 60 123 2 – 87 60 – – 11 – – – 1 Loop 431 313 9037 8168 90% 0.08 0.06 – 79 65 14 7 106 96 – 1 8 – 17 40 – Matilda Rose 315 132 2200 1749 80% 0.03 0.01 – 232 169 5 – 77 33 – – – 6 4 3 – Nora 207 1 699 2 0.3% 0.12 0.00 – 25 41 – 187 242 352 20 1 1 1 38 296 1 Olifant 39 0 637 0 0 0.02 0.00 – – – – – – – – – – – – – 8 Ovis 96 0 1513 0 0 1.18 0.00 – – – – – – – – – – – – – – Pottery 1034 5 3384 17 1% 0.14 0.00 – 132 158 1 614 391 265 13 – 7 11 78 1659 4

Rhino 962 172 65330 55

200 84% 0.27 0.05 – 52 101 – – 68 31 – – 4 1 – – – Shelly 1088 1 2748 4 0.1% 0.20 0.00 2 115 92 9 220 1091 1903 96 – 10 4 – – 6

Snoek 878 45 10

687 1489 14% 0.19 0.01 – 181 264 15 1 45 58 1 – 14 1 – – –

Stella 2083 819 75

587 52

233 69% 0.22 0.09 – 110 1124 4 5 18 10 1 2 9 1 – – – Stonering 492 197 4905 1974 40% 0.06 0.02 – 87 78 7 – 55 71 – – 3 – 66 1484 –

Toaster 992 60 118 489 5797 5% 0.25 0.02 1 32 38 3 13 174 379 8 6 15 12 418 1331 13

TOTAL 13476 3662 378 968

172 383 15 2984 5196 148 1047 2919 3625 144 13 101 48 627 4843 40

AVERAGE 43% 0.17 0.03


36

The overall archaeological character of Bleached Bone is late LSA, based on one diagnostic, broken, backed point made out of silcrete. This isolated tool could indicate a successful or failed hunt that would leave projectile equipment scattered across the landscape. Yellen’s (1977) ethno-archaeological research suggests that projectiles would rarely be abandoned at residential sites and are most likely to be lost during a hunt, that is, “the one that got away”. A radiocarbon date of 570–360 cal BC (Beta-240261) on a white mussel (Donax serra) scraper dates a phase of occupation that is frequent at Geelbek. The small collection of mineralised fauna are not clearly archaeological in nature based on a lack of anthropogenic modification, however, bone preservation here was poor. Three of the non-mineralised fauna show impact fractures, although the majority of the fauna does not show other signs of anthropogenic modification. Locality Homo / Bay 35 The combined locality Homo/Bay 35 is located 200 m northeast of EDM and Bleached Bone (Fig. 2). It started out as two separate deflation hollows that grew together over the course of fieldwork. Bay 35 was the first Geelbek locality investigated and is the only locality to keep its numerical designation, while Homo (Bay 36) was named after human remains discovered there. Collections in Bay 35 began in 1998, and in Homo in 1999. By January 2000 dune movement merged the two bays, which were thereafter considered as a single entity. Over the course of study a combined area of more than 7700 m2 was investigated. Bay 35 began as two elongated strips along the eastern side of the dune neck, while Homo started as a 50 m long irregular patch on the western side. When the two parts joined, Homo/Bay 35 became a large, east-west oriented, oval bay, approximately 150 by 30 m, that moved northwards at a rate of 11.25 m/yr in Bay 35 and 7 m/yr in Homo. The exposed position of Bay 35 in the narrow neck between the southern and northern dune lobes is responsible for the highest average rate of movement in the dunes. The rate of dune migration at Homo is likely lower than expected because it results from just two years of measurement. The materials collected from Homo/Bay 35 are varied and stem from a well-studied geological context. Several U-series dates from Homo/Bay 35 establish a chronological history for the calcrete and sand horizons. The oldest calcrete formation dates to 175–200 ka, a laminar crust to 100–115 ka, and a reworked surface to 19–27 ka. OSL dates from locality Alice, the neighbouring locality to the north, of 9.91 ± 0.54 ka for AD1 and 4.74 ± 0.39 ka for AD2, complete our understanding of the geological sequence. The overall archaeological character of Homo/Bay 35 contains elements of both MSA and late LSA. The MSA component is based on the presence of mostly silcrete parallel and inclined cores, a silcrete bifacial point similar to Still Bay, and denticulates. Numerous heavily mineralised faunal remains and ostrich eggshell fragments are not clearly archaeological in nature.

The late LSA designation is based on a microlithic stone tool assemblage of 11 backed pieces and an adze. Seven ground stone tools include two upper grinders, one hammerstone and four combination grinders and hammerstones (Fig. 7). Calibrated radiocarbon results show at least three phases of LSA occupation, the first, an eland pelvis dated to 40 cal BC–80 cal AD (Beta-240258), the second, a white mussel shell dated to 490–650 cal AD (Beta-240260) and the third, a human femur dated to 900–1030 cal AD (GrA-13530). The latter was assayed and yielded a δ15N value of +12.9‰ (UCT-8069) according to results provided by J. Sealy at UCT. This value suggests a lower level of marine resources in the diet than the sample from locality Loop. Two bone tools decorated with chevrons and crosshatch patterns are link-shafts that are typical finds of the late LSA (Fig. 8). Also interesting in Homo/Bay 35 is the presence of 196 marine shells of which 22 (11%) are fissurellid keyhole limpets. These naturally perforated shells offer themselves as premade ornaments and their association with late LSA contexts on the south coast of South Africa is clear (Hall & Binneman 1987). A single well-preserved roasting platform resting on AD2 and surrounded by faunal remains, lithics and shell supports the LSA determination (Fig. 9). Locality Alice Located one dune crest or 60 m north of Homo/Bay 35, the locality Alice (Bay 9) (Fig. 2) was named in honour of the birth of J. Parkington’s daughter. Alice was collected due to the presence of numerous, complete, fossilised bones of large mammals. Work began in 1998 and more than 9500 m2 were investigated. Alice is an east-west oriented, oval shaped bay, approximately 120 by 50 m, that moved north at a rate of 10.25 m/yr. Its exposed position in the narrow neck between the southern and northern dune lobes is responsible for this second highest average rate of movement. Based on data from Homo/Bay 35, the neighbouring locality to the south, the oldest calcrete formations date to 175–200 ka, a laminar crust to 100–115 ka, and a deflated surface to 16.39 ± 0.78 ka. OSL dates from Alice of 9.91 ± 0.54 ka for AD1 and 4.74 ± 0.39 ka for AD2 complete the geological sequence. Despite the presence of older calcrete layers, the overall archaeological character of Alice is late LSA, based on two microlithic backed stone tools, which should be viewed spatially as a continuation of the backed pieces from Homo/Bay 35. A scatter of cow bones dating to 1700–1930 cal AD (Beta-240257) appears unrelated, and instead likely represents a lost farm animal. The heavily mineralised faunal remains from Alice are unlikely to be archaeological in nature based on their lack of anthropogenic modification. Locality Olifant Continuing 350 m north along the western flank of the dune field, we arrive at the locality Olifant (Bay 13) in a small side valley on the edge of the dune field between


37

Figure 7: Illustrations of ground stone tools from localities Equus, Homo, Pottery and Shelly, (1) bored stone, G-PO-619 and 3473; (2) platform core used as hammerstone and

grinder, G-SH-3411; (3) upper grinder, G-HO-467; (4) knapper, G-PO-3293; (5) knapper, G-SH-482 and 3397; (6) bored stone fragment, G-EQ-1072. Artefacts (1) and

(3) are made on granite, while the rest are made on metamorphic stone (drawings S. Feine & E. Ghasidian).

Figure 8: Illustration of bone tools from Crow, Shelly and Homo. The size and form of these finds supports their

interpretation as linkshafts: (1) G-CR-915; (2) G-CR-914; (3) G-CR-916; (4) G-CR-917; (5) G-CR-943; (6) G-CR-

942; (7) G-SH-858.15; (8) G-HO-136 (drawings S. Feine).

