A Preliminary Report on Fieldwork in the Northern Tuli Game Reserve, Northeastern Botswana

South African Archaeological Bulletin 68 (197): 63–71, 2013 63

Field and Technical Report A PRELIMINARY REPORT ON FIELDWORK IN THE NORTHERN TULI GAME RESERVE, NORTHEASTERN BOTSWANA

TIM FORSSMAN St. Hugh’s College, University of Oxford, OX2 6LE, United Kingdom E-mail: [email protected] (Received July 2012. Revised February 2013)

INTRODUCTION There has been a great deal of archaeological research conducted on the Greater Mapungubwe Landscape, an area which includes northern South Africa, northeastern Botswana and southwestern Zimbabwe (Fig. 1). Most of this research has been conducted on the South African side of the border (Fouché 1937; Hanisch 1980, 1981a; Eastwood & Cnoops 1999; Calabrese 2000a,b; Hall & Smith 2000; Huffman 2000, 2007, 2008, 2009; Wood 2000; Kuman et al. 2005; Van Doornum 2005, 2007, 2008; Eastwood & Eastwood 2006; Schoeman 2006; Kempson 2007; Forssman 2010). Immediately to the north of the Limpopo River, in Botswana and Zimbabwe, little work has been conducted even though a number of sites have been identified (see Huffman 2008). Zimbabwe has seen some studies focusing on the agricultural sequence (e.g. Robinson 1966; Garlake 1967; Manyanga et al. 2000; Manyanga 2006) and excavations have been performed at a small number of FIG. 1. The Greater Mapungubwe Landscape and sites mentioned in the text. LSA sites (e.g. Robinson 1964; Cooke & Simons 1969; Walker & Thorp 1997; Thorp 2010). The least studied part of this landscape is in Botswana where only a few excavations at agriculturalist homesteads (Van Waarden 1979, 1980; Voigt & Plug 1981; Kinahan et al. 1998; Kinahan 2000; Mosothwane 2011) and a single LSA excavation (Walker 1994) have been reported. The Greater Mapungubwe Landscape witnessed significant cultural changes between AD 900 and 1300 leading to the apparent disappearance of the indigenous foraging culture (Van Doornum 2005: 196) and the formation of complex agro-pastoral state societies (Huffman 2000). In this paper recent findings from a large project in the Northern Tuli Game Reserve, Botswana (Fig. 2) are presented in order to demonstrate the rich archaeology of the region and contribute to our understanding of the local prehistory.

BACKGROUND

LOCAL ENVIRONMENT FIG. 2. Location of Northern Tuli. Northern Tuli is a 72 000 ha game reserve most of which is located between the Motloutse and Shashe Rivers in northeastern tree (Xanthecercis zambesiaca) and vachellia. The area is charac- Botswana. Its southern and northeastern boundaries are teristically hot and dry with an average maximum temperature the international borders with South Africa and Zimbabwe, of around 32°C in summer and 22°C during winter (Hanisch respectively. The area is composed primarily of undulating 1981b) and an average rainfall of between 320 and 350 mm per basalt ridges except along the Limpopo and Motloutse Rivers annum (Huffman 2008). where shear deformation along the Limpopo Mobile Belt (Mc- Carthy & Rubridge 2005: 37) has exposed the underlying PREVIOUS ARCHAEOLOGICAL RESEARCH Clarens Sandstone Complex and formed koppies and sand- There have been a number of surveys conducted in Northern stone ridges (Le Baron et al. 2010). Dense stands of Mopane tree Tuli, but most of these findings remain unpublished. The bulk (Colophospermum mopane) dominate the local vegetation with of the surveying was performed by Grant Hall during his vachellia (previously known as acacia) shrubland in flood- employment at the Mashatu Game Reserve. From his work he plains and on disturbed land (Alexander 1984). The northern produced an unpublished report (G. Hall 2003) in which he basalt zone is only broken by the Pitsani, Njaswe and Mojale discusses the Mmamagwa complex, 46 agriculturalist home- Rivers along which fertile floodplains can be found in certain steads that he was able to identify which include Zhizo, K2, areas. Along the river networks are riverine forests comprising Mapungubwe, Khami and Venda period sites (Table 1), and a mostly Croton spp., apple-leaf (Lonchocarpus capassa), nyala glass bead assemblage collected from the Mmamagwa complex 64 South African Archaeological Bulletin 68 (197): 63–71, 2013

