Missing pieces: Later Stone Age surface assemblages on the … · 2020. 3. 30. · Southern African Later Stone Age (LSA) archaeology has generally focused on rock shelter excavations

65

Southern African Humanities 25: 65–85 September 2013 KwaZulu-Natal Museum

http://www.sahumanities.org.za; ISSN (online): 2305-2791

Missing pieces: Later Stone Age surface assemblages on the greater Mapungubwe landscape, South Africa

Tim ForssmanSchool of Archaeology, St Hugh’s College, University of Oxford, OX2 6LE, United Kingdom;

[email protected]

ABSTRACTNew evidence demonstrates a difference between forager lithics at rock shelters and open-air assemblages on the Greater Mapungubwe Landscape, northern South Africa. This paper examines the relationship between these two contexts. A field survey was conducted with the goal of identifying Later Stone Age surface scatters, and twenty-five of the total assemblages identified were collected and analysed. A comparison between these assemblages and those from five excavated and dated rock shelters showed discontinuities in the forager cultural record of the region. It was found that two assemblage types exist on the landscape, one dominated by cryptocrystalline silicates, the other by quartz. Possible factors that may have produced these different assemblage types include access to raw material, special-purpose sites, the formal use of space, forager relationships with agriculturalists, the production of the quartz-dominated assemblages by Khoekhoe herders or agriculturalists, or some combination of these. The data reported here suggest that an important aspect of the total forager record is missed when the entire landscape is not considered. Adopting a landscape approach that involves excavating a series of sites in a variety of contexts—rock shelters, open-air sites, rain-control sites and farmer-appropriated rock shelters—is proposed as a way forward.KEY WORDS: Later Stone Age, Mapungubwe landscape, raw materials, interactions.

Southern African Later Stone Age (LSA) archaeology has generally focused on rock shelter excavations (Arthur 2008). It is here where information regarding the Holocene occupation of southern Africa can easily be found, including a range of lithic and non-lithic artefacts, both floral and faunal assemblages, datable material and rock art. In addition, rock shelters are relatively simple to identify as they are restricted to specific and often distinct geological horizons. Further, because of their limited space, it is easy to define sites. However, on the Greater Mapungubwe Landscape, an area including northern South Africa, north-eastern Botswana and south-western Zimbabwe (Fig. 1), the vast majority of identified LSA surface scatters are located in an open-air context.

Studying archaeological surface assemblages has been heavily criticised. It has, for example, been suggested that they lack a complete artefactual record due to post-depositional processes (Sampson 1986; Bintliff & Snodgrass 1988; but see Reynolds 1982). The resulting impression is that surface scatters are often out of context (Lewarch & O’Brien 1981). However, open-air sites are a frequent feature of most archaeological landscapes (Orton 2007; Vanmontfort 2008) and often represent expressions of the LSA not found in rock shelters (see Orton 2007; Sadr & Gribble 2010). It is suggested here that evidence collected at open-air surface scatters should be used to establish a forager cultural record for an area. For this reason the entire landscape needs to be taken into account; hunter-gatherers operated over the whole region and studies should include a variety of their site types (see Vita-Finzi & Higgs 1970; Binford 1983), such as aggregation and dispersal camps (e.g. Wadley 1989) or trading workshops (see Hall & Smith 2000).

66 SOUTHERN AFRICAN HUMANITIES 25: 65–85, 2013

The distribution of sites within landscapes has been studied across Africa to investigate seasonal mobility (e.g. Parkington 1972; Sampson 1985; Sealy 2006), the introduction of Khoekhoe herders (e.g. Ehret 1982), the spread of their artwork (Smith & Ouzman 2004) and the arrival of agriculturalists (e.g. Huffman 1970; Whitelaw & Moon 1996). This study contributes to our understanding of the history of human occupation of the Greater Mapungubwe Landscape (e.g. Hall & Smith 2000; van Doornum 2005) by incorporating findings from surface assemblages in both open-air and rock-shelter contexts, and relating them to the excavated sequences. In so doing, an improved understanding of the regional archaeological record can be gained.

BACKGROUND

The Greater Mapungubwe Landscape is a savannah biome dominated mostly by Mopane (Colophospermum mopane) veld and Vachellia (Acacia) shrubland in floodplains and disturbed areas (Alexander 1984). Riparian forest exists along the major river networks, which includes the Limpopo, Pitsani, Kolope, Motloutse and Shashe Rivers, where large stands of Croton sp. can be found along with Apple-leaf (Lonchocarpus capassa), Nyala tree (Xanthecercis zambesiaca) and Vachellia. Along the Limpopo River, shear deformation (McCarthy & Rubidge 2005: 37) has exposed the underlying Clarens Sandstone Complex, forming koppies and sandstone ridges (Le Baron et al. 2010). The area to the north is mostly characterised by undulating basalt ridges (Alexander 1984) whereas to the south there is the Kolope River drainage basin and a sandveld plateau (Le Baron et al. 2010). The area is characteristically hot and dry, with average temperatures of 32˚C in summer and 22˚C in winter (Hanisch 1981), and annual rainfall between 320 and 350 mm (Huffman 2008).

Fig. 1. The Greater Mapungubwe Landscape and sites mentioned in the text: B, Baobab; B2, Balerno Shelter 2; B3, Balerno Shelter 3; BMS, Balerno Main Shelter; DS, Den Staat AB32; K2, Bambadyanalo; LK, Leokwe Hill; LMS, Little Muck Shelter; MPG, Mapungubwe; SC, Schroda and TS, Tshisiku Shelter.

FORSSMAN: MISSING PIECES 67

The area is well known for its Iron Age archaeology, of which the most prominent site is Mapungubwe, capital of southern Africa’s first state-level society (Huffman 2000), but over 1150 farmer homesteads have also been identified, dating from about AD 900 until the 1800s (Huffman 2012). The area’s rock art has been well recorded and over 170 sites have been identified (e.g. Eastwood & Cnoops 1998). Preliminary studies have been conducted of the Early and Middle Stone Ages (see Kuman et al. 2005 for a review) and the LSA sequence has been well established by van Doornum (2005, 2007, 2008) following the earlier work of Hall and Smith (2000).

Van Doornum (2005: 188–9) identified five cultural phases in the local forager sequence (Table 1). These were established through excavations conducted at Balerno Main Shelter, Tshisiku Shelter, Balerno Shelter 2 and 3 (van Doornum 2000, 2005, 2007, 2008), and Hall and Smith’s (2000) excavation at Little Muck Shelter. The earliest evidence for an LSA occupation of the area is around 11 120 BC at Balerno Main (van Doornum 2008). The site was occupied until about 6060 BC, after which there seems to have been a hiatus in occupation until the fifth century BC (van Doornum 2008). Tshisiku, about 6 km northwest of Balerno Main, was occupied from about 5660 BC, with an artefact peak between 4330 and 1220 BC (van Doornum 2007). From about AD 100 the foraging record begins to change: there is a proliferation of scrapers, items such as ostrich eggshell beads and linkshafts increase, and ceramics appear in forager assemblages (Hall & Smith 2000). Van Doornum (2005: 190) refers to this phase as the early contact period and it is the first sign of forager interactions with Bantu-speaking farmers and possibly Khoekhoe herders on the Greater Mapungubwe Landscape. Evidence indicating a herder presence is, however, questionable as no evidence of their settlement has been identified, although their artwork is present in the region (Eastwood & Smith 2005). On the other hand there is ample evidence of a farmer occupation beginning with the appearance of ceramics, around AD 350, at rain-control

TABLE 1Calibrated dates of the forager cultural phases on the Mapungubwe landscape.