Figure 9: Geelbek Dunes. This map shows the structured distribution of finds and features in Homo. Radiocarbon ages of fauna and shell date two phases of occupation to the middle of the first millennium cal BC. The diamond shaped symbols represent individual

blocks of burnt calcrete that we interpret as a single roasting platform. The cultural remains include lithics, fauna and shell that are concentrated throughout the feature. We interpret this feature as a campsite where people sat around a fireplace, chipping lithics

and discarding faunal remains. low, rolling dunes (Fig. 2). This locality owes its name to the Afrikaans word for elephant. Collections in Olifant of several lamella of elephant molars occurred only in 2000 and amounted to an area of more than 2000 m2. Olifant is situated on the south-western edge of the northern dune lobe. A north-south oriented, oval bay, approximately 35 by 85 m, the rate of movement could not be calculated because study was limited to one year. Olifant lacks diagnostic lithic artefacts for an age determination. Neither the small collection of non-mineralised fauna, nor the few mineralised fauna show traces of anthropogenic

modification. The field crew documented eight roasting platforms 120–200 m southeast of Olifant. While these distinctive features lacked diagnostic artefacts, their proximity to Olifant parallels a similar occurrence described at locality Toaster. Locality Bovid The locality Bovid (Bay 14) was discovered 200 m northwest of Olifant on a level rise of calcrete and named on account of the many moderately mineralised (MC3)


38

large bovid remains recovered there (Fig. 2). In 2000 we collected finds from an area of more than 3800 m2. Bovid is situated on the south-western corner of the northern dune lobe. This north-south oriented, rectangular bay is approximately 100 by 60 m. The rate of movement could not be calculated because study was limited to one year. Most of the lithic artefacts from Bovid are not diagnostic with the exception of one quartz end scraper and a broken upper grindstone on quartzite that was subsequently used as a platform core. Although these artefacts and three platform cores suggest an LSA affinity, further evidence is lacking. Locality Stella Located 400 m east of Bovid, Stella (Bay 40) is named after the project director’s oldest daughter on the occasion of her first birthday (Fig. 2). Finds were collected mainly in 1998 and 2000 and covered an area of more than 9200 m2. Stella includes both parts of a very large deflation hollow that remained separated by about 30 m of modern sand and vegetation throughout the field work. We conducted excavations over an area of 150 m2 at Stella. This locality is situated in the middle of the northern lobe where long, east-west oriented valleys separated by high transverse dunes cross the entire breadth of the dune field. The western part of this east-west oriented bay is 100 by 50 m, while the eastern part is 100 by 40 m. Dune movement at Stella averaged 8.75 m to the north annually with a maximum of 12 m recorded in one year. The materials collected from Stella are varied and stem from a well-studied, geological context. U-series dates on the basal calcrete yield results between 146–162 ka, and IRSL dates for the overlying ancient dune AD0 are consistent with a date of 147 ± 12 ka. Fossil root casts in AD0 are much younger and date to 21–25 ka with U-series. IRSL dates of 11 ± 1 ka for AD1 and 5 ± 1 ka for AD2 are consistent with the results for the same units in Alice. A rhizolith in AD2 dating to 90–250 cal AD (Pta-7759) suggests that the area was stabilised by vegetation. In Stella, multiple phases of occupation are suggested. An MSA presence is confirmed by two Howiesons Poort segments and nine denticulates, as well as five parallel and three inclined cores (Dietl et al. 2005). This is one of just three examples of such segments in the Geelbek Dunes, the others coming from localities Rhino and Equus. The mineralised fauna indicates some human modifications, although biogenic modifications predominate. A late LSA signature was also observed in Stella, with four silcrete backed bladelets, one backed point, six ground stone tools combining hammerstones with upper grinders, and five ostrich eggshell bead blanks. One radiocarbon date on a turban shell operculum (T. cidaris cidaris) yielded an age of 1170–950 cal BC (Beta-240263). A portion of a smooth-edged abalone (H. midae) fragment represents a rare example of a shell ornament at Geelbek. Also interesting in a locality with

55 shells is the presence of nine (16%) fissurellid keyhole limpets. These naturally perforated shells offer themselves as premade ornaments and their association with late LSA contexts on the south coast of South Africa is known (Hall & Binneman 1987). Locality Hetero Less than 150 m northeast of Stella, four, small, irregular deflation areas exposed four human remain fragments (Fig. 2). Hetero (Bay 52) was named after these remains since the name Homo was already taken. Finds were collected in the years 2000 and 2002 over a total area of ca. 900 m2. The largest collection area was circular with a circumference of 20 m. Hetero moved northwards at an average of 9.5 m per year. The geological sequence at Hetero can be seen as equivalent to Stella, due to its proximity. However, no diagnostic finds were recovered. A human tibia dated to 1450–1620 cal AD (GrA-17558), indicates a relatively recent age. The tibia yielded a δ15N value of +11.0‰ (UCT-8070) according to the results provided by J. Sealy at UCT. This value suggests a lower level of marine resources in the diet compared to the samples from Homo and Loop. Locality Check About 500 m east of Stella in the south-eastern corner of the northern dune lobe, a relatively stable area of calcrete and sand is exposed. This north-south oriented 30 by 160 m locality was named Check (Bay 66) because we expected our grant funding to arrive that day (Fig. 2). Finds were collected in Check in 1999 and 2002 covering a surface of more than 6300 m2. Check was the only Geelbek locality that did not move northwards over the course of the GAASP field campaigns. Although Check’s eastern boundary expanded by 20 m, its other boundaries remained as originally measured. This atypical movement can be explained by Check’s location on the trailing, southern edge of the dune field. Since the dunes already moved across the area, this locality represents what remains in their aftermath. The finds from Check are some of the few from Geelbek to suggest an occupation by herders. Radiocarbon dates of 250–380 cal AD (Beta-240259) on a cow vertebra place occupation at a time when herders are already known from other sites in the Western Cape (Sealy & Yates 1994, Henshilwood 1996, Vogel et al. 1997, Sadr et al. 2003, Sadr 2004, Webley 2007). Black mussel (Choromytilus meridionalis) shell document use of Check at 720–900 cal AD (Pta-8707). Despite the presence of 240 g of marine shell from a moderate variety of species more than 5 km from the shore, the total weight of shellfish amounts to just 5% (Buchanan 1988) of the modern daily calorie allowance of 2000 kcal for one individual. Three perforated plough shells (Bullia digitalis) are likely examples of shell ornaments. The diagnostic lithic artefacts include a silcrete backed bladelet, a quartz bipolar core and three ground stone tools including a hammerstone, upper grinder and lower


39

grinder. Still, these finds do not clarify the hunter versus herder question. Two distinct roasting platforms present in areas where finds are absent represent a common late LSA feature in the landscape of the south-western Cape (Avery 1974). Three fragments of non-diagnostic pottery may support a connection to herders, but the evidence is limited. A subsequent use of Check is documented by a radiocarbon date of 1290–1380 cal AD (KIA-17745) on a bone tool, while a Dutch clay pipe (Fig. 10), 18th century Chinese provincial ware and a green glass container demonstrate further examples of post-contact (later than 1652 AD) site use.

Figure 10: Geelbek Dunes. Illustrations of the bowls of Dutch clay pipes from Check, Shelly and Toaster: (1) G-SH-3169; (2) G-CH-234; (3) G-TO-2461 (Kügler 2002) (drawings H. Frey).

Locality Frikkie About 70 m east of Check in a small deflation hollow, Frikkie (Bay 67) was named after the used car salesman who sold us our field vehicle (Fig. 2). Collections in Frikkie occurred mainly in 1999 and 2002 over an area of more than 1200 m2. This rectangular shaped, 25 by 35 m locality moved an average of 7 m to the north over the years of study. The find density in Frikkie was very low, and none were diagnostic of a specific time period, with the exception of a Dutch clay pipe stem. Locality Toaster On the west side of the dune field, the locality Toaster (Bay 27) is located 200 m northwest of Stella (Fig. 2). Collections in this deep, linear valley between high transverse dunes occurred mainly between 2000 and 2002, but continued through 2006. The locality owes its name to a home appliance that the project director asked for, but never got. Collections in Toaster covered over 3900 m2 in an east-west oriented, 100 by 45 m deflation bay. Movement to the north averaged 8.5 m each year. The basal calcrete layer exposed in Toaster yielded U-series dates between 211–228 ka. With the exception of locality Stonering, these ages are older than all other calcretes at Geelbek. A younger episode of carbonate mobilization is recorded within the same calcrete at 122–133 ka. Finds from both MSA and LSA contexts are present. MSA artefacts include one inclined core, one parallel core, and two denticulates. Well-preserved, highly-mineralised faunal remains include rhinoceros, large bovids and equids, despite Toaster’s low average