TABLE 1. Chronology of major ceramic facies found in the area. as the artefact’s size and colour (Banning et al. 2006). This has led some researchers to question the quality of data produced Ceramic facies Date (AD) Reference by archaeological surveys (Cowgill 1989) and ask what in fact these finds actually represent (Wandsnider & Camilli 1992). Happy Rest 350–600 Hanisch 1981a Zhizo 750–1050 Calabrese 2000a Most researchers, however, do agree that many of the issues Leokwe 1050–1220 Calabrese 2000a surrounding an archaeological survey can be overcome by a Toutswe 1050–1300 Huffman 2007: 151 well-structured survey technique (Sampson 1985) and there K2 1000–1200 Huffman 2007: 279 are a number of techniques that a surveyor can choose from. Transitional K2 1200–1250 Huffman 2007: 279 Mapungubwe 1250–1300 Calabrese 2000b Google Earth has been used to plot stone-walled settlements Icon 1350–1400 Huffman 2010 and to identify their different architectural styles (Sadr & Khami 1400–1800 Huffman 2012 Rodier 2012) as well as aerial photography to identify agricul- Venda (Letaba) 1600 until present Loubser 1991 turalist homesteads (Denbow 1984). Chemical analysis of soils such as phosphate testing (Wells et al. 2000) is also a useful and analysed by Marilee Wood (2005). There have been two technique as is the use of test pits (Nance & Ball 1986) to identify additional surveys conducted by Ed Eastwood along the areas of human activity that are not visible on the surface. How- Motloutse River and by Cynthia Mooketsi (2009) in the western ever, foot surveying is argued as the best technique for any portion of Mashatu near to the Mmamagwa complex. archaeological survey (Foley 1981; Reid & Segobye 2000). There In addition to the surveys there have been a few excava- are a number of ways that a foot survey can be undertaken and tions. The earliest of these was Van Waarden’s (1979, 1980) includes surveying vegetation or geological zones (Sampson excavations at Leeukop as part of a project sponsored by Trent 1985; Lane 1996), specific features of the landscape such as University. The site is about 1.6 km northwest of Mmamagwa koppies (Forssman 2010) or walking in transects (Bintliff & and was occupied by agriculturalists during the 19th century Snodgrass 1988). with a possible early 17th century component and may have been used as a rain-control site. Over 40 pole and daga struc- SURVEY METHOD tures were found along with several middens and a large as- Two survey techniques were used in this project: foot and semblage of ceramics, beads and a number of human burials. vehicle surveys. Not all of Northern Tuli was surveyed due to Commando Kop, which is situated approximately 22 km north- its size and the project’s time constraints. Only Tuli Safari east of Leeukop and along the Pitsani River, was excavated as Lodge (2143 ha) and Uitspan South (794 ha) were surveyed part of the same project (Voigt & Plug 1981). The site has a completely on foot with the northern portion of Tuli surveyed primary Zhizo occupation with 11 associated human burials in a vehicle. The Mashatu Game Reserve (25 018 ha) was also but also has Leopard Kopje ceramics. During the Second Anglo- surveyed but not in its entirety. Mashatu was divided up into Boer War the site was used as a canon station (Huffman 2012) 1 × 1 minute squares using a 1:50 000 topographic map (Fig. 3). because of its elevated position on Pitsani Hill overlooking In each square terrain ruggedness (as indicated by the total Bryce’s store 2 km south along the river where an intense battle length of contour lines), hill ranges, the dominant geology and occurred in 1899 (Hickman 1972). Fortunately, the temporary soil type, the number of koppies, known sites and possible sites war-time use of Pitsani Hill did not damage the site signifi- were recorded. Based on the presence, absence and combina- cantly (Voigt & Plug 1981). tion of these attributes certain squares were selected for foot More recently Nick Walker (1994) excavated Tuli Lodge and surveying. The intention was to cover not only areas where Morongwa Mosothwane (2011) Cut Line Rock, both near the sites were known or expected to occur but also a representative Limpopo River. Mosothwane’s (2011) excavation in 2008 was sample of the different ecological zones. The area around limited to the removal of a naturally mummified human body Mmamagwa was completely surveyed in order to gain a better later found to be a male between the ages of 40 and 60 years. understanding of site distribution around this important The remains were removed from a homestead that contains farmer site and to test whether there is a greater density of Khami ceramics and was likely occupied between AD 1400 forager sites from within the last 2000 years around the and 1800. At Tuli Lodge Nick Walker (1994) excavated two 1 × Mmamagwa settlement (see Reid & Segobye 2000). Figure 4 1 m squares to a depth of 30 cm and recovered 14 379 lithics presents a map of the foot and vehicle surveys. which include 4912 chips (34.16%) and 441 formal tools (4.66%). He noted a decline of backed pieces relative to scrapers in upper levels. The assemblage also included Bambata sherds and what Walker (1994) calls Mapungubwe/Zimbabwe ware as well as large amounts of copper slag.

ARCHAEOLOGICAL SURVEY Surveying for archaeological sites is the primary phase of any archaeological work (Tartaron 2003), and the inspection and identification of sites is an integral step before any excavation (Wandsnider & Camilli 1992). Some scholars have even suggested that surveying is more important than excavating (Alcock & Cherry 2004) but it is not without its limitations (Lewarch & O’Brien 1981; Sullivan et al. 2007), which if not taken into consideration, have the potential to distort results. These limitations can include vegetation cover (Foley 1981), survey procedures (Cherry et al. 1988), substrate type (Shennan 1985), site identification (Fish & Kowalewski 1990; Orton 2007) FIG. 3. The 1 × 1 minute grid placed over Mashatu also includes Tuli and and the distance between the surveyor and the artefact as well Uitspan. South African Archaeological Bulletin 68 (197): 63–71, 2013 65