Phase Period Assemblage characteristics

Early pre-contact 6000–1220 BC

CCS dominates; there is a high variety of tools including scrapers (dominate), segments, backed bladelets and adzes at Balerno Main Shelter and Tshisiku Shelter.

Late pre-contact 1220 BC–AD 100CCS dominates most sites but quartz marginally dominates at Little Muck Shelter; scrapers dominate and backed microlith numbers are low.

Early contact AD 100–900CCS dominates; steep increase in scrapers and drop-off in all other tool categories; increase in bead production.

Zhizo contact AD 900–1010 CCS dominates; there is a decrease in tools at most sites but scrapers still dominate.

Leopard Kopje contact AD 1010–1300

CCS dominates; LSA assemblages become more ephemeral during this period but at Balerno Main Shelter they remain constant; scrapers dominate.


sites, indicating that even though no homesteads have been identified (Hall & Smith 2000), farmers were settled in the immediate vicinity (Huffman 2000). From AD 900 Zhizo agriculturalists occupied the region in large numbers (van Doornum 2005: 192). In van Doornum’s sequence this phase is characterised by the further dominance of scrapers in the lithic record, and an increase in trade items as well as in forager sites. The following phase, the Leopard Kopje period, includes the occupation of the major farming centre of K2 at ca. AD 1010 and Mapungubwe from ca. AD 1220 until AD 1300 (Huffman 2000). During this 300-year period, evidence of the forager presence on the landscape diminished and eventually disappeared altogether (Hall & Smith 2000; van Doornum 2005: 196). However, the lack of LSA material does not necessarily mean that foraging people abandoned the area completely.

METHOD

A survey area of approximately 160 km² within the Greater Mapungubwe Landscape was selected with the objective of recording as many LSA sites as possible (Fig. 2). The survey area includes the Venetia Limpopo Nature Reserve (VLNR), the farms Haakdoring and Little Muck, and part of the Mapungubwe National Park. This area was chosen because it had already been extensively surveyed for Iron Age sites by Tom Huffman and his students, and because access was permitted to properties owned by De Beers Consolidated Mines (VLNR and Little Muck) and the Venter family (Haakdoring). The VLNR also covers all three major ecological zones in the area: the sandstone belt (where Little Muck and the surveyed portion of the Mapungubwe National Park are found), the sandveld plateau (where Haakdoring is found) and the

Fig. 2. The survey zone and prominent sites.


Kolope River drainage basin. In a preliminary survey conducted on Haakdoring it was found that a large number of sites and discrete lithic assemblages occurred at the koppies (sandstone inselbergs), whereas neither was found in the sandveld area. It was therefore decided to survey all of the koppies in the survey zone and only part of the sandveld area. Sites and lithic scatters visible on the surface were recorded and named numerically in the order that they were found. At each point the GPS co-ordinates, soil horizon, location, habitat type and cultural materials were recorded and no artefacts were removed.

In total, 153 LSA lithic scatters were identified (Fig. 3) and were grouped into three classes: class A (n = 49) consists of large LSA assemblages with ceramics or organic (shell or bone), glass or metal beads; class B (n = 70) consists of LSA lithic assemblages with no other cultural materials; and class C (n = 34) consists of indeterminate lithic scatters that could not be unequivocally linked to the LSA.

Twenty class A and five class B assemblages were analysed; ten class A assemblages were intentionally selected and the remaining class A and B assemblages were randomly selected. At each lithic scatter a grid of 1 x 1 m squares was set up over the entire assemblage and 20–40 % of the squares within the grid were selected for analysis using a random stratified method, which entails creating zones in the grid (such as columns or blocks) and randomly selecting a representative or equal number of squares for analysis in each. The lithic analyses used van Doornum’s (2005) typology, which is based on Walker’s (1994, 1995a, b) and Deacon’s (1984) typologies. This was done for comparative purposes so as to compare the surface assemblages with the previously excavated regional LSA sequence. Using the lithic inventory for each assemblage

Fig. 3. Assemblages analysed within the survey zone.


(Appendix A), links based on the similarity of raw material frequency, formal tools and formal tool raw material were made with the excavated sequence, and assemblages were placed into the early pre-contact, late pre-contact, early contact, Zhizo and Leopard Kopje periods identified and named by van Doornum (2005; Table 1). This method is known as typological cross-referencing (Gelfand 1971). The surface finds at the dated sites (see van Doornum 2000, 2005) were also compared to the sequence in the same way as the surface assemblages. Analysis of assemblage 144 showed that it was a Middle Stone Age (MSA) site and it is not considered further here.

FINDINGS

For my purposes, an assemblage is considered dominated by a particular raw material type, such as cryptocrystalline silicates (CCS) or quartz, if the total percentage of that material is 5 % or more than any other material type in the assemblage. Using this method I found that the 25 lithic assemblages are dominated either by CCS (n=14; 53.9 %) or quartz materials (n=9; 34.6 %; Table 2), with three (11.54 %) assemblages having no clearly dominant material type. In the former category, most of the assemblages contain over 50 % CCS, with only three containing less: assemblages 5 (48.47 %), Balerno 3 (49.07 %) and Tshisiku (37.71 %). The difference between the average frequency of CCS and quartz in CCS-dominated assemblages is 31.13 %. The only exception is Tshisiku, where the surface assemblage contains a high frequency of dolerite (33.93 %), almost as high as CCS (35.71 %), but it is the only example where this occurs on the surface or in the excavated record. All of the quartz assemblages contain more than 50 % quartz, with some assemblages having particularly high percentages of quartz, such as 17 (78.43 %) and 113 (74.46 %). The difference between the average frequency of quartz and CCS is 34.79 %. In all of these assemblages, barring Tshisiku, quartzite, agate and dolerite frequencies remain low and no discernible pattern was noted.

Using typological cross-referencing, assemblages can be placed into date brackets, a technique used in this study. It is possible to do so in the Greater Mapungubwe Landscape because of the well-dated LSA sequence (see Hall & Smith 2000; van Doornum 2005). Typological cross-referencing relies on time-related diagnostic criteria—chronological markers—and for the region they were identified as raw material utilisation, formal tool types and formal tool frequency (see Forssman 2010). Using these markers the two assemblage types can be compared to the excavated sequence. CCS assemblages are similar to the excavated assemblages as both are dominated by CCS materials and have a high frequency of formal tools (2.11 %), dominated by scraping tools. Quartz assemblages, however, are dominated by a different material type when compared to the excavated sequence and have fewer formal tools (0.7 %). In both CCS and quartz assemblages the morphology of the formal tools and the ratio between scrapers and backed tools are similar. Unfortunately, in most cases assemblages contained chronological markers of multiple periods and could not be placed into a single phase. In some cases, a single component was identified and assemblages could be placed into a date bracket (Table 3).