state of mineralization (1.31) (Fig. 4). The signal from 60 heavily mineralised bones is overwhelmed by a large assemblage of non-mineralised (MC1) very large mammals (SC6, n = 804). The late LSA is well represented at Toaster. The presence of two backed bladelets, an adze and a drill support this designation, as do five hammerstones and grinders. One unperforated, smooth-edged, oval disk of abalone (H. midae) and the perforated valve of a dog cockle (Glycymeris queketti) represents some rare examples of a shell ornaments at Geelbek. Also interesting in a locality with 174 shells is the presence of 15 (8%) fissurellid keyhole limpets. These naturally perforated shells offer themselves as premade ornaments and they are often associated with the late LSA. Radiocarbon dates on a white mussel scraper gave a result of 30–200 cal AD (KIA-17749), and a large, finished ostrich eggshell bead dated to 870–1140 cal AD (KIA-17760). Twelve large ostrich eggshell beads and one small bead representing later stages of production are usually considered as indicators of LSA herders (Yates 1995, Kandel & Conard 2005). These finds are associated with 418 pottery fragments, the highest number at Geelbek, some of which are red in colour and incised with parallel patterns, confirming an association with herders (Sampson & Sadr 1999). The locality Toaster also yielded 24 fragments of bright red ochre. The identification of thirteen sheep-sized post-cranial remains is suggestive of herding, but none of the finds could be identified to species level. Stone features in Toaster include at least 13 roasting platforms associated with a rich concentration of very large (SC6) burnt bones. In fact, the numerous fireplaces may be similar to the occurrence of such features at the locality Olifant. At Toaster, the almost complete remains of a male elephant with only its tusks missing were found amongst eight slugs of poured lead gunshot. Remains of two Dutch pipes (Fig. 10), a European gunflint and a hand wrought, iron stake and hook round out this post-contact (later than 1652 AD) assemblage. These finds document a clear instance of elephant hunting. It seems the tusks were harvested, but the animal itself was left behind. We interpret the 13 roasting platforms and the resulting burnt faunal remains as evidence for the processing of this downed elephant. The elephant bone was tested for isotopes by F. Radloff of Stellenbosch University, who reported in August 2008 that a diet rich in C4 grasses accounted for ca. 10% of its food, with the rest of its diet composed of either C3 grasses or browse (pers. comm.) Locality Shelly Located 120 m northwest of Toaster, Shelly (Bay 26) is situated in a deep, circular valley in between high transverse dunes (Fig. 2). The locality Shelly was named for the frequent whole marine shells observed on its surface. Collections were made mainly between 2000 and 2002 but continued through 2006 over an area of more


40

than 5300 m2. This 70 m diameter bay moved north at a rate of 8.0 m per year. Thanks to the exposure of geological unit AD2 across almost the entire locality, Shelly, like other localities such as Pottery and Nora (see below), is one of a few exclusively late LSA sites at Geelbek. Based on IRSL, AD2 dates to 6 ± 1 ka in the locality Pottery, 300 m to the north. The age for AD2 is consistent with the results from Alice and Stella.

Figure 11: Geelbek Dunes. Illustration of shell tools made on

white mussel (Donax serra) shells from Equus, Nora and Shelly. The regularity and direction of retouch supports their

interpretation as scrapers: (1) G-SH-2272; (2) G-SH-1658; (3) G-SH-1653; (4) G-SH-1473; (5) G-SH-2590; (6) G-SH-1251; (7) G-NO-951; (8) G-EQ-485; (9) G-EQ-2508 (drawings S.

Feine). Radiocarbon dates of 720–510 cal BC (KIA-17748) on a white mussel scraper (Fig. 11), 620–400 cal BC (Pta-8688) on a limpet (Cymbula granatina), and 570–370 cal BC (Pta-8689) on black mussel overlap, suggesting a single occupational phase at Shelly. A small ostrich eggshell bead dating to 480–170 cal BC (KIA-17757) partly overlaps with this phase. The lithic assemblage is strictly late LSA with 12 backed pieces, five small scrapers, four adzes and a spokeshave. Ground stone tools are significant at Shelly with a total of 48, including mostly hammerstones and grinders, as well as four anvils and three knappers (Fig. 7). There are 24 ochre fragments present. Small ostrich eggshell beads are also significant, with 220 present at all stages of production. In contrast to locality Toaster, all of the beads are small, which suggests a late hunter-gatherer affinity (Yates 1995, Kandel & Conard 2005). One fragment of a bone tool did not contain sufficient collagen for radiocarbon dating (KIA-17747), but appears to be part of a linkshaft (Fig. 8). Despite the presence of more than 4.5 kg of marine shell from a wide variety of species more than 5 km from the shore, the total weight of shellfish (Buchanan 1988) amounts to no more than the modern daily calorie allowance of 2000 kcal for one individual. Of the 1091

shells from Shelly, ten (1%) are fissurellid keyhole limpets which may have served as ornaments. Ninety-six (30%) of the 327 white mussel shells are retouched into scrapers, showing their utility not only as a food resource, but also as a tool. The association of a single ulna of a Cape fur seal (Arctocephalus pusillus) within a scatter of ostrich eggshell beads and lithics suggests that LSA people exploited this marine resource as well. Six roasting platforms are present in Shelly and seem to be located in places where finds are absent (Fig. 12). This pattern was also noted at locality Check (cf. Avery 1974.) There is no pottery in Shelly. Modern finds include a Dutch clay pipe (Fig. 10), plus a harness buckle and a spur of unknown age, perhaps representing an equestrian mishap. Locality Rhino Located 250 m north of Shelly, the locality Rhino (Bay 22) was named for the heavily mineralised (MC5) remains of at least two black rhinoceros (Diceros bicornis) that were first observed here (Fig. 2). Rhino is situated on the western end of a long, transverse dune valley which contains the next string of localities, namely Pottery, Snoek, Ovis and Matilda Rose. Collections in Rhino occurred mainly from 1999 to 2002 and continued through 2004 over an area of more than 3500 m2. This elongate, east-west oriented bay 110 by 20 m moved north at an average of 7.0 m and a maximum of 12 m per year. The exposed hardpan calcrete yielded U-series dates between 67–83 ka while the overlying sand gave a consistent IRSL date of 61 ± 5 ka. A radiocarbon date on a Trigonephrus land snails eroded from the calcrete was considered to be a minimum age of 40,890 ± 1060 uncal BP (GrA-13527). IRSL dating of AD1 resulted in an age of 10 ± 1 ka, and in Pottery, the neighbouring site to the east, AD2 was IRSL dated to 6 ± 1 ka. In locality Rhino, the direct date of an eland (Taurotragus oryx) metacarpal of MC3 gave a result of 5770–5520 cal BC (Pta-8770). Further dates include a fissurellid keyhole limpet dated to 1670–1490 cal BC (Beta-240264) and Trigonephrus land snails collected from the surface that dated to 690–770 cal AD (Pta-7949). Two Howiesons Poort silcrete segments were documented in Rhino and distinguish themselves from LSA segments through their typology and technology (Deacon 1995, Wadley & Mohapi 2008). Other diagnostic MSA lithics were not documented. A concentration of refitting quartz porphyry artefacts, including two large scrapers with maximum dimension of 109 and 150 mm is spatially associated with the two rhinoceros. It was not possible to date the rhinoceros remains to a specific time period. The remaining lithics from Rhino have a strong late LSA character. The most diagnostic pieces of the late LSA assemblage are 11 backed pieces and one drill. The two


41

Figure 12: Geelbek Dunes. Map showing the distribution of finds and features in the locality Shelly. Radiocarbon ages of shells and OES beads date occupation to the middle of the first millennium cal BC. These man-made stone features cluster in areas where find

density is very low. While most features are burnt and interpreted as roasting platforms, one is unburnt and may represent an unused roasting platform, a wind break or a working surface. We interpret this site as a living area where people knapped lithics and made