to archaeologists. For example, of the 186 LSA points of interest 30 (16.13%) are at raw material outcrops, which contain a small number of lithics that are scattered over a large area and mixed with naturally occurring raw material resources. It is not possible to say when these lithics were produced and for how long the site was occupied and conducting a full study may prove diffi- cult because of the lack of deposit, mixing of artefacts with naturally broken nodules and the wide distribution of the low density stone tool scatter. Many of the LSA points of interest are ephemeral or discrete collections of lithics. Some offered no option for further enquiry due to a number of reasons, one of which is erosion; a total of 51 (27.42%) LSA points of interest were subjected to either extensive or considerable erosion. There are two reasons why erosion is so prevalent on the landscape. First, the area is covered with aeolian sand, possibly a mantle of the eastern FIG. 4. Foot and vehicle survey tracklogs. palaeo-Kalahari (Le Baron et al. 2010). When flooding is experi- enced, to which the area is prone (Plug 2000), alluvial action SITE DEFINITION AND RECORDING results in the removal of the deposit and formation of erosional Among other reasons, survey data needs to be recorded so gullies. Second, the region was used for intensified local cattle as to inform future decisions on which sites are to be excavated ranching during the early part of the 20th century (Mazonde (Foley 1981). Important information to record includes site 1987) contributing to localised land degradation. Both of position, the condition of the site and deposit, the approximate the factors have contributed to the eroded landscape. At depth of the deposit, a list of artefact types present such as 32 (17.20%) LSA points of interest, however, no erosion was LSA lithics, ceramics or bead types with an estimate of artefact recorded and the sites appear to be in primary context. Such density and site size. In parts of Europe, North America and ephemeral scatters are often overlooked by archaeologists Africa the concept of a ‘site’ is rejected (Ebert 2001) because (Orton 2007) and archaeological projects should include them defining sites in areas with a widespread artefact distribution is as they may provide us with valuable information on short- not always possible (Krakker et al. 1983). For this reason some term camps. have instead suggested that archaeology needs to be studied One might expect that there would be a concentration of across the landscape (Thomas 1975; Foley 1981). The definition LSA points of interest (including rock art sites) near to major of a site draws on other debates in archaeology including - agriculturalist centres such as has been found elsewhere in special purpose sites (Binford 1983) and aggregation and Botswana (Reid & Segobye 2000). Figure 5 shows the location of dispersal camps (Wadley 1989), the discussion of which is the Leokwe Hill and Mmamagwa agriculturalist centres with beyond the scope of this paper. Modifying the definition of a 3 km radius buffer zone around each site and the known Bevan & Conolly (2004) who refer to any accumulation of LSA points of interest, rock art sites and excavations carried out artefacts as a site, I consider sites to be any archaeological point on the landscape. The density of LSA points around of interest. While this includes what are traditionally accepted Mmamagwa is 1.17/km² (n = 33) and 1.03/km² (n = 29) at as sites such as agriculturalist homesteads it also includes Leokwe Hill. In the area of Nldulamithi Kraal and Shawu discrete artefact accumulations like lithic or ceramic scatters, Camp, two sites excavated in this project, the LSA density is isolated finds of stone tools and rock markings such as mankala 2.76/km² (n =78) in the same sized buffer zone. It is clear that gaming boards or grinding hollows. In this survey, a survey there is a large amount of forager sites around the agriculturalist card was developed with the categories GPS coordinates, site centres but their contemporaneity with the agriculturalist number, substrate, soil depth, erosion, location, habitat type, centres is unknown. natural features, animal and human activity and archaeological features listed. The list of categories has a corresponding list of conditions which are numbered to simplify the recording process. For example, when recording the category soil depth the conditions and corresponding numbers are:1=<5cm; 2 = 5–20 cm 3 = 20–50 cm; 4 = 50–100 cm and 5 = >100 cm. Only the number representing the applicable depth measure- ment is recorded. This technique allows for rapid recording in the field and a concise, systematic and consistent recording system easily stored in a digital format and manipulated there- after.

FINDINGS

SURVEY RESULTS The survey revealed 428 archaeological points of interest. From the findings some interesting points can be made and are summarised below. FIG. 5. LSA point density around Leokwe Hill and Mmamagwa withina3km LSA POINTS OF INTEREST radius: B2, Balerno Shelter 2; B3, Balerno Shelter 3; BMS, Balerno Main Shel- LSA artefacts are abundant in Northern Tuli and they are ter; DS, Dzombo Shelter; JS, João Shelter; KC, Kambaku Camp; LMS, Little found in a range of contexts. However, most are of little interest Muck Shelter; MS, Mafunyane Shelter; and SC, Shawu Camp. 66 South African Archaeological Bulletin 68 (197): 63–71, 2013

A number of LSA lithic scatters were also found within agriculturalist homesteads (n = 32; 17.20%) but without excavations their association cannot be determined. Similar occurrences have been found in other parts of southern Africa where the LSA material was associated with the agriculturalist homestead; for example, in Madikwe in the North West Prov- ince (S. Hall 2000), near Gaborone at Magagarape (Walker 1994), at Msuluzi in KwaZulu-Natal (Maggs 1980) and at Baobab, a Zhizo settlement in northern South Africa (Calabrese 2005: 84). It has been suggested that foragers moved into agriculturalist homesteads possibly through marriage (S. Hall 2000), labour relations or assimilation (Wadley 1996). The same shift occurring on the Greater Mapungubwe Landscape might explain the apparent disappearance of the foraging record in rockshelter sites around AD 1300 (Van Doornum 2005: 196). There are historical accounts of foragers living with farmers or in permanent homesteads that might indicate assimilation had FIG. 6 occurred by the mid to late 1800s (see Elton 1872; Dornan 1917). . The distribution of agriculturalist homesteads in the survey zone. Ovals demarcate homestead clusters In order to test whether foragers assimilated into farmer society, excavations need to be conducted at homesteads where lithics them are a number of kraals. A seasonal stream has eroded occur. through the occupation revealing middens, ceramics, beads and human burials (all recorded while surveying the area and IRON AGE SETTLEMENTS by G. Hall 2003).The main palace is on the eastern koppie and The so-called Iron Age of northern South Africa has been terracing is found at its southeastern base and near to the hill- well-studied (for a review see Huffman 2007). In the 2229A map top in this same area. On top of the hill is an area with grain bin zone, over 1150 agriculturalist sites have been identified foundations and a number of grinding stones and at the back or through surveys carried out by Tom Huffman and his students the western edge of the koppie there are at least 15 cupules in (Huffman 2012). Almost all of these sites are found in South the bedrock. The hill has various terraced areas as well as exten- Africa with 46 located in Northern Tuli. An additional 78 agri- sive walling with loopholes. Another large walled complex is culturalist homesteads, defined by the presence of a single or located approximately 2.15 km south of Mmamagwa on top of several kraals, were identified during the survey for this project. a koppie. Only undiagnostic ceramics were found here during In addition, 13 homesteads of those identified by G. Hall and the survey along with a Bohemian glass bead which was Huffman were revisited either because they were located in the produced in the 19th century AD (Wood 2005). square being surveyed or found during the vehicle survey. A number of homestead clusters were also identified during Therefore, 91 homesteads were visited during the survey and a the survey and in different parts of the research area (Fig. 7). total of 124 homesteads have been identified on Mashatu, Tuli North of the Pont Drift border a cluster of eight sites with Icon and Uitspan. At some of the homesteads diagnostic ceramic and Khami ceramics were identified and one with Transitional sherds or glass beads were found and it was possible to identify K2 (TK2) sherds. At one of the Khami sites walling was found in which period the site was occupied (Table 2). The home- on a natural terrace south of the main kraal, which may demar- steads are situated across the entire survey zone (Fig. 6) and cate a girls’ initiation zone (see Huffman 1984). Further north there does not seem to be a difference in the location of sites near to Ndlulamithi Kraal are three agriculturalist homestead from the Middle (AD 900–1300) and Late Iron Age (AD 1300 clusters, situated on the edge of the Mojale and Pitsani River onwards; Fig. 7) periods. floodplains, and another is located northwest of the Of particular interest are the larger sites in the survey zone Mmamagwa complex along the Motloutse River. The reason such as the Mmamagwa complex. This large political centre has why homestead clusters exist and whether all of the settle- Zhizo, K2, Mapungubwe, Khami and Venda ceramics present ments within the cluster are contemporaneous are unknown at as well as glass beads from the respective periods (see Wood this stage and should be the focus of future research. 2005). However it has not been studied in detail and little can be said about the site other than to give a basic description. Two main ridges occur at the Mmamagwa complex and between