Decorated ceramics can also be used as chronological indicators, if they can be associated with the lithics (see Sadr 2009), because the various ceramic facies on the Greater Mapungubwe Landscape have been well dated (see Huffman 2007). Lithic


assemblages with ceramics are found at 32, 95, 108, 113, 119 and 152. Some of these assemblages had ceramics (95 and 119) and glass beads (95, 110 and 119), for which a typology and dated sequence exists (Wood 2000), and dated to the Late Iron Age, from AD 1300 to the 1820s (Huffman 2007: xi). These sites are not considered here because glass beads are extremely susceptible to post-depositional processes and the known LSA record does not date to this period. Unfortunately, at the remaining assemblages the ceramics found in most cases derive from various chronologically distinct facies. For example, at site 152, Bambata ceramics and the lithic assemblage, when compared to van Doornum’s (2005) sequence, suggest that the site was occupied in the early contact period sometime between AD 100 and 900. However, the lithic assemblage also suggests a late pre-contact date, prior to AD 100, and a Leopard Kopje-period occupation, post-AD 1000. Other ceramics found with the lithic assemblage at 152

TABLE 2Shelter and open air assemblages.

CCS assemblagesSites (14) CCS Agate Quartz Dolerite Quartzite

1 51.11 – 40 4.44 4.445 48.47 1.05 38.58 1.22 10.6714 84.85 – 7.79 2.16 5.1932 59.8 – 30.6 7.31 2.2850 55.56 – 40 2.22 2.2256 54.69 1.95 42.19 0.78 0.3976 55.26 – 42.11 – 2.63151 56.86 2.35 30.2 3.14 7.45152 54.61 2.13 27.66 4.02 11.58

Bal 2 53.63 6.85 20.97 9.27 9.27Bal 3 49.07 9.94 26.4 5.59 9.01BMS 51.02 15.31 19.39 11.22 3.06LM 100 – – – –TS 35.71 16.07 8.93 33.93 5.36

Total 57.9 3.98 26.77 6.09 5.26

Quartz assemblagesSites (9) CCS Agate Quartz Dolerite Quartzite

17 13.99 0.58 78.43 3.5 3.545 43.75 – 56.25 – –55 36.52 0.43 56.09 0.43 6.5265 32.71 1.12 59.29 0.56 6.3266 31.52 0.94 62.67 0.73 4.14110 24.94 9.67 62.6 0.25 2.54113 14.29 3.03 74.46 2.16 6.06119 27.37 6.39 54.48 0.77 11126 26.82 1.75 60.78 1 9.65

Total 27.99 2.66 62.78 1.05 5.53


TABL

E 3

Chr

onol

ogic

al in

dica

tors

: X in

dica

tes w

here

an

asse

mbl

age’s

crit

eria

(raw

mat

eria

l, fo

rmal

tool

s and

form

al to

ol’s

raw

mat

eria

l [Fo

rmal

RM

]) ar

e sim

ilar t

o th

ose

of a

dat

ed a

ssem

blag

e. N

ote

that

mos

t ass

embl

ages

can

not b

e pl

aced

into

a si

ngle

per

iod

(per

iods

: Leo

pard

Kop

je c

onta

ct, L

KC

; Zhi

zo c

onta

ct, Z

C;

Ear

ly c

onta

ct, E

C; L

ate

pre-

cont

act,

LPC

and

Ear

ly p

re-c

onta

ct, E

PC).

For t

he fu

ll de

tails

of

this

tabl

e se

e Fo

rssm

an 2

010.

Assemblage

Raw

Mat

eria

lFo

rmal

Too

lsFo

rmal

RM

Tota

lPe

riod

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

1

X

X

2

E

C

5

X

XX

X

22

E

C/L

PC

14

X

1

LK

C

17

X

X

1

1

LK

C/E

PC

32

XX

X

X

X

11

1

1LK

C/Z

C/E

C/E

PC

45

X

X

1

1

LK

C/E

C

50

X

X

1

1

LK

C/E

C

55X

X

X

2

1LK

C/E

PC

56

XX

XX

X

11

11

1A

ll

61

X

X

2

E

C

65

X

X

2

E

C

66

X

X

X

X

2

2Z

C/E

PC

71

X

X

2

E

C


TABL

E 3

(con

tinue

d)C

hron

olog

ical

indi

cato

rs: X

indi

cate

s whe

re a

n as

sem

blag

e’s c

riter

ia (r

aw m

ater

ial,

form

al to

ols a

nd fo

rmal

tool

’s ra

w m

ater

ial [

Form

al R

M])

are

simila

r to

thos

e of

a d

ated

ass

embl

age.

Not

e th

at m

ost a

ssem

blag

es c

anno

t be

plac

ed in

to a

sing

le p

erio

d (p

erio

ds: L

eopa

rd K

opje

con

tact

, LK

C; Z

hizo

con

tact

, ZC

; E

arly

con

tact

, EC

; Lat

e pr

e-co

ntac

t, LP

C a

nd E

arly

pre

-con

tact

, EPC

). Fo

r the

full

deta

ils o

f th

is ta

ble

see

Fors

sman

201

0.

Assemblage

Raw

Mat

eria

lFo

rmal

Too

lsFo

rmal

RM

Tota

lPe

riod

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

LKC

ZC

EC

LPC

EPC

76

X

1E

PC

108

X

1

LKC

110

X

1

EPC

113

X

1

EC

119

X

X

2

EC

126

X

X

1

1

LKC

/EC

151

X

X

X

2

EC

152

X

X

X

X

1

2

1

LKC

/EC

/LPC

B2X

X

2

LK

C

B3X

X

X

X

2

2

LK

C/E

C

BMS

X

X

2

LKC

LM

X

1

LK

C

TS

X

X

X

11

1

LK

C/Z

C/E

C


are Zhizo (AD 900 to 1000) or Leokwe sherds (AD 1000 to 1300). It can be assumed from these data that the site was likely occupied, or at least used, at varies stages from the late pre-contact through until the Leopard Kopje period.

In some areas CCS and quartz assemblages are found in close proximity to one another. For example, in the eastern portion of the survey area CCS assemblages 50 and 56 and quartz assemblages 45, 65 and 66 are found at the same koppie and within 100 m of one another. In the central region, Little Muck (LMS in Fig. 3) contains a CCS assemblage (Hall & Smith 2000). Little Muck is 60 m from 113 and 800 m from 110, which are two quartz-dominated assemblages. Therefore, in some cases the distance between CCS and quartz assemblages is negligible. In other areas, mostly on the sandveld plateau in the northeastern portion of the survey area, quartz assemblages do not exist, such as around sites 1, 5, 14, 32, 151 and 152, all dominated by CCS materials (see Fig. 3).