OES beads around several fireplaces where they also ate shellfish. ground stone tools are both hammerstones and upper grinders. The less mineralised (MC1 and MC2) fauna in Rhino cannot be placed in time, but must be younger than AD2 based on their preservation and stratigraphic association. Locality Pottery The locality immediately east of Rhino was named Pottery (Bay 23) after a large concentration of prehistoric potsherds (Fig. 2). Pottery is divided into two separate areas separated by a steep, small valley. The western one is elongate and 100 by 30 m, while the eastern one is rectangular and 70 by 60 m. Field activities began in 1999 and continued through 2004 covering an area of more than 7600 m2. The locality moved north at a rate of 6.0 m per year and in 2004 began merging with Rhino to the west. Thanks to the exposure of geological unit AD2 across the entire locality, Pottery joins Shelly and Nora as one of a few, completely late LSA sites at Geelbek. AD2 produced IRSL dates of 6 ± 1 ka, consistent with results from Alice and Stella. Dates from the locality Pottery are similar to those from Shelly, just 300 m to the south, and the localities share many other similarities as well. A black mussel shell dated to 730–480 cal BC (Pta-8691) and a whale barnacle (Coronula diadema) dated to 620–390 cal BC (GrA-28388) appear to belong to one occupational phase, as

does a small ostrich eggshell bead dated to 530-220 cal BC (KIA-17761). A white mussel scraper dated to 360–200 cal BC (KIA-17751) partly overlaps with this phase of occupation. The small, late LSA lithic assemblage of four diagnostic pieces includes one adze, one small tanged unifacial point and two small scrapers (Deacon 1984). Fourteen ground stone tools include two complete bored stones, several hammerstones and grinders, and two knappers (Fig. 7). Small ostrich eggshell beads are significant at Pottery with 614 present in all stages of production and document a preference for burnt beads. Small finished beads, defined as smaller than 5 mm in diameter, are often associated with occupation by hunter-gatherers (Yates 1995, Kandel & Conard 2005). Despite the presence of more than 0.5 kg of marine shell from a wide variety of species more than 5 km from the shore, the total weight of shellfish (Buchanan 1988) amounts to just a ninth of the modern daily calorie allowance of 2000 kcal for one individual. Thirteen (19%) of 67 white mussel shells are retouched into scrapers showing their further utility as tools. A distinct concentration of burnt calcrete blocks signifies a single roasting platform surrounded by a clear concentration of lithics, bone and shell (Kandel & Conard 2003). This pattern contrasts with roasting pits lacking finds, and therefore represents a second category of features, which Avery (1974) described. Distributed amongst the finds surrounding the roasting platform are the remains of 35


42

whale barnacles (C. diadema), indicating that this was a place where humpback whale meat or blubber was prepared (Kandel & Conard 2003). Two unburnt concentrations of calcrete denote a third type of feature in Pottery. Although they could represent unused roasting platforms, their unusual crescent shape suggests another function. Such forms may indicate weights for a former shelter, such as those used by Nama groups in South Africa and Namibia to secure their huts (MacCalman & Grobbelaar 1965, Avery 1974). The authors have often seen examples of similar features left behind by Bedouin herders in Syria. A fourth type of feature consists of three, 30–50 cm large, flat, unburnt calcrete blocks arranged next to each other that may have served as hard working surfaces on the soft sandy substrate of AD2. These finds are associated with 78 pottery fragments, many of which refit into three-quarters of the mid-section of a plain large vessel about 28 cm in diameter. These kinds of pots are usually associated with the first LSA herders (Sampson & Sadr 1999). However, all of the radiocarbon dates for the main settlement period at locality Pottery predate the advent of pottery in the Western Cape at about 2000 BP (Sealy & Yates 1994, Henshilwood 1996, Vogel et al. 1997). Therefore, the pottery clearly represents a later phase of occupation. M. Copley at the University of Bristol tested nine of the potsherds for absorbed lipid residues and reported in December 2004 that marine derived fats were present in two samples and unidentified lipid sources in five others (pers. comm.). Locality Snoek In the same deflation valley 350 m east of Pottery, the locality Snoek (Bay 57) owes its name to the previous evening’s delicious fish braai (barbecue in Afrikaans) (Fig. 2). Artefacts from Snoek were collected mainly between 1999 and 2002 and continued through 2006 over an area of more than 4600 m2. This oval, northwest-southeast oriented 90 by 50 m bay moved northeast at an average of 2.7 m, considerably below the average movement of 7.23 m/yr observed in the dunes. Snoek is situated in one of the deepest valleys of the dunes and was perhaps more sheltered from dune movement. The undated calcrete that outcrops in Snoek may correlate with the upper calcrete of its northern neighbour locality Stonering (see below), which yielded a U-series date of 109–119 ka. The reworked sand that covers the calcrete in Snoek dated to 1.15 ± 0.10 ka using OSL. These data fit with radiocarbon dates of several finds at Snoek. Two examples of perforated ostrich eggshell dated between 1230–1460 cal AD (KIA-17762) and 1310–1540 cal AD (KIA-17759). Such finds may represent ostrich eggshell containers, but without further evidence, are more likely examples of hyena scavenged ostrich eggshell (Kandel 2004). The remains of two large (SC4) unmineralised (MC1) elands, one adult and one juvenile, appear contemporaneous, dating to 1430–1610 cal AD (Pta-8406) and 1450–1620 cal AD (Pta-8768), while a large ostrich eggshell bead postdates the elands at

1640–1910 cal AD (KIA-17758). Snoek was remarkable for its 30 backed lithic pieces, the largest collection from a single locality at Geelbek. The question arises as to why so many tools are present in one place. This toolkit is typically associated with late LSA hunter-gatherers (Deacon 1984). Yet the high number of tools suggests that Snoek is more than an example of the “one that got away” (Yellen 1977). Instead of a failed hunting attempt, Snoek represents a camp where retooling occurred. Two ground stone tools, a knapper and a lower grinder, complete the lithic assemblage. Snoek yielded 45 shells, 18 of which were white mussels. One of these was retouched into a shell scraper. Perhaps more interesting in a locality with so few shells was the presence of 14 (31%) keyhole limpets which may have been used as ornaments. Locality Ovis To the east of Snoek lies Bay 58, one of the largest Geelbek localities measuring approximately 200 by 50 m (Fig. 2). Within Bay 58, we designated the locality Ovis as a single scatter of a modern sheep skeleton. Unlike all other Geelbek localities, we collected only selected material from Ovis in a limited area of just over 80 m2 in 2002. Thus, Ovis represents a spatial and taphonomic study of what the skeleton of a single animal looks like after it died in the dunes. The rate of movement could not be calculated because study was limited to one year. Locality Mathilda Rose At the eastern end of this transverse valley, we find the locality Mathilda Rose (Bay 59) situated 250 m east of Ovis (Fig. 2). Mathilda Rose is named in honour of the birth of the project director’s youngest daughter. Finds were collected in 1999 and 2006 and covered an area of more than 9600 m2. Mathilda Rose started out as an elongate bay 170 by 50 m but shrunk over time, ending up as two separate oval areas of 80 by 40 m and 50 by 20 m. Dune movement to the north averaged 9.3 m/yr. We correlate the undated calcrete that outcrops in Mathilda Rose with the upper calcrete of its northern neighbour Equus, which yielded a U-series date of 105–111 ka, and with its western neighbour Stonering, which yielded a U-series date of 109–119 ka. No further radiometric dates are available. The diagnostic lithic inventory of Mathilda Rose include three bifacially retouched points that have an affinity to certain MSA artefacts, as well as three backed microliths that fit with a late LSA designation. Mathilda Rose distinguishes itself as the locality with the most post-contact finds (n=52) including a coin from 1843 AD, 19th century British stoneware, a brass tinderbox and a European gunflint. Locality Stonering About 200 m northwest of Mathilda Rose and 100 m directly north of Snoek, the locality Stonering (Bay 79) was named after a complete bored stone found on the


43

western edge of the locality in association with a scatter of prehistoric potsherds (Fig. 2). Finds were collected in 2000 and covered an area of more than 8300 m2. Stonering started out as an elongate bay of 190 by 40 m but grew and separated over time, ending up as two elongate areas of 120 by 45 m and 100 by 15 m. Dune movement to the north averaged 4.5 m, considerably lower than the average movement of 7.0 m per year observed throughout the dunes. Like Snoek, Stonering is situated in one of the deepest valleys of the dunes and was perhaps more sheltered from dune movement. The deeper of two calcretes in Stonering yielded U-series dates of 241–251 ka, while the upper calcrete gave dates of 109–119 ka. It should be noted that the lower calcrete is not exposed at the surface in Stonering, but was only reached by excavating a 2 m pit through zones of the upper calcrete layer. While an ancient dune (AD0) overlying the upper calcrete gave a date of 57 ± 4 ka, no further dates are available from Stonering. One backed bladelet, three scrapers and three ground stone tools, including the eponymous bored stone and two grinders, comprise the diagnostic lithic artefacts and suggest a late LSA age. These finds are associated with 66 pottery fragments, 31 of which refit into a small, lugged and spouted, pointed-base vessel 15 cm high and 17 cm in diameter (Fig. 13). These kind of pots are usually associated with LSA herders (Sampson & Sadr 1999). Next to Matilda Rose, Stonering distinguishes itself as the locality with the second highest number of post-contact finds (n=45) including fragments of a 19th century British stoneware vessel and 41 glass shards.