TABLE 2. Agricultural homesteads identified based on ceramics and glass beads.

Period Iron Age settlements

Zhizo 7 Leokwe 2 K2 10 Transitional K2 3 Mapungubwe 6 Icon 1 Khami 30 Venda 2 Unknown 63 Total 124 FIG. 7. Middle and Later Iron Age homesteads identified during the survey. Circles demarcate homestead clusters. South African Archaeological Bulletin 68 (197): 63–71, 2013 67

FIG. 8. Rock art sites in Northern Tuli: Bushman, herder and farmer sites. FIG. 9. A so-called Khoekhoe herder apron motif painted behind Dzombo Shelter. BUSHMAN AND KHOEKHOE ROCK ART Eastwood and colleagues (e.g. Eastwood & Cnoops 1999; The Bushman rock art tradition in Northern Tuli is similar Eastwood & Eastwood 2006) have identified more than to rock art in northern South Africa and Zimbabwe (Walker 150 Bushman rock art sites between the Mapungubwe eastern 2009; see Eastwood & Eastwood 2006 for an overview).There is boundary and the confluence of the Motloutse and Limpopo also so-called Khoekhoen herder art on the Greater Mapun- Rivers. In Northern Tuli, however, only eight sites have been gubwe Landscape (see Eastwood & Smith 2005). Notably, identified with an additional two so-called Khoekhoen herder herder art differs in the way images are produced and in the art sites (Fig. 8) and 25 locations with engraved markings which content; all are finger painted and are outlined geometric include mankala boards, grooves, cupules or hollows. The images (Smith & Ouzman 2004) many of which may represent Tuli Block as a whole, which includes Northern Tuli, lacks the aprons or loin cloths (Eastwood & Smith 2005). There are two density of Bushman rock art sites found in South Africa even known sites in Northern Tuli which have so-called herder though there are many paintable rock surfaces present (Walker images. One site, JB Shelter, is located behind Dzombo Shelter 2009). A cursory investigation of the size of the paintable zone and has paintings similar to what has been identified else- shows that in South Africa, the portion of the sandstone koppie where as aprons (Fig. 9). The only other recording of finger- belt that has been surveyed extends for 40.41 km along the painted art is at Cut Line Rock and although not mentioned Limpopo River and 9.43 km inland, covering approximately in Mosothwane’s (2011) report it was identified during the 381.01 km2. The sandstone koppie belt within Northern Tuli is survey along with a nearby panel of Bushman rock art. A less extensive, covering an area of about 71.76 km2 (24.16 km by further analysis of this is beyond the scope of this paper. 2.97 km). Therefore, in South Africa there is approximately a rock art site every 2.54 km2 (total = 150) and one every 8.97 km2 EXCAVATIONS (total = 8) in Northern Tuli. It is clear that even though Northern Six excavated sites produced archaeological materials. Each Tuli has a smaller area where paintable surfaces occur it has a excavation is briefly described below and Table 3 is a summary lower density of rock art sites. The relative lack of sites may of the findings at each site. indicate cultural differences existing across the Limpopo River, Dzombo Shelter (Fig. 10) is situated about 600 m west of the but even so, it would not exclude the possibility of cross-border Mmamagwa complex. It is a fairly large (h: 3 m; d: 10 m; w: 8 m) trade, the sharing of ideas, migration and conflict (cf. Flynn north-facing rockshelter and was excavated because of its deep 1997). deposit (Fig. 11). Two areas of the site were excavated: inside

TABLE 3. A brief summary of the excavated assemblage at each site: DS, Dzombo Shelter; JS, João Shelter; SC, Shawu Camp; MS, Mafunyane Shelter; KC, Kambaku Camp and NK, Ndlulamithi Kraal.