The proximity between both CCS and quartz assemblages and nearby agriculturalist homesteads may be significant. Before the Kolope River joins the Limpopo River it passes through the sandstone belt in an area where a large number of Middle Iron Age (AD 900–1300; Huffman 2007: xi) homestead sites occur. Three CCS (21.43 %) and five quartz assemblages (55.56 %) occur in this zone, in addition to a large number of quartz-dominated assemblages recorded during the fieldwork but not analysed further (Figs 4–5).

DISCUSSION

Typological cross-referencingIn order to give the surface assemblages an approximate age they were typologically cross-referenced with the dated sequence. This proved to be useful as assemblages which would otherwise be undated could now be placed into a date bracket. In order to do so, ratios of the different types of formal tools, ratios of the raw materials used to produce the entire assemblage, and the ratio of raw materials used to produce only the formal-tool component of the entire assemblage were used as chronological markers. The results show that there are two categories of assemblages: those that can be placed into a single phase and those that fit into multiple phases. Fourteen assemblages (53.85 %) can be placed into a single phase, of which seven are from the early contact period (50 %), five from the Leopard Kopje period (35.71 %) and two from the early pre-contact period (14.29 %). The remaining 12 assemblages (46.15 %) represent multiple-occupation sites, suggesting that foragers living on the landscape reused certain camps. In order to confirm these findings radiocarbon dates are needed, possibly using shell artefacts such as OES beads (see Sadr 2009). Until then these results should be viewed with caution.

Quartz-dominated assemblagesPrevious research on the Greater Mapungubwe Landscape has identified CCS-dominated LSA assemblages at each of the five excavated sites (see Hall & Smith 2000; van Doornum 2005). Quartz-dominated assemblages, on the other hand, are poorly understood because they occur in a range of contexts with no strong chronological associations. Hall and Smith (2000) identify a basal quartz assemblage at Little Muck that closely resembles the surface assemblages identified in this study, in which no single dominant raw-material type exists. While no radiocarbon dates are associated with Little


Fig. 4. CCS and quartz assemblage distribution.

Fig. 5. CCS and quartz assemblages related to Middle Iron Age homesteads.


Muck’s basal assemblage, the lack of ceramics suggests a pre-contact date, although this should be accepted cautiously since a lack of ceramics does not necessarily mean that the site was occupied before the arrival of ceramic-producing agriculturalists or pastoralists. Quartz assemblages are also identified by Schoeman (2006) in the farmer rain-control sites of M3S and EH Hill. She argues that foragers played a central role in the rain-control rituals undertaken at K2 rain-control sites. The only other previously recorded quartz assemblage comes from an Iron Age site known as Den Staat AB32, occupied continually between AD 900 and AD 1300, and again in more recent centuries (Forssman 2010). Unfortunately, the highly disturbed deposit there does not allow for an association to be made between the lithic assemblage and any of the site’s agriculturalist phases (discussed in more detail below). These three quartz assemblage types all come from different contexts—a rock shelter, a rain-control site and a possible agriculturalist homestead—but they nevertheless have certain features in common, including a low frequency of formal tools.

Quartz assemblages on the surface are at present anomalous and cannot easily be explained. Unlike those found at rain-control sites, they contain a diverse assemblage of formal tools, albeit in low numbers, and are more like the assemblage found at Den Staat AB32. There are a number of possibilities that might explain their appearance in the archaeological record of the area, whereas the CCS assemblages seem to be comparable and related to the rock-shelter sequence. The possible reasons that quartz assemblages exist are listed and discussed below.

Proximity to raw-material outcrops and material selectionThe most natural explanation would be one involving access to raw materials, with foragers exerting the least effort by utilising nearby raw materials rather than travelling great distances to acquire them. Sadr and Gribble (2010) argue that a portion of the sites they analysed in the Vredenburg Peninsula, Western Cape, are dominated by quartz because of their proximity to quartz material sources. Assemblages that lacked quartz were in areas where quartz material was not readily available. In northeastern Botswana, I report on Shawu Camp, a forager camp/manufacturing site on a small koppie surrounded by a large CCS outcrop mostly composed of agate material. Unsurprisingly, over 90 % of the lithic assemblage found there had been produced using that same CCS material (Forssman 2013). Unfortunately, the location of raw-material outcrops was not recorded during the fieldwork undertaken for the project presented here and so it remains impossible to link the distribution of CCS- and quartz-dominated assemblages with that of raw material sources over the landscape. However, CCS and quartz assemblages were found in close proximity to one another, sometimes, as in parts of the eastern and central area of the survey zone, as little as 100 m apart. Therefore, the distance between the different assemblage types and raw material outcrops is negligible as the two assemblages are, in certain parts of the survey zone, in close proximity to one another. Foragers are also known to have carried nodules or partially worked stones from a material source to their camp, at times covering large distances (e.g. Barham 1992). As a result, a CCS-dominated assemblage may in fact be located a great distance from any CCS material sources and possibly in an area where quartz raw material is more readily available. With this in mind, it does not seem likely that access to raw material determines the dominant material type in each assemblage.


Different site types, settlement and mobility patternsSome features of the late-Holocene record cannot be identified at rock-shelter excavations. First, foragers are highly mobile and use a variety of site types, such as manufacturing sites (Forssman 2013), aggregation and dispersal camps (Wadley 1989; Barham 1992), and trading posts (Hall & Smith 2000), all for variable lengths of time (e.g. Yellen 1976). It should be expected, then, that on the Greater Mapungubwe Landscape a variety of site types exist possessing different expressions of the LSA record. For example, van Doornum (2005) strongly argues using Wadley’s (1989) findings that Balerno Main is an aggregation site and Balerno 2 and 3 are both dispersal camps. Hall and Smith (2000) show Little Muck to be a workshop and trade depot, and Tuli Lodge may have been similar in this respect (Walker 1994). At Den Staat AB32, foragers may have been living within an agriculturalist homestead (Forssman 2010), while at JC, M3S and EH Hill the assemblages are part of the farmer rain-control ritual (Schoeman 2006). With such a diversity of site types it may be that quartz assemblages form another component of the regional forager-settlement pattern. They may be related to CCS assemblages in that they are part of the larger forager system (for example see Binford 1983), but represent different occupation phases, such as aggregation, dispersal (e.g. Wadley 1989) or manufacturing camps, as argued for elsewhere. Site diversity has also been recorded within single site types. For example, Yellen (1976) comments on the varied nature of sites within the dispersal phase, particularly with respect to gift manufacture and specialist activities. While such activities are generally not undertaken during this phase, at times gifts, such as ostrich eggshell beads, are produced in preparation for the aggregation phase. How this manifests in the archaeological record, considering the striking difference between open-air and rock-shelter sites as shown here and elsewhere (see Barham 1992; Plug et al. 2010), is a point of discussion.