Figure 13: Geelbek Dunes. Illustration of a small, refitted, lugged and spouted, pointed-base vessel from Stonering

(Conard & Kandel 2006) (drawing S. Feine). Locality Equus Located 200 m northeast of Stonering and 150 m north of Mathilda Rose, the locality Equus (Bay 77) was named after numerous heavily mineralised (MC5) remains of Cape zebra (E. capensis) (Fig. 2). Finds were collected mainly in the years 2000–2002 and continued through

2006 covering more than 10 800 m2. Equus is an elongate bay 200 by 30 m and moved north at an average rate of 9.0 m per year. The deeper of two calcretes in Equus yielded U-series dates of 138–150 ka, while the upper calcrete gave dates of 105–111 ka. It should be noted that the lower calcrete is not exposed on the surface in Equus, but was only reached after excavating a 1.0 m pit. The upper calcrete provides a maximum age for the finds in Equus. Two Howiesons Poort silcrete segments documented in Equus distinguish themselves from LSA segments through their typology and technology. These and two cores, one parallel and one inclined (Conard et al. 2004) provide evidence for MSA occupations of the site. The average mineralization at Equus (4.30) is the highest observed in the Geelbek Dunes (Fig. 4). Mineralised (MC4 and MC5) fauna account for 86% of the assemblage, and large animals such as black rhinoceros (D. bicornis), large bovids and equids comprise 88% of the mineralised remains. Calcrete encrusted ostrich eggshell fragments were common in Equus (n=1304) with 76 (6%) showing signs of perforation. Two radiocarbon dates on perforated ostrich eggshell fragments gave minimum ages of 37 050 ± 310 uncal BP (GrA-19666) and 44 600 + 3500/ -2400 uncal BP (Pta-8382, uneven error margins). Such finds may represent ostrich eggshell containers, but without further evidence, are more likely examples of hyena scavenged ostrich eggshell (Kandel 2004). LSA occupation is documented through four backed microlithic pieces, an adze, a bored stone fragment, one combination hammerstone and grinder (Deacon 1984), and two white mussel scrapers (Fig. 11). The radiocarbon age of a burnt Argenville’s limpet (Scutellestra argenvillei) dates an occupation to 1500–1650 cal AD (Beta-240265). A single piece of non-diagnostic pottery may support a connection to herders, but the evidence here is limited. Beside Mathilda Rose and Stonering, Equus distinguishes itself as the locality with the third highest number of post-contact finds (n=44) including mostly pieces of mid-to-late 19th century British ironstone. Locality Loop Located in the same transverse valley 100 m east of Equus on the north-eastern edge of the dunes (Fig. 2), Loop (Bay 76) describes what airplanes from the Langebaanweg military flight school performed over the locality as we began work here in 2000, that is, loop-de-loops. From this circular deflation hollow about 80 m in diameter we collected finds through 2002 and covered more than 5500 m2. During this time Loop moved north at an average rate of 7 m/yr. Similar to Equus, the average mineralization at Loop is the second highest (4.17) observed in the Geelbek Dunes (Fig. 4). Mineralised (MC4 and MC5) fauna account for 80% of the assemblage, and large animals such as rhinoceros, large bovids and equids comprise 79% of the mineralised remains.


44

The lithic assemblage from Loop is the third largest at Geelbek and contains 19 microliths, among them ten backed points, three backed bladelets and six segments. Eleven side and end scrapers and one adze complete this assemblage. A total of seven ground stone tools include two knappers, two anvils and three combination hammerstones and grinders (Fig. 7). Nineteen (42%) of 45 cores are bipolar on quartz. Taken together, these artefacts suggest a strong late LSA component at Loop. A radiocarbon date on a large ostrich eggshell bead resulted in an age of 1930–1670 cal BC (KIA-17756), which is older than expected for beads larger than 5 mm (Yates 1995, Kandel & Conard 2005). In Loop, four large ostrich eggshell beads and three small beads were found in several stages of production. Based on such small numbers of beads, no conclusion can be drawn about their manufacturers (Yates 1995, Kandel & Conard 2005). Also interesting in a locality with 106 shells is the presence of eight (8%) fissurellid keyhole limpets. A human radius was radiocarbon dated to 40 cal BC–60 cal AD (GrA-17565) and yielded a δ15N value of +16.7‰ (UCT-8071). According to communication from J. Sealy in May 2001, this is the most positive value from the West Coast region and suggests a diet that included large quantities of marine resources and some C4 based terrestrial foods (pers. comm.) This result compares with the most marine rich diets of the South Coast (Sealy 2006). Seventeen pieces of non-diagnostic pottery may support a connection to herders, but the evidence is limited. Finally, 14 post-contact finds from Loop include a European gunflint, gunshot and shards of glass. Locality Crow Crow (Bay 92) is situated 100 m north of Equus in the northeast corner of the Geelbek Dunes on the east end of the northernmost transverse dune (Fig. 2). Crow was named for the frequent visits of corvid avifauna circling and calling overhead. Finds were collected mainly between 2000 and 2002 covering an area of more than 7600 m2. Crow is an irregular shaped, elongate bay 150 by 60 m that moved at an average northward rate of 8.0 m/yr. While radiometric dates are not available, the proximity to the well-dated sequence at site Equus serves as a reasonable guideline. Similar to nearby Equus and Loop, the average mineralization at Crow is the fourth highest (3.38) observed in the Geelbek Dunes (Fig. 4). Mineralised (MC4 and MC5) fauna account for 70% of the assemblage, and large animals such as rhinoceros, large bovids and equids comprise 78% of the mineralised remains. The lithic assemblage from Crow contains four microliths, among them three segments and one backed point, plus a side scraper and an end scraper. A total of six ground stone tools include four hammerstones, one knapping tool and one upper grinder. Six (40%) of 15 cores are bipolar on quartz. Taken together, these artefacts indicate a significant late LSA use of the locality. A radiocarbon date on a turban shell operculum

(T. cidaris cidaris) gave a result of 160–350 cal AD (Beta-240262). Most of the 15 bone tools preserved at Geelbek come from Crow (Fig. 8). These include nine polished bone tools, interpreted as linkshafts, and indicate a later phase of occupation based on a radiocarbon date of 1400–1430 cal AD (KIA-17746). Two fragments of non-diagnostic pottery may support a connection to herders, but the evidence is limited. Finally, seven post-contact finds from Crow include a Dutch clay pipe and pieces of glass and metal. Locality Nora The last locality that we collected is Nora (Bay 88), situated in the northwest of the dunes 500 m west of Crow and 350 m northwest of Stonering (Fig. 2). Nora is named after the sister of a former graduate student, Mr. T. Prindiville. Finds were collected mainly between the years 2000–2002 covering almost 1800 m2. Nora is a circular bay about 50 m in diameter that moved north at an average rate of 5.0 m per year. This area of the dunes is noticeably different from the long transverse valleys to the south. Here, on the leading edge of the dunes, the landscape is hummocky, often covered with vegetation and containing few deflation hollows. This area represents a state before the large transverse dunes move north and begin to deflate new sites. Due to the exposure of only geological units AD1 and AD2 across the entire locality, Nora, like Shelly and locality Pottery, is one of a handful of exclusively late LSA sites at Geelbek. Although no sediment dates are directly available, about 500 m to the south, AD1 dates to 10 ± 1 ka in Rhino and AD2 dates to 6 ± 1 ka in Pottery. Given the consistency of the dates for AD1 and AD2 across the entire dune field from Alice, Stella and Rhino, this correlation is reasonable. Radiocarbon dates match well with those from Nora’s sister localities, Pottery and Shelly. A white mussel scraper dates to 730–530 cal BC (KIA-17750), a black mussel dates to 730–540 cal BC (Pta-8695), and a small ostrich eggshell bead dates to 630–350 cal BC (KIA-17755). Lithic tools include five backed microliths, two adzes, a small unifacial point with a weakly developed tang and two small scrapers (Deacon 1984). Four ground stone tools include a knapper, two upper grinders and a lower grinder. Four (80%) of five cores are bipolar on quartz, indicating a significant late LSA component. Despite the presence of more than 1.3 kg of marine shell from a wide variety of species more than 5 km from the shore, the total weight of shellfish (Buchanan 1988) amounts to just a quarter of the modern daily calorie allowance of 2000 kcal for one individual. Twenty (35%) of 57 white mussel shells are retouched into scrapers showing their further utility as tools (Fig. 11). One white mussel shell shows traces of ochre staining in a bay where seven ochre fragments were recovered. A distinct concentration of burnt calcrete blocks signifies a single roasting platform surrounded by a clear