DS MS JS NK KC SC

Chips 6210 3391 3351 240 66 299 Chunks 814 411 454 44 58 197 Cores 128 82 88 2 8 33 Flakes 738 347 257 1 25 19 Broken flakes 3108 1867 1961 22 121 49 Formal tools 509 207 138 2 19 17 Other 17 44 268 11 48 179 Total 11524 6349 6517 322 345 793

Plain ceramics 199 28 1037 372 559 0 Diagnostic ceramics 13 3 80 24 45 0 Total 212 31 1117 396 604 0

Organic beads 86 50 87 13 0 0 Glass beads 20 3 150 3 1 0 Metal beads 0 1 0 1 0 0 Total 106 54 237 17 1 0 68 South African Archaeological Bulletin 68 (197): 63–71, 2013

FIG. 10. Dzombo Shelter. Black shapes represent boulders. FIG. 12. João Shelter. the rock shelter two1×1msquares were dug (trench 1) and a the substantial amount of iron and copper slag found in the large artefact assemblage was recovered, and three squares deposit. were excavated in the outside living area (trench 2). Although In excavations at João Shelter (Fig. 12) a fairly large assem- no radiocarbon dates are currently available, the sequence of blage of over 6500 lithics (n = 53 lithics/13l bucket) was recov- decorated ceramics and glass beads appears to be consistent ered from seven1×1 msquares. The rockshelter itself, where with the dated sequence at farmer sites: in the upper levels four of the squares were excavated, is relatively shallow (h: Leopard’s Kopje ceramics, Mapungubwe oblates and K2 beads 5 m; d: 4 m; w: 18 m) and is situated along a portion of a large were found with Zhizo ceramics and glass beads below these. ridge (h: 14 m; l: 55 m) less than 300 m southeast of the The data recovered from the site are comparable to Van Door- Mmamagwa complex. Immediately outside the rockshelter is num’s (2005) findings in South Africa: the lithic assemblage is an agriculturalist homestead with K2 decorated ceramics dominated by crypto-crystalline (CCS) materials with a high and K2, Mapungubwe, Indo-Pacific and European glass beads. frequency of formal tools (4.16%) and there are few ceramic At the site there are also at least three grainbin foundations, sherds and glass beads present. three possible human burials, a midden and stone-walling The highest density of artefacts came from Mafunyane inside the rockshelter and along the koppie ridge next to the Shelter (n = 595 lithics/13 l bucket), which elsewhere may be midden (trench 4). There is also a fine-lined painting of a kudu named Tuli Lodge (the distinction was made since it was not at the site and an indeterminate antelope. Most of the lithics certain that the shelters were the same because Walker’s [1994] (n = 5405) came from inside the rockshelter and are dominated report has no GPS coordinates for the site). On the surface a by CCS materials but a lithic assemblage dominated by quartz large assemblage containing lithics, ceramics and a variety of came from two squares placed in the homestead at a grainbin beads was recorded and excavations inside the rockshelter foundation (trench 2) and in the midden (trench 4). Charcoal produced a large array of lithic artefacts including backed samples have been selected for radiocarbon dating to establish bladelets, segments, scrapers and a variety of cores. In total a a chronology for the rockshelter and homestead. sample of over 6000 lithics was collected in three 50 × 50 cm Ndlulamithi Kraal was excavated to pursue the notion that squares. Also found were glass beads, various organic beads, foragers may have been living within farmer homesteads. At ceramics, copper items and a portion of a figurine. Of interest is the site K2 and Leokwe decorated pottery was found, and scattered on the surface was an LSA lithic assemblage which included small and medium scrapers. The lithics were found in the kraal and particularly in the northern portion of the homestead near the koppie edge where trench 3 is located (Fig. 13). The site also has a possible second kraal (l: 12 m) that is slightly smaller than the upper kraal (l: 18 m; see Huffman 2009). The excavation revealed additional K2 and Leokwe decorated sherds and ostrich eggshell, bone, land snail, glass and copper beads. No diagnostic LSA artefacts were found in the excavation. However, at Kambaku Camp, a homestead with K2 and Khami decorated ceramics, a small diagnostic LSA assemblage was recovered from behind the hut zone where a single 1 ×1 m square was excavated, and from within the kraal in which three 1 × 1 m squares were dug. Like Dzombo and João Shelter, FIG. 11. The stratigraphy from trench 1 inside Dzombo Shelter. Kambaku Camp is in the southwestern sandstone belt not South African Archaeological Bulletin 68 (197): 63–71, 2013 69

larly Pete le Roux, Andrei Snyman, Quentin Fraser-Jones, Francois du Toit, Abraham Ramonwana and Piet van Rensberg. Two anonymous reviewers are thanked for their in- put as well as Karim Sadr and Jill Kinahan for their help with editing and Wendy Voorvelt for the maps and section drawing. All errors are my own.