Forager versus herder signaturesOver the past 2000 years, two stone-tool producing communities have occupied southern Africa: foragers and herders (Eastwood & Smith 2005). In the Northern Cape at Doornfontein (Humphreys & Thackeray 1983; Sampson 1986; Parsons 2003) it has been suggested that local quartz-dominated assemblages were produced by herding communities (see Parsons 2003, 2007), and in the Western Cape at Kasteelberg (Smith et al. 1991) similar assemblages have been found. In both regions the quartz assemblages lack well-crafted formal tools. This may be the case on the Greater Mapungubwe Landscape where foragers produced CCS-dominated assemblages like those found by Hall and Smith (2000) and van Doornum (2005), and herders possibly produced quartz-dominated assemblages. At present the only evidence indicating a herder occupation of the area is the existence of finger-painted geometric artwork (Eastwood & Smith 2005), which, it has been argued, was produced by Khoekhoe herders (Smith & Ouzman 2004). The authorship of the finger-painted geometric tradition, however, is in question (Mitchell’s and Morris’s comments in Smith & Ouzman 2004), along with the chronology of the artwork and whether it fits the timing of the initial herder migration into southern Africa (Sadr 2008). Also present on the Greater Mapungubwe Landscape are forager paintings of fat-tailed sheep (Walker 1991), a species that may have been introduced into southern Africa by arriving herding communities. If so, this indicates encounters between foragers and Khoekhoe people (Eastwood & Eastwood


2006). Therefore, assuming that the herder migration did in fact also occur ca. 2000 BP (see Sadr 2008) on the Greater Mapungubwe Landscape (Eastwood & Smith 2005), it may be the case that the CCS- and quartz-dominated assemblages indicate not so much a difference in the way foragers used the landscape, but rather the result of different people occupying the same space.

The lack of data does not permit clear associations to be made between Khoekhoe herders and quartz-dominated assemblages. At a number of sites locally and in other parts of southern Africa, quartz assemblages are strongly associated with foraging people. At Leholamogoa Shelter in the Makgabeng Plateau, about 150 km southwest of the Greater Mapungubwe Landscape, Bradfield et al. (2009) found a lithic assemblage comprising 80–90 % quartz material. The quartz assemblage pre-dates 2000 BP and shows relative continuity in terms of formal tool production, until the nineteenth century, even though there is evidence of interaction with farming people for much of this period. While at Little Muck interaction with farmers heavily influenced the forager cultural signature, an ephemeral basal quartz-dominated assemblage (n = 687) is thought to have been produced by foragers, and the lack of associated ceramics suggests that it pre-dates the arrival of herding or farming people (Hall & Smith 2000). In the vicinity of Little Muck are M3S (2.5 km southeast) and EH Hill, which are rain-control sites that contain quartz-dominated assemblages believed to be produced by foragers involved in the ritual (Schoeman 2006). With these examples in mind, it is clear that quartz alone cannot be used to indicate a herder occupation. Likewise, on the Greater Mapungubwe Landscape the formal tool frequencies and the ratio between scrapers and backed tools are comparable in both CCS- and quartz-dominated assemblages, indicating a continuity or affinity between the two assemblage types.

A shared landscapeThe outcome of forager interactions with agriculturalists varies across southern Africa (Kent 1992). On the Greater Mapungubwe Landscape at sites such as Little Muck, interaction led to changes in the forager stone-tool assemblage and the introduction of ceramics and iron (Hall & Smith 2000). It has also been found here (van Doornum 2008) and elsewhere such as in Thamaga, Botswana (Sadr 2002) that the distance between forager camps and farmer homesteads tempers the degree of change and exchange between these two ethnic groups; moving towards farmers affords foragers various amenities that a settled way of life offers (Wadley 1996; Lee 2002), including food security, access to livestock, iron tools and trade wealth. As a result, foragers might settle in homesteads and begin to take on the appearance of farmers in the archaeological record, and unless they maintain parts of their own economy (see Hall 2000), they may become archaeologically invisible (see Sadr & Plug 2001; Sadr 2002). Such findings of LSA assemblages associated with agriculturalist homesteads have been made, for example, in the Waterberg (Van der Ryst 2003), Madikwe (Hall 2000), the Magaliesberg (Wadley 1996), at Msuluzi Confluence in KwaZulu-Natal (Maggs 1980), and at Magagarape near Gaborone (Walker 1994). Perhaps the best example is Hall’s (2000) identification of two stone tool caches in the residential zone of a Madikwe phase site (AD 1500–1700). Here Lithic Scatter 1 (LS1) was probably situated behind a hut in a back-courtyard commonly associated with women and children and which, in this case, included a small copper furnace. The Madikwe evidence suggests foragers


may have been incorporated into the farming economy, whether through marriage or assimilation (cf. Hall 2000: 47–8).

It has been argued that on the Greater Mapungubwe Landscape a similar pattern occurred, in which foragers moved between farmer-settled areas and those relatively unoccupied by farmers. Little Muck was specifically occupied for its proximity to the Leokwe Hill Iron Age centre (Hall & Smith 2000), roughly 1.5 km away, for trade and exchange purposes. On the other hand, Balerno Main, located more than 3 km from any farmer settlement, was utilised as an aggregation site (van Doornum 2008). Balerno Main shows comparably less evidence of exchange with farmers than Little Muck, possibly because it was used during a formal phase of the foraging cycle or because of its distance from farmers (van Doornum 2008). Van Doornum (2005: 195) suggests that foragers may have been moving between Balerno Main and dispersal camps, possibly also Little Muck, or even relied on farmers during certain phases of the forager cycle. Thus, it may be that quartz-dominated assemblages represent a phase in the foraging cycle during which foragers moved towards farmer homesteads, beginning in the early second millennium AD, a period during which they had access to farming tools and so only required an informal, expedient tool technology. Without dates this cannot be shown for certain but the near proximity between Middle Iron Age homesteads and quartz-dominated assemblages may suggest the two are associated. Furthermore, the only evidence suggesting that quartz assemblages were produced by foragers rather than farmers is the presence of formal tools.

If quartz-dominated assemblages are linked to agriculturalist homesteads, it should be expected that they would for the most part not be found in rock shelters, but instead near farmer settlements. There are a few sites in the region where LSA assemblages were found and excavated within homesteads (Forssman 2013), such as Den Staat AB32 (-22.2272; 29.25492; see Forssman 2010 for details). Den Staat AB32 was occupied from AD 900 into the Khami period and is situated in the Limpopo floodplain. In a trench situated in what would have been the back-courtyard of a hut, an assemblage of 2469 lithics was recovered. Quartz dominates the assemblage (47.63 %) followed by CCS (34.39 %), dolerite (10.49 %), agate (4.54 %) and quartzite (2.96 %). The formal-tool component accounts for 1.05 % (n = 26) of the total assemblage and includes backed bladelets (23.1 %), scrapers and segments (each 15.4 %), and adzes (7.69 %). The remainder of the formal-tool component consisted of miscellaneous retouched pieces, awls and segmented backed bladelets, together comprising the majority of the formal tool component (38.5 %). While quartz is consistently dominant throughout the deposit, formal tools are mostly made from CCS materials (53.8 %), followed by agate (30.8 %) and quartz (15.4 %). The similarities between the Den Staat AB32 assemblage and the quartz-dominated assemblages are clear (Table 4): both are dominated by quartz and they have few formal tools, which are mostly made on CCS materials. Unfortunately, aardvark (Orycteropus afer) and springhare (Pedetes capensis) burrows have disturbed Den Staat AB32 considerably and the association between the lithics and the Iron Age occupation could not be firmly established.