45

concentration of lithics, bone and shell (Kandel & Conard 2003). This pattern differs from the roasting pits that fill areas where finds are absent and represents another type of feature described by Avery (1974). About 15 m away, a single whale barnacle fragment (C. diadema) suggests that whale meat or blubber may have been prepared here. The evidence from Nora is less convincing than from locality Pottery, but still it is unusual to find whale barnacles at all. Since we know of no non-anthropogenic means of transporting whale barnacles 5 km from the ocean, we attribute this find to human agency. Thus, we argue that humpback whale was also processed and/or eaten at Nora (Kandel & Conard 2003). These finds are associated with 38 pottery fragments. Two pieces refit, and with the exception of three lug fragments, few pieces were diagnostic. Eight pottery fragments have red exteriors and may be related to the seven pieces of ochre found. If this is a simple lugged vessel, it may be similar to the one from locality Pottery. Lugged pots are usually associated with LSA herders (Sampson & Sadr 1999). Since all of the radiocarbon dates for the main settlement phase at Nora predate the introduction of pottery in the Western Cape at about 2000 BP, the pottery likely represents a later phase of occupation. Finally, ten post-contact finds from Nora include a European gunflint, some metal finds and a mid-to-late 19th century ironstone fragment. Discussion The data collected from the Geelbek Dunes are voluminous. After five seasons of fieldwork and several years of analysis, we can make sense out of this mountain of information. To begin with, we can examine the summary of pertinent data (Tables 4 & 5) to assess some of the parameters that we use to analyse settlement patterns at Geelbek during the MSA, LSA and post-contact period. It is clear that without the action of the wind and the movement of the dunes, the finds at Geelbek would not have been exposed. However, it is important to separate the modern dune field from the palaeo-landscape. Compton & Franceschini (2005) estimate that the modern dunes began their trajectory towards Geelbek about 6000 years ago during a high stand of sea level. Our recent measurements of dune movement support this hypothesis and demonstrate that the modern dunes migrated more than 700 m across the underlying landscape during the last 1000 years alone. Thus, during the main phases of occupation at many of the Geelbek localities, the modern dunes had not yet arrived. We see the archaeological and paleontological occurrences at Geelbek as chance glimpses into the past. The assemblages are examples of finds scattered on the landscape in the broadest sense, not by the action of the wind or through the process of deflation, but simply because materials have been preferentially exposed in certain places and not in others. In other words, at Geelbek the landscape is the site (Foley 1981). The

localities at Geelbek show views of specific activities that happened over long expanses of time away from major foci of occupation. This landscape approach brings with it strengths as well as limitations. While cultural remains such as lithic artefacts, worked bone, ostrich eggshell beads, ornaments, marine shell, pottery and stone features are clear indicators of human activities, the accumulators of terrestrial fauna including bone and ostrich eggshell are harder to nail down. In this open-air environment, animals die naturally, and scavengers transport their bones. While burning may suggest a human accumulator, brush fires cannot be ruled out. Ostrich eggshell can tell us of ostrich nests, hyena scavengers or human activity. The presence of fish on a site may result from the activities of carnivores, scavengers, birds or humans. For these reasons, we tend to steer away from interpretations that rely solely on these types of faunal assemblages, unless unambiguous signs of human activity exist. This constraint does not apply to marine shells because they offer a clear indication that humans accumulated them. MSA The data confirm that Geelbek was occupied during the MSA, but the settlement pattern we observe is diffuse. Scattered across a broad landscape are 96 cores and tools diagnostic of the MSA, including segments, bifacial points, and parallel and inclined cores. Limited evidence suggests an association between the artefacts and the mineralised (MC4 and MC5) faunal assemblages. These fauna shows low frequencies of anthropogenic modification such as burning, green breaks, cut marks and impact fractures, supporting that people were partly responsible for their accumulation. Radiocarbon dating of ostrich eggshell shows that some of the eggshell is older than 40 000 BP, but like the terrestrial fauna, the old ostrich eggshell does not positively correlate with the distribution of lithic artefacts. Perforated pieces of old ostrich eggshell do not confirm human modification, as these holes can result from hyena predation (Kandel 2004). The clearest MSA signature comes from three localities, Equus, Rhino and Stella, each of which yielded two Howiesons Poort segments (Dietl et al. 2005). Two localities, Homo and Matilda Rose, yielded bifacially retouched points suggestive of the Still Bay phase of the MSA, and Homo added one unifacially retouched point to the collection. Localities with parallel and inclined cores attributed to the MSA include Equus, Homo, Snoek, Stella and Toaster. Dating the MSA finds offers a significant challenge. While an excellent geological framework exists through the dating of calcretes and ancient dunes, these generally offer large brackets of time. The dates provide a maximum age for the assemblages lying upon them. Mineralised remains of MC4 and MC5 are clearly fossils (Conard et al. 2008), but their exact age is difficult to assess. The presence of certain extinct species such as E.


46

capensis confirms the antiquity of the assemblage, but not the specific age. The best chance to ascribe a date range to MSA finds comes from the Howiesons Poort segments and bifacial points. However, these types of artefacts do not only occur in the narrow timeframes that these MSA sub-phases represent (Jacobs et al. 2008, Wadley & Mohapi 2008, Tribolo et al. 2009, Porraz & Roberts 2010). Thus, we interpret the MSA data as suggesting that population density was low relative to the LSA occupation of the area. Based on the distances raw materials were transported, the MSA groups that lived here were highly mobile. The signature that these people left behind on the landscape was very diffuse and likely fluctuated through time. LSA In comparison to the MSA signature at Geelbek, the LSA localities have well defined distributions of artefacts, and stone features are frequent. Many of the scatters of LSA artefacts appear to be intact. Artefacts cluster in specific areas, and their spatial distribution suggests minimal displacement. This observation is consistent with the results from GOME, in which we found that assemblages of artefacts move together as a package when eroded by wind. The impact of deflation on the assemblages is not enough to scatter the package (Kandel et al. 2003). Most of the LSA occupations occurred within the last 4000 years (Fig. 6). A typical late LSA assemblage at Geelbek contains backed microlithic artefacts, mainly segments, points, and bladelets, as well as scrapers and adzes. These are present amongst a backdrop of ground stone tools, including knappers, hammerstones, upper and lower grinders, and anvils. Clustered accumulations of fauna, marine shell and ostrich eggshell beads are often found in association with burnt stone features, for example at localities Check, Homo/Bay 35, Nora, Pottery, Shelly and Toaster. The role of terrestrial fauna in the diets of the inhabitants of Geelbek is difficult to assess because multiple modes of accumulation must always be considered. Natural forces such as brush fire, scavengers and random deaths can rarely be overruled. However, in a few localities including localities Check, Homo/Bay 35, Nora, Pottery, Shelly and Toaster, clear concentrations of terrestrial faunal remains cluster near stone features. This spatial patterning strongly suggests that humans accumulated the fauna. Bone tools at Check, Crow, Homo, Shelly and Toaster appear to be linkshafts that would have been part of the hunting toolkit. The use of marine resources is documented at several localities and reliably indicates occupation. Shellfish are most common at Check, Loop, Homo/Bay 35, Nora, Pottery, Shelly and Toaster. Nonetheless, the amount of shellfish consumed is small compared to shell middens located directly on the coast (e.g. Parkington et al. 1992). This suggests that late LSA people carried small portions of shellfish away from the coast, perhaps to supplement their diet. Shell scrapers made on white mussels are