REFERENCES Alcock, S.E. & Cherry,J.F.2004. Side-by-Side Survey: Comparative Regional Studies in the Mediterranean World. Oxford: Oxbow Books. Alexander, G.J. 1984. A preliminary investigation into the relationships between geology, soils and vegetation in the Eastern Tuli Block, Botswana. Unpublished BSc Honours Research Report. Pieter- maritzburg: University of Natal. Banning, E.B., Hawkins, A.L. & Stewart, S.T.2006. Detection functions for archaeological survey. American Antiquity 71(4): 723–742. Bevan, A. & Conolly, J. 2004. GIS, archaeological survey and landscape archaeology on the island of Kythera, Greece. Journal of Field Archaeol- ogy 29(1/2): 123–138. 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(3): 506–513. Calabrese, J.A. 2000a. Interregional interaction in southern Africa: Zhizo and Leopard’s Kopje relations in northern South Africa, southwestern Zimbabwe and eastern Botswana, AD 1000 to 1200. African Archaeological Review 17(4): 183–210. Calabrese, J.A. 2000b. Metals, ideology and power: the manufacture and control of materialised ideology in the area of the Limpopo- Shashe confluence, c. AD 900 to 1300. The South African Archaeological Society Goodwin Series 8: 100–111. FIG. 13. Ndlulamithi Kraal. Calabrese, J.A. 2005. Ethnicity, class, and polity: the emergence of social and political complexity in the Shashi-Limpopo valley of southern far from the Limpopo River (500 m southeast) and the Africa, AD 900 to 1300. Unpublished PhD thesis. Johannesburg: Mmamagwa complex (3 km west). The formal component of University of the Witwatersrand. the lithic assemblage was relatively high (6.81%) when Cherry, J.F., Davis, J.L., Demitrack, A., Mantzourani, E., Strasser, T.F. & Talalay,L.E. 1988. Archaeological survey in an artifact-rich landscape: compared to Dzombo Shelter (4.16%) and Shawu Camp a Middle Neolithic example from Nemea, Greece. American Journal of (3.44%) as well as other local excavations (also see Van Archaeology 92(2): 159–176. Doornum 2007, 2008). Of great interest is the relatively late Cooke, C.K. & Simons, H. 1969. Mpato Shelter: Sentinel Ranch, Beit age estimate of the site and with radiocarbon dates it will be Bridge, Rhodesia: excavation results. Arnoldia (Rhodesia) 4(18): 1–9. possible to associate the lithic assemblage with either the K2 or Cowgill, G. 1989. Formal approaches in archaeology. In: Lamberg- Khami occupation of the homestead. Karlovsky, C.C. (ed.) Archaeological Thought in America: 74–88. Cam- The smallest excavated site where a single 1.5×1msquare bridge: Cambridge University Press. Denbow,J.R. 1984. Cows and kings: a spatial and economic analysis of a was dug is Shawu Camp. The site is located on top of a koppie hierarchal Early Iron Age settlement system in eastern Botswana. In: in the northern basalt zone in Mashatu. The koppie is Hall, M., Avery, G., Avery, D.M., Wilson, M.L. & Humphreys, A.J.B. surrounded by a massive CCS outcrop, composed mostly of (eds) Frontiers: Southern African Archaeology Today: 24–39. Cambridge: agate, and over 90% of the lithic assemblage is made using Cambridge Monographs in African Archaeology, BAR International CCS materials. The excavated assemblage is fairly small, with Series 207. less than 1000 artefacts (n = 97 lithics/13l bucket), and no ceramics Dornan, S.S. 1917. The Tati Bushmen (Masarwas) and their language. or organic material were found. Journal of the Anthropological Institute 47: 37–112. Ebert, J.I. 2001.Distributional Archaeology. Salt Lake City: University of Utah Press. SUMMARY Eastwood, E.B. & Cnoops, C.J.H. 1999. Capturing the spoor: towards This paper has presented an interim report of an archaeo- explaining kudu in San rock art of the Limpopo-Shashi confluence logical survey in Northern Tuli, Botswana, near Mapungubwe, area. South African Archaeological Bulletin 54: 107–119. K2 and Schroda, as well as excavated LSA sites such as Balerno Eastwood, E.B. & Eastwood, C. 2006. Capturing the Spoor: an Exploration Main and Little Muck Shelter. The final results will help us of Southern African Rock Art. Claremont: New Africa Books. better understand and interpret the archaeological record on Eastwood, E.B. & Smith, B.W. 2005. Fingerprints of the Khoekhoe: geometric and handprinted rock art in the central Limpopo basin, the Greater Mapungubwe Landscape. Future studies will southern Africa. South African Archaeological Society Goodwin Series 9: expand on the LSA and agricultural sequence in the Northern 63–76. Tuli and address questions of rock art authorship and variability Elton, F.1872. Journal of exploration on the Limpopo River. Journal of the on both sides of the Limpopo River,incorporating this informa- Royal Geographical Society of London 42: 1–49. tion into the broader archaeological framework for the region. Fish, S.K. & Kowalewski, S.A. 1990.The Archaeology of Regions: A Case for Full-Coverage Survey. Washington: Smithsonian Institute Press. ACKNOWLEDGEMENTS Foley, R. 1981. Off Site Archaeology and Human Adaptation in Eastern Africa. Cambridge: Cambridge Monographs in African Archaeology, I would like to thank the Palaeontological Scientific Trust BAR International Series 97. and its Scatterlings of Africa programmes for financial support Forssman, T.R. 2010. The Later Stone Age Occupation and Sequence of and the Mashatu Game Reserve for lodging. All members of the Mapungubwe Landscape. Unpublished MSc dissertation. Northern Tuli must be thanked for their assistance and particu- Johannesburg: University of the Witwatersrand. 70 South African Archaeological Bulletin 68 (197): 63–71, 2013