Den Staat AB32 may represent a settlement shift on the part of foragers during the Middle Iron Age, at which time Little Muck was appropriated by K2 farmers (Hall & Smith 2000). Farmers perhaps wished to ‘own’ a place like Little Muck because of its association with powers attributed to foragers as ‘first people’ (see Loubser & Laurens


1994). Farmers would have been aware of the power inherent in rock art, and perhaps believed that foragers had control over the rhythms of nature (e.g. Dowson 1994). It remains to be shown whether Little Muck demonstrates a process of appropriation and exclusion of foragers, or the complete assimilation of foragers into the farming economy. Whatever the case, during this time the LSA content at all excavated sites diminished and then eventually disappeared altogether.

Using rock shelters exclusively to explain the foraging signature may not be the most adequate approach to the late-Holocene record (see Orton 2007). It is, however, necessary when first establishing a cultural sequence, as was done by Hall and Smith (2000) and van Doornum (2005). With their studies as a foundation, research can now be performed at a variety of sites in different contexts both inside and outside of rock shelters. The varied nature of sites and their distribution across the region require that archaeologists adopt a landscape approach and excavate a range of site types in different ecological locations. Only then will we have a full understanding of the local forager sequence.

CONCLUSION

The archaeological record varies and changes across the landscape (Foley 1981) and in order to develop a fully inclusive archaeological sequence, sites distributed across the region and in a variety of contexts need to be studied. It has been shown that two LSA assemblage types are present on the Greater Mapungubwe Landscape: CCS- and quartz-dominated. A number of possible explanations are posited, yet without dates none can be argued with certainty. Assemblage differences may be due to access to raw materials, special-purpose sites, the formal use of space, relationships with local agropastoralists or the difference between forager and Khoekhoe herder-produced lithic assemblages. Each explanation has ramifications for the way the foraging record is viewed, particularly if the use of quartz-dominated assemblages is due to external

TABLE 4Den Staat AB32 and surface assemblages dominated by quartz.

Site Quartz Quartzite CCS Agate Dolerite Total Formal toolsN % N % N % N % N % N %

17 269 78.43 12 3.50 48 13.99 2 0.58 12 3.50 343 4 1.17

113 276 74.46 17 6.06 50 14.29 12 3.03 8 2.16 363 1 0.28

66 863 62.67 57 4.14 432 31.52 13 0.94 10 0.73 1375 4 0.29

110 315 62.60 17 2.54 134 24.94 53 9.67 3 0.25 522 3 0.57

126 485 60.78 77 9.65 214 26.82 14 1.75 8 1.00 798 4 0.50

65 319 59.29 34 6.32 176 32.71 6 1.12 3 0.56 538 4 0.74

45 45 56.25 – – 35 43.75 – – – – 80 1 1.25

55 129 56.09 15 6.52 84 36.52 1 0.43 1 0.43 230 2 0.87

119 231 54.48 47 11.00 126 27.37 35 6.39 6 0.77 445 3 0.67

AB32 1176 47.63 73 2.96 849 34.39 112 4.54 259 10.49 2469 26 1.05


factors. However, the possibility that a combination of these explanations may shed light on the nature of quartz assemblages cannot be excluded, particularly if quartz assemblages, like CCS assemblages, consist of a series of phases. Through future research and typological studies on a range of quartz assemblages, along with dating, such cultural phases may be established. Studies should also focus on extending the survey zone and including the sandveld areas to increase the sample size and better represent site distribution within the landscape. Only then will the relationship between CCS and quartz assemblages be adequately explained.

ACKNOWLEDGEMENTSFunding for this project was provided by the Palaeontological Scientific Trust, National Research Foundation and the University of the Witwatersrand. De Beers Consolidated Mines generously sponsored accommodation. I would like to thank Karim Sadr for his help and guidance over the course of this project, Peter Mitchell and Karim Sadr for their insightful comments on a previous draft of this paper, and Simon Hall and an anonymous reviewer for their comments. All errors are my own.

REFERENCES

Alexander, G.J. 1984. A preliminary investigation into the relationships between geology, soils and vegetation in the Eastern Tuli Block, Botswana. Honours Report, University of Natal: Natal.

Arthur, C. 2008. The archaeology of indigenous herders in the Western Cape of South Africa. Southern African Humanities 20: 205–20.

Barham, L.S. 1992. Let’s walk before we run: an appraisal of historical materialist approaches to the Later Stone Age. South African Archaeological Bulletin 47: 44–51.

Binford, L.R. 1983. In pursuit of the past: decoding the archaeological record. London: Thames & Hudson.Bintliff, J. & Snodgrass, A. 1988. Off-site pottery distributions: a regional and interregional perspective.

Current Anthropology 29: 506–13.Bradfield, J., Holt, S. & Sadr, K. 2009. The last of the LSA on the Makgabeng Plateau, Limpopo Province.

South African Archaeological Bulletin 64: 176–83.Deacon, J. 1984. The Later Stone Age of southernmost Africa. Oxford: British Archaeological Reports.Dowson, T.A. 1994. Reading art, writing history: rock art and social change in southern Africa. World

Archaeology 23 (3): 332–44.Eastwood, E.B. & Cnoops, C.J.H. 1998. The rock art of the Limpopo-Shashi confluence. South Africa

and Zimbabwe: an overview. Pictogram 10 (1):1–17.Eastwood, E.B. & Eastwood, C. 2006. Capturing the spoor: an exploration of southern African rock art. New

Africa Books, Claremont. Eastwood, E.B. & Smith B.W. 2005. Fingerprints of the Khoekhoen: geometric and handprinted rock

art in the Central Limpopo Basin, southern Africa. South African Archaeological Society Goodwin Series 9: 63–76.

Ehret, C. 1982. The first spread of food production to southern Africa. In: C. Ehret & M. Posnansky, eds, The archaeological and linguistic reconstruction of African history. Berkeley: University of California Press, pp. 158–81.

Foley, R. 1981. Off-site archaeology and human adaptation in eastern Africa. Oxford: British Archaeological Reports.Forssman, T.R. 2010. The Later Stone Age occupation and sequence of the Mapungubwe landscape. MSc dissertation,

University of the Witwatersrand.Forssman, T.R. 2013. A preliminary report on fieldwork in the Northern Tuli Game Reserve, northeastern

Botswana. South African Archaeological Bulletin 68: 63–71.Gelfand, A.E. 1971. Seriation methods for archaeological materials. American Antiquity 36: 263–74.Hall, S.L. 2000. Forager lithics and Early Moloko homesteads at Madikwe. Natal Museum Journal of

Humanities 12: 33–50.Hall, S.L. & Smith, B. 2000. Empowering places: rock shelters and ritual control in farmer-forager

interactions in the Northern Province. The South African Archaeological Society Goodwin Series 8: 30–46.