associated with areas of greater settlement activity, for example at Nora, Pottery and Shelly. Rare shell ornaments include a rounded disk of abalone and a few perforated gastropods and molluscs. Fissurellid keyhole limpets are present in greater numbers than expected for the amount of shell material present, and may represent ready-made shell ornaments (Hall & Binneman 1987). Another use of marine resources can be found in the presence of whale barnacles at Nora and Pottery. These crustaceans attest to the use of beached whales from the coast (Kandel & Conard 2003). A single ulna from a Cape fur seal in Shelly represents the only marine mammal remain from Geelbek. Found amidst a dense scatter of ostrich eggshell beads and lithics artefacts, its clear association with human activity supports that seal was incorporated into the subsistence patterns of late LSA people (e.g. Sadr et al. 2003). However, scavengers cannot be ruled out as the accumulator of occasional fish remains. The manufacture of small ostrich eggshell beads less than 5 mm in diameter played an important role at Nora, Pottery and Shelly. We observed complete chains of production at these localities, suggesting occupations of longer duration (Kandel & Conard 2005). Beads and marine shell from these localities date to the middle of the first millennium cal BC and form the strongest argument that these sites were part of the same settlement system. Stone features (Avery 1974) are most common at Olifant, Pottery, Shelly and Toaster and represent focal points of activity. Most examples at Geelbek are accumulations of burnt calcrete that we interpret as roasting platforms on which people cooked food. At Check and Shelly roasting platforms are located where other finds are absent, while at Homo, Nora and Pottery, artefacts cluster around the features. These examples suggest a structured use of space. Unburnt features are less common and may represent unused roasting platforms (Shelly), weights to hold down windbreaks or huts (Pottery), or stable working platforms on the loose, sandy surface (Pottery). This spatial patterning indicates places around which specific activities occurred, such as the roasting of whale meat, the production of ostrich eggshell beads or the knapping of lithics. When pottery is present in a locality, it often refits and represents part of one broken vessel. Scatters of pottery often spatially overlap with late LSA artefacts that date to the first millennium cal BC, for example at Nora and Pottery. However, potsherds are unlikely to be directly associated with these assemblages because the technology had not reached this region at this time (e.g. Henshilwood 1996, Sadr 2004, Webley 2007). Human remains from an LSA context were found in the localities Homo, Hetero and Loop. The human bones were fragmentary, not representing complete individuals, and younger than many of the find categories that date to first millennium cal BC. Isotopic analyses of these


47

individuals suggests that a marine diet was important, an idea echoed in the discovery of marine lipids adhering to some fragments of pottery. These remains may represent disturbed burials, but the evidence is not clear. Post-contact We consider the post-contact period to begin in 1652 AD with the arrival of permanent European settlement in the Cape, although first contact with the Portuguese occurred earlier in Saldanha Bay just north of Langebaan Lagoon. Under this rubric we include finds of historic ceramics, ironstone, china, Dutch clay pipes, European flints, gunshot, coins, glass and metal indentified by Antonia Malan and Jane Klose associated with UCT. Although the post-contact finds are relatively few, they place the occupation of Geelbek into a greater narrative and document the continued use of the near-coastal environment into the historic period. Our research has documented a broad range of post-contact activities including elephant hunting at locality Toaster, animal husbandry at Check, as well as, cooking, dining, drinking, shooting and smoking at many localities in the strandveld. Conclusions The MSA and LSA sites at Geelbek indicate brief periods of occupation and highlight specific activities that occurred at different points on the landscape. Geelbek is situated where people could exploit the diversity of the near-coastal environment, hunting and trapping terrestrial animals, as well as gathering coastal resources. Lithic raw materials came from up to 25 km away. A wide variety of activities occurred, for example, people knapped stone artefacts, retooled for hunting, ground ochre and plant material, roasted whale meat, rendered blubber, manufactured ostrich eggshell beads and shell ornaments, collected and ate shellfish, and buried their dead. The signatures of these activities are preserved at Geelbek and show that traces of human activities during the MSA and LSA are by no means restricted to the caves, rockshelters and shell middens that have long been the focus of archaeological research (cf. Conard et al. 1999). This kind of large-scale research represents one of a few examples (e.g. Parkington et al. 1992) of detailed studies of open-air localities in southern Africa. The data demonstrate how people made use of the landscape as a single entity and were flexible in their ability to exploit all available resources. These snapshots in time help us understand aspects of human behaviour that cannot be extracted from denser horizons of occupation. We hope that with this endeavour into large-scale landscape archaeology, we have demonstrated the utility and potential of this approach. Acknowledgements We are grateful for the years of discussion and hands on experience we gained in South Africa. Without the support of the following people, we could not have

conducted this project: Graham Avery, Margaret Avery, John Compton, Janette Deacon, Holger Dietl, Susanne Feine, Peter Felix-Henningsen, Markus Fuchs, Pippa Haarhoff, Richard Klein, Mary Leslie, John Parkington, Tim Prindiville, Dave Roberts, John Rogers, Judy Sealy, Steven Walker, Stephan Woodborne, Sarah Wurz and Royden Yates. We are grateful to students from Tübingen and Cape Town for their tremendous help in conducting this project. Special thanks are due to the following people who contributed their expertise on a wide variety of topics: Miranda Armour-Chelu (equids), Mark Copley (lipids), Holger Dietl (lithics), Werner Fourie (hyena), Gavin Kanigowsky (ostrich eggshell), Richard Klein (fauna), Jane Klose (post-contact), Martin Kügler (clay pipes), Antonia Malan (post-contact), Briana Pobiner (hyena), Cedric Poggenpoel (fish), Frans Radloff (isotopes), Essi Rönkkö (pottery), Judy Sealy (isotopes), John Spense (hyena), Ralf Vogelsang (ostrich eggshell), Christina Weiss (ostrich eggshell) and Lars Werdelin (carnivores). We are grateful to two anonymous reviewers who invested a significant amount of time to help improve this manuscript. Funding for this research came mainly from the Deutsche Forschungsgemeinschaft (German Research Foundation), with significant contributions from the Department of Early Prehistory and Quaternary Ecology of the University of Tübingen and the Heidelberg Academy of Sciences and Humanities. References Acocks, J.P.H. 1988. Veld Types of South Africa.

Memoirs of the Botanical Survey of South Africa 57. Pretoria: Botanical Research Institute.

Apps, P. (ed.) 2000. Smithers’ Mammals of Southern Africa: A Field Guide. Cape Town: Struik.

Avery, G. 1974. Open Station shell midden sites and associated features from the Pearly Beach area, South-western cape. South African Archaeological Bulletin 29: 104–114.

Avery, G., Kandel, A.W., Klein, R.G., Conard, N.J. & Cruz-Uribe, K. 2004. Tortoises as food and taphonomic elements in palaeo « landscapes ». In: J-P. Brugal & Desse, J. (eds) Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires: XXIVe rencontres internationales d’archéologie et d’histoire d’Antibes: 147–161. Antibes: Editions APDCA.

Behrensmeyer, A.K. 1978. Taphonomic and ecologic information from bone weathering. Paleobiology 4: 150–162.

Brain, C.K. 1974. Some suggested procedures in the analysis of bone accumulations from southern African Quaternary sites. Annals of the Transvaal Museum 29: 1–8.

Branch, G.M., Griffiths, C.L., Branch, M.L. & Beckley, L.E. 2005. Two Oceans: A Guide to the Marine Life of Southern Africa. Cape Town: David Philip.

Bronk Ramsey, C. 2009. Bayesian analysis of radiocarbon dates. Radiocarbon 51 : 337–360.

Buchanan, W. 1988. Shellfish in Prehistoric Diet: Elands Bay, SW Cape Coast, South Africa. Cambridge


48

Monographs in African Archaeology 31. Oxford: British Archaeological Reports International Series 455.

Compton, J. & Franceschini, G. 2005. Holocene geoarchaeology of the Sixteen Mile Beach barrier dunes in the Western Cape, South Africa. Quaternary Research 63: 99–107.

Conard, N.J. & Kandel, A.W. 2006. The economics and settlement dynamics of the later Holocene inhabitants of near coastal environments in the West Coast National Park (South Africa). In: Wotzka, H.-P. (ed.) Grundlegungen. Beiträge zur europäischen und afrikanischen Archäologie für Manfred K.H. Eggert: 329–355. Tübingen: Franke.

Conard, N.J., Prindiville, T.J. & Kandel, A.W. 1999. The 1998 Fieldwork on the Stone Age Archaeology and Paleoecology of the Geelbek Dunes, West Coast National Park, South Africa. South African Field Archaeology 8: 35–45.

Conard, N.J., Soressi, M., Parkington, J.E., Wurz, S. & Yates, R. 2004. A unified lithic taxonomy based on patterns of core reduction. South African Archaeological Bulletin 59: 12–16.

Conard, N.J., Walker, S.J. & Kandel, A.W. 2008. How heating and cooling and wetting and drying can destroy dense faunal elements and lead to differential preservation. Palaeogeography, Palaeoclimatology, Palaeoecology 266: 236–245.

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

Deacon, H.J. 1995. Two late Pleistocene–Holocene depositories from the southern Cape, South Africa, South African Archaeological Bulletin 50: 121–131.

Dewar, G. 2010. Late Holocene burial cluster at Diaz Street Midden, Saldanha Bay, Western Cape, South Africa. South African Archaeological Bulletin 65: 26–34.

Dewar, G. & Pfeiffer, S. 2010. Approaches to estimating marine protein in human collagen for radiocarbon date calibration. Radiocarbon 52: 1611–1625.