Fouché, L. 1937. Mapungubwe: Ancient Bantu Civilization on the Limpopo. Maggs, T.M.O’C. 1980. Msuluzi confluence: a seventh century Early Reports on Excavations at Mapungubwe (Northern Transvaal) from Iron Age site on the Tugela River. Annals of the Natal Museum 24: February 1933 to June 1935. Cambridge: Cambridge University Press. 111–145. Flynn, D.K. 1997. ‘We are the border’: identity, exchange, and the State Manyanga, M. 2006. Resilient landscape: socio-environmental along the Bénin-Nigeria Border. American Ethnologist 24(2): 311–330. dynamics in the Shashi-Limpopo basin, southern Zimbabwe Garlake, P.S.1967. Iron Age archaeology. In: 17th Expedition: Maramani: c. AD 800 to the present. Unpublished PhD thesis. Uppsala : Uppsala 22–25. Salisbury: Rhodesian Schools Exploration Society. University. Hall, G. 2003. A compendium of archaeological sites and artifacts from Manyanga, M., Pikirayi, I. & Ndoro, W. 2000. Coping with dryland Mashatu Game Reserve, north eastern Tuli Block, Botswana. environments: preliminary results from Mapungubwe and Zimba- Unpub lished report for Mashatu Game Reserve. bwe phase sites in the Mateke Hills, south-eastern Zimbabwe. The Hall, S.L. 2000. Forager lithics and Early Moloko homesteads at South African Archaeological Society Goodwin Series 8: 69–77. Madikwe. Natal Museum Journal of Humanities 12: 33–50. Mazonde, I.N. 1987. The development of ranching and economic Hall, S.L. & Smith, B. 2000. Empowering places: rock shelters and ritual enterprise in eastern Botswana. Unpublished PhD thesis. Manches- control in farmer-forager interactions in the Northern Province. The ter: University of Manchester. South African Archaeological Society Goodwin Series 8: 30–46. McCarthy, T. & Rubidge, B. 2005. The Story of Earth and Life: a Southern Hanisch, E.O.M. 1980. An Archaeological Interpretation of Certain Iron African Perspective on a 4.6-Billion-Year Journey. Cape Town: Struik. Age Sites in the Limpopo/Shashi Valley. Unpublished MSc disserta- Mooketsi, C.O. 2009. Archaeological surveys in the Shashi-Limpopo tion. Pretoria: University of Pretoria. Basin, eastern Botswana. In: Pwiti, G., Radimilahy, C. & Macamo, S. Hanisch, E.O.M. 1981a. Schroda: a Zhizo site in the Northern (eds) Dimensions of African Archaeology: 8–18. Dar es Salaam: E & D Transvaal. In: Voigt,E. (ed.) Guide to Archaeological Sites in the Northern Vision Publishing Limited. and Eastern Transvaal: 37–53. Pretoria: Transvaal Museum. Mosothwane, M. 2011. The Tuli mummy: a preliminary report from Hanisch, E.O.M. 1981b. The Northern Transvaal: environment and northeastern Botswana. South African Archaeological Bulletin 66(194): archaeology. In: Voigt, E.A. (ed.), Guide to Archaeological Sites In The 157–160. Northern and Eastern Transvaal. Pretoria: Transvaal Museum. Nance, J.D. & Ball, B.F.1986. No surprises? The reliability and validity of Hickman, A.S. 1972. The fight on Bryce’s store and other incidents test pit sampling. American Antiquity 51(3): 457–483. during the Boer War. Rhodesiana 26: 53–68. Orton, J. 2007. The sampling of ephemeral shell scatters in Huffman, T.N. 1984. Expressive space in the Zimbabwe culture. Man Namaqualand, South Africa. South African Archaeological Bulletin 19(4): 593–612. 62(185): 74–78. Huffman, T.N. 2000. Mapungubwe and the origins of the Zimbabwe Plug, I. 2000. Overview of Iron Age fauna from the Limpopo Valley. culture. The South African Archaeological Society Goodwin Series 8: 14–29. South African Archaeological Society Goodwin Series 8: 117–126. Huffman, T.N. 2007. Handbook to the Iron Age: the Archaeology of Reid, A. & Segobye, A. 2000. Politics, society and trade on the eastern Pre-Colonial Farming Societies in Southern Africa. Pietermaritzburg: margins of the Kalahari. South African Archaeological Society Goodwin University of KwaZulu-Natal Press. Series 8: 58–68. Huffman, T.N. 2008. Climate change during the Iron Age in the Robinson, K.R. 1964. Dombozanga Rock Shelter, Mtetengwe River, Beit Shashe-Limpopo Basin, southern Africa. Journal of Archaeological Bridge, southern Rhodesia: excavation results. Arnoldia (Rhodesia) Science 32: 2032–2047. 1(17): 50–51. Huffman, T.N. 2009. Mapungubwe and Great Zimbabwe: the origin Robinson, K.R. 1966. The Leopard’s Kopje culture, its position in the and spread of social complexity in southern Africa. Journal of Anthro- Iron Age of southern Rhodesia. South African Archaeological Bulletin pological Archaeology 28(1): 37–54. 21: 5–51. Huffman, T.N. 2010. Intensive El Niño and the Iron Age of south- Sadr, K. & Rodier, X. 2012. Google Earth, GIS and stone-walled struc- eastern Africa. Journal of Archaeological Science 37: 2572–2586. tures in southern Gauteng, South Africa. Journal of Archaeological Huffman, T.N. 2012. Historical archaeology of the Mapungubwe area: Science 39: 1034–1042. Boer, Birwa, Sotho-Tswana and Machete. Southern African Humanities Sampson, C.G. 1985. Atlas of Stone Age Settlement in the Central and Upper 24: 33–59. Seacow Valley. Bloemfontein: National Museum. Kinahan, J. 2000. Fifteenth century agropastoral responses to a Schoeman, M.H. 2006. Clouding power? Rain-control, space, land- disequilibrial ecosystem in southeastern Botswana. In: Barker, G. & scapes and ideology in Shashe-Limpopo state formation. Unpub- Gilbertson D. (eds) The Archaeology of Drylands: Living at the Margin: lished PhD thesis. Johannesburg: University of the Witwatersrand. 233–251. London: Routledge. Shennan, S.J. 1985. Experiments in the Collection and Analysis of Archaeo- Kinahan, J., Kinahan J.H.A. & Van Waarden, C. 1998.The archaeology logical Survey Data: the East Hampshire Survey. New York: Academic and symbolic dimensions of a thirteenth century village in eastern Press. Botswana. Southern African Field Archaeology 7(2): 63–71. Smith, B.W. & Ouzman, S. 2004. Taking stock: identifying Khoekhoen Kempson, H. 2007. Late Earlier Stone Age sites in the Mapungubwe herder rock art in southern Africa. Current Anthropology 45(4): National Park, South Africa: a Technological Study. Unpublished 499–526. MSc dissertation. Johannesburg: University of the Witwatersrand. Sullivan III, A.P., Mink II, P.B. & Uphus, P.M. 2007. Archaeological Krakker, J.J., Shott, M.J. & Welch, P.D. 1983. Design and evaluation of survey design, units of observation, and the characterization of shovel-test sampling in regional archaeological survey. Journal of Field regional variability. American Antiquity 72(2): 322–333. Archaeology 10(4): 469–480. Tartaron, T.F.2003. The Archaeological Survey: sampling strategies and Kuman, K., Gibbon, R.J., Kempson, H., Langejaans, G., Le Baron, J.C., field methods. Hesperia Supplements 32: 23–45. Pollarolo, L. & Sutton, M. 2005. Stone Age signatures in northernmost Thorp, C. 2010. Induna Cave, southwestern Zimbabwe: a ‘contact South Africa: early archaeology of the Mapungubwe National Park period’ assemblage in a changing social landscape. Southern African and vicinity. In: D’Errico, F.& Backwell, L. (eds) From Tools to Symbols, Humanities 22: 113–147. From Early Hominids to Modern Humans: 163–182. Johannesburg: Thomas, D.H. 1975. Nonsite sampling in archaeology: up the creek Witwatersrand University Press. without a site? In: Mueller, W.J. (ed.) Sampling in Archaeology: 61–81. Lane, P.J. 1996. Archaeological survey and excavation in southeast Tucson: University of Arizona Press. Botswana, 1992. Nyame Akuma 45: 11–23. Van Doornum, B.L. 2005. Changing places, spaces and identity in the Le Baron, J.C., Grab, S.W.& Kuman, K. 2010. Grain-size characteristics Shashe-Limpopo region of Limpopo Province, South Africa. Unpub- of surface sands and sandstones: indicators of potential sediment lished PhD thesis. Johannesburg: University of the Witwatersrand. sources for the southern Tuli Basin, South Africa. South African Van Doornum, B.L. 2007. Tshisiku Shelter and the Shashe-Limpopo Geographical Journal 92(1): 7–21. confluence area hunter-gatherer sequence. Southern African Human- Lewarch, D.E. & O’Brien, M.J. 1981. The expanding role of surface ities 19: 17–67. assemblages in archaeological research. Advances in Archaeological Van Doornum, B.L. 2008. Sheltered from change: hunter-gatherer Method and Theory 4: 297–342. occupation of Balerno Main Shelter, Shashe-Limpopo confluence Loubser, J.N. 1991. The ethnoarchaeology of Venda-speakers in south- area, South Africa. Southern African Humanities 20: 249–284. ern Africa. Navorsinge van die Nasionale Museum, Bloemfontein 7(8): Van Waarden, C. 1979. Archaeological investigation of Leeukop: a 145–164. functional approach. Botswana Notes and Records 12: 151–164. South African Archaeological Bulletin 68 (197): 63–71, 2013 71