Hanisch, E.O.M. 1981. The northern Transvaal: environment and archaeology. In: E.A. Voigt, ed., Guide to archaeological sites in the Northern and Eastern Transvaal. Pretoria: Transvaal Museum, pp. 1–6.


Huffman, T.N. 1970. The Early Iron Age and the spread of the Bantu. South African Archaeological Bulletin 25: 3–21.

Huffman, T.N. 2000. Mapungubwe and the origins of the Zimbabwe culture. South African Archaeological Society Goodwin Series 8: 14–29.

Huffman, T.N. 2007. Handbook to the Iron Age: the archaeology of pre-colonial farming societies in southern Africa. Scottsville: University of KwaZulu-Natal Press.

Huffman, T.N. 2008. Climate change during the Iron Age in the Shashe-Limpopo Basin, southern Africa. Journal of Archaeological Science 32: 2032–47.

Huffman, T.N. 2012. Historical archaeology of the Mapungubwe area: Boer, Birwa, Sotho-Tswana and Machete. Southern African Humanities 24: 33–59.

Humphreys, A.J.B. & Thackeray, A.I. 1983. Ghaap and Gariep: the Later Stone Age in the northern Cape. Cape Town: South African Archaeological Society Monograph Series.

Kent, S. 1992. The current forager controversy: real versus ideal views of hunter-gatherers. Man 28: 45–70.

Kuman, K., Gibbon, R.J., Kempson, H., Langejans, G., Le Baron, J.C., Pollarolo, L. & Sutton, M. 2005. Stone Age signatures in northernmost South Africa: early archaeology of Mapungubwe National Park and vicinity. In: F. D’Errico & L. Backwell, eds, From tools to symbols, from early hominids to modern humans. Johannesburg: University of Witwatersrand Press, pp. 163–82.

Le Baron, J.C., Kuman, K. & Grab, S. 2010. The landscape distribution of Stone Age artefacts on the Hackthorne Plateau, Limpopo River valley, South Africa. South African Archaeological Journal 65: 123–31.

Lee, R.B. 2002. Solitude or servitude? Ju/’hoansi images of the colonial encounter. In: S. Kent, ed., Ethnicity, hunter-gatherers and the “other”: association or assimilation in Africa. Washington: Smithsonian Institution Press, pp. 184–205.

Lewarch, D.E. & O’Brien, M.J. 1981. The expanding role of surface assemblages in archaeological research. Advances in Archaeological Method and Theory 4: 297–342.

Loubser, J. & Laurens, G. 1994. Depictions of domestic ungulates and shields, hunter-gatherers and agropastoralists in the Caledon River Valley area. In: T.A. Dowson & J.D. Lewis-Williams, eds, Contested images: diversity in southern African rock art research. Johannesburg: University of the Witwatersrand Press, pp. 83–118.

Maggs, T. 1980. Msuluzi confluence: a seventh century Early Iron Age site on the Tugela River. Annals of the Natal Museum 24: 111–45.

McCarthy, T. & Rubidge, B. 2005. The story of Earth and life: a southern African perspective on a 4.6-billion-year journey. Cape Town: Struik.

Orton, J. 2007. The sampling of ephemeral shell scatters in Namaqualand, South Africa. South African Archaeological Bulletin 62: 74–8.

Parkington, J.E. 1972. Seasonal mobility in the Late Stone Age. African Studies 31: 223–43.Parsons, I. 2003. Lithic expressions of Later Stone Age lifeways in the Northern Cape. South African

Archaeological Bulletin 58: 33–7.Parsons, I. 2007. Hunter-gatherers and herders? Reconsidering the Swartkop and Doornfontein Industries

of the Northern Cape Province, South Africa. Before Farming 4: 123–32.Plug, I., Mitchell, P. & Bailey, G. 2010. Late Holocene fishing strategies in southern Africa as seen from

Likoaeng, highland Lesotho. Journal of Archaeological Science 37: 3111–23.Reynolds, P.J. 1982. The ploughzone. Festschrift zum 100 Jahrigen Jubiläum der Abteilung Vorgeschichte der Natur-

historischen Gesellschaft Nurnberg, pp. 315–40.Sadr, K. 2002. Encapsulated Bushmen in the archaeology of Thamaga. In: S. Kent, ed., Ethnicity, hunter-

gatherers, and the ‘Other’: association or assimilation in Africa. Washington: Smithsonian Institution Press, pp. 28–47.

Sadr, K. 2008. Invisible herders? The archaeology of Khoekhoe pastoralists. Southern African Humanities 20: 179–203.

Sadr, K. 2009. Marine shell dates and surface lithic assemblages on the west coast of South Africa. Journal of Archaeological Science 36: 2713–29.

Sadr, K. & Gribble, J. 2010. The stone artefacts from the Vredenburg Peninsula archaeological survey, west coast of South Africa. Southern African Humanities 22: 19–88.

Sadr, K. & Plug, I. 2001. Faunal remains in the transition from hunting to herding in southeastern Botswana. South African Archaeological Bulletin 56: 76–82.

Sampson, C.G. 1985. Atlas of Stone Age settlement in the central and upper Seacow Valley. Bloemfontein: National Museum.


Sampson, C.G. 1986. Models of a prehistoric herder-hunter contact zone: a first approximation. South African Archaeological Society Goodwin Series 5: 50–6.

Schoeman, M.H. 2006. Clouding power? Rain-control, space, landscapes and ideology in Shashe-Limpopo state formation. PhD thesis, University of the Witwatersrand.

Sealy, J. 2006. Diet, mobility, and settlement pattern among Holocene hunter-gatherers in southernmost Africa. Current Anthropology 47 (4): 569–95.

Smith, A.B., Sadr, K., Gribble, J. & Yates, R. 1991. Excavations in the south-western Cape, South Africa, and the archaeological identity of prehistoric hunter-gatherers within the last 2000 years. South African Archaeological Bulletin 46: 71–91.

Smith, B.W. & Ouzman, S. 2004. Taking stock: identifying Khoekhoen herder rock art in southern Africa. Current Anthropology 45: 499–526.

Van der Ryst, M.M. 2003. The so-called Vaalpense or Masele of the Waterberg: the origins and emergence of a subordinate class of mixed descent. Anthropology Southern Africa 26: 42–52.

van Doornum, B.L. 2000. Spaces and places: investigating proximity between forager and farmer sites. MSc dissertation, University of the Witwatersrand.

van Doornum, B.L. 2005. Changing places, spaces and identity in the Shashe-Limpopo region of Limpopo Province, South Africa. PhD thesis, University of the Witwatersrand.

van Doornum, B.L. 2007. Tshisiku Shelter and the Shashe-Limpopo confluence area hunter-gatherer sequence. Southern African Humanities 19: 17–67.

van Doornum, B.L. 2008. Sheltered from change: hunter-gatherer occupation of Balerno Main Shelter, Shashe-Limpopo confluence area, South Africa. Southern African Humanities 20: 249–84.