Dibble, H.L. & McPherron, S.P. 1996. A Multimedia Companion to The Middle Paleolithic Site of Combe-Capelle Bas (France). Philadelphia: University of Pennsylvania Museum.

Dietl, H., Kandel A.W. & Conard, N.J. 2005. Middle Stone Age settlement and land use at the open-air sites of Geelbek and Anyskop, South Africa. Journal of African Archaeology 3: 233–244.

Eitel, B. 1994. Kalkreiche Decksedimente und Kalkkrustengenerationen in Namibia: Zur Frage der Herkunft und Mobilisierung des Calciumcarbonates. Stuttgarter Geographische Studien 123: 1–193.

Felix-Henningsen, P., Kandel, A.W. & Conard, N.J. 2003. The significance of calcretes and paleosols on ancient dunes of the Western Cape, South Africa, as stratigraphic markers and paleoenvironmental indicators. In: Füleky, G. (ed.) Papers of the 1st International Conference on Soils and Archaeology, Százhalombatta, Hungary, 30 May-3 June 2001. British Archaeological Reports International Series 1163: 45–52. Oxford: Archaeopress.

Foley, R. 1981. Off-site archaeology and human adaptation in Eastern Africa: An analysis of regional artefact density in the Amboseli, southern Kenya. Cambridge Monographs in African Archaeology 3, British Archaeological Reports International Series 97. Oxford: Archaeopress.

Fuchs, M., Kandel, A.W., Conard, N.J., Walker, S.J. & Felix-Henningsen, P. 2008. Geoarchaeological and chronostratigraphical investigations of open-air sites in the Geelbek Dunes, South Africa. Geoarchaeology 23: 425–449.

Hall, S. & Binneman, J. 1987. Later Stone Age burial variability in the Cape: a social interpretation. South African Archaeological Bulletin 42: 140–152.

Henshilwood, C.S. 1996. A revised chronology for pastoralism in southernmost Africa: new evidence of sheep at c. 2000 b.p. from Blombos Cave, South Africa. Antiquity 70: 945–949.

Jacobs, Z., Roberts, R.G. Galbraith, R.F., Deacon, H.J., Grün, R., Mackay, A., Mitchell, P. Vogelsang, R. & Wadley, L. 2008. Ages for the Middle Stone Age of southern Africa: Implications for human behavior and dispersal. Science 322: 733–735.

Kandel, A.W. 2004. Modification of ostrich eggs by carnivores and its bearing on the interpretation of archaeological and paleontological finds. Journal of Archaeological Science 31: 377–391.

Kandel, A.W. & Conard, N.J. 2003. Scavenging and processing of whale meat and blubber by Later Stone Age people of the Geelbek Dunes, Western Cape. South African Archaeological Bulletin 58: 91–93.

Kandel, A.W. & Conard, N.J. 2005. Production sequences of ostrich eggshell beads and settlement dynamics in the Geelbek Dunes of the Western Cape, South Africa. Journal of Archaeological Science 32: 1711–1721.

Kandel, A.W., Felix-Henningsen, P. & Conard, N.J. 2003. An overview of the spatial archaeology of the Geelbek Dunes, Western Cape, South Africa. In: G. Füleky (ed.) Papers of the 1st International Conference on Soils and Archaeology, Százhalombatta, Hungary, 30 May-3 June 2001. British Archaeological Reports International Series 1163: 37–44. Oxford: Archaeopress.

Klein, R.G., Cruz-Uribe, K. & Beaumont, P. 1991. Environmental, ecological, and paleoanthropological implications of the Late Pleistocene mammalian fauna from Equus Cave, Northern Cape Province, South Africa. Quaternary Research 36: 94–119.

Kügler, M. 2002. Tonpfeifen aus Geelbek/RSA. Knasterkopf 15: 98–100.

MacCalman, H.R. & Grobbelaar, B.J. 1965. Preliminary report of two stone-working Ova Tjimba groups in the northern Kakaoveld of the South West Africa. Cimbebasia 13: 1–30.

McCormac, F.G., Reimer, P.J., Hogg, A.G., Higham, T.F.G., Baillie, M.G.L., Palmer, J. & Stuiver, M. 2002. Calibration of the radiocarbon time scale for the southern hemisphere: AD 1850–950. Radiocarbon 44: 641–651.


49

Netterberg, F. 1969. The interpretation of some basic calcrete types. South African Archaeological Bulletin 24: 117–122.

Parkington, J.E., Nilssen, P., Reeler C. & Henshilwood, C. 1992. Making sense of space at Dunefield Midden campsite, Western Cape, South Africa. South African Field Archaeology 2: 63–71.

Porraz, G. & Roberts, D.L. 2010. Preliminary characterization of an MSA lithic assemblage preceding the ‘classic’ Howiesons Poort complex at Diepkloof rock-shelter (Western Cape Province, South Africa). Current Themes in Middle Stone Age Research. Goodwin Series Volume in Honour of Lyn Wadley 10: 105–121.

Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Hajdas, I., Heaton, T. J., Hogg, A.G., Hughen, K.A., Kaiser, K. F., Kromer, B., McCormac, F. G., Manning, S.W., Reimer, R.W., Richards, D.A., Southon, J.R., Talamo, S., Turney, C.S.M., van der Plicht, J. & Weyhenmeyer, C.E. 2009. IntCal09 and Marine09 radiocarbon age calibration curves, 0-50,000 years cal BP. Radiocarbon 51: 1111–1150.

Rogers, J. 1980. First report of the Cenozoic sediments between Cape Town and Eland’s Bay. Report Geological Survey South Africa 165: 1–64.

Sadr, K. 2004. Feasting on Kasteelberg? Early herders on the west coast of South Africa. Before Farming (online version) 3, Article 2: 1–17.

Sadr, K., Smith, A.B., Plug, I., Orton, J. & Mutti, B. 2003. Herders and foragers on Kasteelberg: interim report of excavations 1999–2002. South African Archaeological Bulletin 58: 27–32.

Sampson, C.G. & Sadr, K. 1999. On the size and shape of Later Stone Age fibre tempered vessels in the upper Seacow River valley. Southern African Field Archaeology 8: 3–16.

Sealy, J. 2006. Diet, Mobility, and Settlement Pattern among Holocene Hunter-Gatherers in Southernmost Africa. Current Anthropology 47: 569–595.

Sealy, J.C. & Yates, R. 1994. The chronology of the introduction of pastoralism to the Cape, South Africa. Antiquity 68: 58–67.

Stuiver, M. & Braziunas, T.F. 1993. Modeling atmospheric 14C influences and 14C ages of marine samples to 10,000 BC. Radiocarbon 35:137–189.

Stynder, D.D., Brock, F., Sealy, J.C., Wurz, S., Morris, A.G. & Volman, T.P. 2009. A mid-Holocene AMS 14C date for the presumed upper Pleistocene human skeleton from Peers Cave, South Africa. Journal of Human Evolution 56: 431–434.

Tribolo, C., Mercier, N., Valladas, H., Joron, J.L., Guibert, P., Lefrais, Y., Selo, M., Texier, P.-J., Rigaud, J.-Ph., Porraz, G., Poggenpoel, C., Parkington, J.E., Texier, J.-P. & Lenoble, A. 2009. Thermoluminescence dating of a Stillbay-Howiesons Poort sequence at Diepkloof Rock Shelter (Western Cape, South Africa). Journal of Archaeological Science 36: 730–739.

Vogel, J., Plug, I. & Webley, L. 1997. New dates for the introduction of sheep into South Africa: The evidence from Spoegrivier Cave in Namaqualand. South African Journal of Science 93: 246–48.

Vogel, J.C., Visser E. & Fuls, A. 2001. Suitability of ostrich eggshell for radiocarbon dating. Radiocarbon 43: 133–137.

Wadley, L. & Mohapi, M. 2008. A segment is not a monolith: evidence from the Howiesons Poort of Sibudu, South Africa. Journal of Archaeological Science 35: 2594–2605.

Webley, R. 2007. Archaeological evidence for pastoralist land-use and settlement in Namaqualand over the last 2000 years. Journal of Arid Environments 70: 629–640.

Yates, R. 1995. Appendix B: report on the analysis of ostrich eggshell beads from Geduld. South African Archaeological Bulletin 50: 17–20.

Yellen, J. E. 1977. Archaeological Approaches to the Present. Models for Reconstructing the Past. New York: Academic Press.

Stone Age economics and land use in the Geelbek Dunes

Documents

Transcript of Stone Age economics and land use in the Geelbek Dunes