Van Waarden, C. 1980. Leeukop: a functional analysis of a refugee Pwiti, G. (ed.) Caves, Monuments and Texts: Zimbabwean Archaeology period settlement in eastern Botswana. Unpublished MA thesis. Today: 9–32. Studies in African Archaeology 14. Uppsala: Uppsala Peterborough: Trent University. University. Voigt, E.A. & Plug, I. 1981. Early Iron Age Herders of the Limpopo Valley. Wandsnider, L. & Camilli, E.L. 1992. The character of surface archaeo- Pretoria: Transvaal Museum. logical deposits and its influence on survey accuracy. Journal of Field Wadley, L. 1989. Legacies from the Later Stone Age. The South African Archaeology 19(2): 169–188. Society Goodwin Series 6: 42–53. Wells, E.C., Terry, R.E., Parnell, J.J., Hardin, P.J., Jackson, M.W. & Wadley, L. 1996. Changes in the social relations of precolonial Houton, S.D. 2000. Chemical analysis of ancient anthrosols in resi- hunter-gatherers after agropastoralist contact: an example from the dential areas at Piedras Negras, Guatamala. Journal of Archaeological Magaliesberg, South Africa. Journal of Anthropological Archaeology 15: Science 27: 449–462. 205–217. Wood, M. 2000. Making connections: relationships between interna- Walker,N.J. 1994. The Late Stone Age of Botswana: some recent excava- tional trade and glass beads from the Shashe-Limpopo area. The tions. Botswana Notes and Records 26: 1–35. South African Archaeological Society Goodwin Series 8: 47–50. Walker, N.J. 2009. Test excavation at Lepokole Cave, eastern Botswana. Wood, M. 2005. Glass beads and pre-European trade in the Botswana Notes and Records 41: 17–24. Shashe-Limpopo region. Unpublished MSc dissertation. Johannes- Walker, N.J. & Thorp, C. 1997. Stone Age Archaeology in Zimbabwe. In: burg: University of the Witwatersrand.