Vanmontfort, B. 2008. Farmer-forager connections in an ‘unoccupied’ land: first contact on the western edge of LBK territory. Journal of Anthropological Archaeology 27: 149–60.

Vita-Finzi, C. & Higgs, E.S. 1970. Prehistoric economy in the Mount Carmel area of Palestine: site catchment analysis. Proceedings of the Prehistoric Society 36 (1): 1–37.

Wadley, L. 1989. Legacies from the Later Stone Age. South African Archaeological Society Goodwin Series 6: 42–53.Wadley, L. 1996. Changes in the social relations of precolonial hunter-gatherers after agropastoralist contact:

an example from the Magaliesberg, South Africa. Journal of Anthropological Archaeology 15: 205–17.Walker, N.J. 1991. Rock paintings of sheep in Botswana. In: S.-A. Pager, B.K. Swartz & A.R. Willcox,

eds, Rock Art: the way ahead. SARARA Occasional Paper 1, pp. 183–95. Walker, N.J. 1994. The Late Stone Age of Botswana: some recent excavations. Botswana Notes and Records

26: 1–35.Walker, N.J. 1995a. Late Pleistocene and Holocene hunter-gatherers of the Matopos: an archaeological study of change

and continuity in Zimbabwe. Societas Archaeologica Upsaliensis, Uppsala.Walker, N.J. 1995b. The archaeology of the San: the Late Stone Age of Botswana. In: A.J.G.M. Sanders,

ed., Speaking for the Bushmen. Gaborone: Botswana Society.Whitelaw, G. & Moon, M. 1996. The ceramics and distribution of pioneer agriculturists in KwaZulu-

Natal. Natal Museum Journal of Humanities 8: 53–79.Wood, M. 2000. Making connections: relationships between international trade and glass beads from the

Shashe-Limpopo area. South African Archaeological Society Goodwin Series 8: 47–50.Yellen, J.E. 1976. Settlement patterns of the !Kung. In: R.B. Lee & I. De Vore, eds, Kalahari hunter-gatherers:

studies of the !Kung San and their neighbors: Cambridge, Harvard University Press, pp. 47–72.

84 SOUTHERN AFRICAN HUMANITIES 25: 65–85, 2013A

ppen

dix

A.1

Ana

lyse

d su

rfac

e co

llect

ion

brea

kdow

nR

aw m

ater

ial r

atio

sFo

rmal

tool

s

Site

Cryptocrystalline

Agate

Quartz

Dolerite

Quartzite

Combined

Scrapers n

Scraper %

Backed bladelets n

Backed bladelet %

Segments n

Segment %

Adzes n

Adzes %

Other tools n

Other tool %

and adzes

including segments

to backed bladelets

Scraper ratio

151

.11

0.00

40.0

04.

444.

4448

.89

00

00

00

00

00

00

00

548

.47

1.05

38.5

81.

2210

.67

51.5

38

61.5

00

17.

691

7.69

323

.11

10

814

84.8

50.

007.

792.

165.

1915

.15

110

00

00

00

00

00

00

117

13.9

90.

5878

.43

3.50

3.50

86.0

12

66.7

00

00

00

133

.30

00

232

59.8

00.

0030

.60

7.31

2.28

40.1

97

77.8

00

00

00

222

.20

00

745

43.7

50.

0056

.25

0.00

0.00

56.2

50

00

00

00

01

100

10

00

5055

.56

0.00

40.0

02.

222.

2244

.44

00

00

00

00

00

00

00

5536

.52

0.43

56.0

90.

436.

5263

.48

150

00

00

00

150

10

01

5654

.69

1.95

42.1

90.

780.

3945

.31

222

.20

02

22.2

444

.41

11.1

21

01

6143

.18

1.62

42.8

67.

474.

8756

.82

555

.61

11.1

00

00

333

.30

01

565

32.7

11.

1259

.29

0.56

6.32

67.2

91

250

00

00

03

750

00

166

31.5

20.

9462

.67

0.73

4.14

68.4

82

33.3

00

233

.30

02

33.3

01

01

7141

.10

2.97

44.0

75.

935.

9358

.90

00

00

00

00

110

00

00

076

55.2

60.

0042

.11

0.00

2.63

44.7

41

500

00

00

01

500

00

110

839

.29

3.57

42.8

67.

147.

1460

.71

150

150

00

00

00

00

11

110

24.9

49.

6762

.60

0.25

2.54

75.0

61

501

500

00

00

00

01

111

314

.29

3.03

74.4

62.

166.

0685

.71

00

00

00

00

110

00

00

011

927

.37

6.39

54.4

80.

7711

.00

72.6

31

33.3

00

00

00

266

.70

00

112

626

.82

1.75

60.7

81.

009.

6573

.18

00

00

00

125

375

10

00

151

56.8

62.

3530

.20

3.14

7.45

43.1

42

400

00

00

03

600

00

215

254

.61

2.13

27.6

64.

0211

.58

45.3

99

81.8

00

19.

090

01

9.09

01

09

FORSSMAN: MISSING PIECES 85A

ppen

dix

A.2

Ana

lyse

d su

rfac

e co

llect

ion

brea

kdow

n

Form

al to

ol ra

w m

ater

ial %

Art

efac

t den

sity

Site

Cryptocrystalline

Agate

Quartz

Dolerite

Quartzite

Total artefacts

Total per square

Formal tools per square

Minimum total artefacts

Ceramics

Beads

10

00

00

459

015

3

5

92.3

10

00

7.69

1143

87.9

21

4748

1410

00

00

023

138

.50.

1777

0

17

100

00

00

343

31.1

80.

2710

29

32

100

00

00

219

733

219

Leok

we

OE

S &

gla

ss45

100

00

00

8016

0.2

240

500

00

00

4522

.50

90

55

5050

00

023

046

0.4

690

5677

.78

22.2

20

00

256

85.3

33

682.

67

61

88.8

911

.11

00

034

538

.33

111

11.6

7

65

750

250

053

848

.91

0.36

2152

6683

.33

016

.67

00

1377

688.

53

5508

Plai

n ea

rthe

nwar

e

7110

00

00

023

659

0.25

708

7610

00

00

038

7.6

0.4

76

10

850

500

00

5614

0.5

56

11

00

100

00

039

339

32

1572

Ve

nda

octa

gona

l bea

d11

310

00

00

023

146

.20.

269

3Le

opar

d K

opje

(TK

2)

119

33.3

033

.333

.30

391

195.

51.

578

2Le

opar

d K

opje

, Kha

mi &

Ven

da

126

100

00

00

798

399

247

88

15

180

200

00

255

36.4

30.

7151

0

15

281

.82

9.09

00

9.09

423

141

3.67

987

Bam

bata

& Z

hizo

/Leo

kwe

OE

S

Documents

Missing pieces: Later Stone Age surface assemblages on the … · 2020. 3. 30. · Southern African Later Stone Age (LSA) archaeology has generally focused on rock shelter excavations