The Wetlands of the Okavango Delta and Boteti River

The central Kalahari basin of Botswana is a sand-filled depression bordered to the north, east, and south by the higher, better-watered landscapes of the Bie Plateau of north-central Angola, the Katanga Plateau of northern Zambia, the Muchinga Mountains of eastern Zambia and Malawi, the Matopo Hills of southwestern Zimbabwe, and the Bushveld or Highveld complex of South Africa and eastern Botswana. To the west, the Kalahari thirstland gradually gives way to the stark landscape of the Namib Desert, where shifting sand dunes and arid, stony flats parallel the Atlantic coastline northward from the Gariep or Orange River on the border with South Africa to cross the Cunene River into southwestern Angola.

In a stunning contrast to this generally arid landscape, the waters of the Okavango Delta—the largest inland delta in the world—spill out over the northern sands of the Kalahari Desert in northern Botswana (Map 1).

Open in new tab

Map 1. Map of Botswana showing the locations of archaeological sites and geographic features mentioned in the text.

What water is not lost to transpiration along the way ultimately disappears in Lake Ngami to the west and the vast salt flats of the Makgadigkadi salt pans to the east. The permanent and seasonal waterways of the delta cover almost fifteen thousand square kilometers at the height of the seasonal floods. The water comes not from local rainfall but from the annual rains that fall on the Bie Plateau between October and May. Carried southward by the Cubango River, the water takes about six months to percolate through the permanent papyrus swamp of the northern Okavango to its southern end, where it rises in Maun during the dry season between May and September. Because the annual rise and fall of flood waters is out of sync with seasonal rainfall in Botswana, water is most prevalent at seasonal pans in the desert during the local rainy season when the waters of the Delta and Boteti River are at their lowest. As the desert dries out during the dry season, the waters of the southern Okavango, Lake Ngami, and Boteti reach their highest levels. These fluctuations produce a contrasting seasonal dynamic as resource availability shifts between wetland and desert micro-environments, a dynamic that foragers, fishers, farmers, and herders have exploited in a variety of ways over the past three millennia.

The Makgadikgadi Salt Pans

The Makgadikgadi pans, among the largest salt pans on earth, contain the evaporate remains of an extensive Pleistocene lake that covered over eighty thousand square kilometers forty thousand years ago.¹⁶ The paleo-lake gradually disappeared over the course of the Holocene and is now dry except in years of exceptional rains such as those in 2014. Open grasslands dotted with occasional clusters of palm trees border the pans on the west; mixed grassland and mopane forest occur to the south and east. Before veterinary cordon fences were built in the 1970s to control the spread of livestock diseases, an annual migration of gregarious herds of water-dependent game that included blue wildebeest (Connochaetes taurinus), zebra (Equus quagga), and Cape buffalo (Syncerus caffer) moved from the grasslands and seasonal pans of the desert in the summer rainy season to drink at the rising waters of the Boteti in the winter.¹⁷ For hunters, game would have been especially abundant along the Boteti River and southern Okavango during the winter dry season as the river rose. During the summer rainy season as the desert pans filled, herds moved back to the desert to exploit fresh pastures, allowing the grazing along the Boteti to regenerate.

Fishing would have been most productive at the beginning of the rainy season as receding floods left behind drying pools in the southern Okavango and Boteti where large numbers of bream (Tilapia sp.) and barbel (Claris sp.) could be efficiently harvested as the Yeyi do today.¹⁸ Along the Boteti River, Khoe speakers used stone-walled traps to capture fish as the flood waters receded.¹⁹ In recent times, traders in dugout canoes took advantage of the winter floods to cross from the permanent swamps of the northern Okavango to the seasonally flooded southern Okavango to exchange fish, dried and salted meat, and wild sycamore figs or mukuchomo/mukutshumo (Ficus sycomorus) for commercial goods in Maun.²⁰

There is also evidence for early trade in lithics through the waterways of the Boteti and Okavango. Silcrete from Middle and Later Stone Age levels at White Paintings Shelter in the Tsodilo Hills, for instance, has been traced to outcrops along the Boteti River 250 kilometers away.²¹ It is not known whether the raw material was directly procured and carried over these distances as part of seasonal or long-term transhumance by highly mobile groups, or whether it was passed from group to group through exchange networks that followed the river systems. Silcrete tools have also been recovered from Iron Age settlements at Tsodilo, Matlapaneng, just outside Maun, and at Bosutswe in eastern Botswana, indicating an overlap in the use of lithic resources by farmers and foragers.²²

When domesticated animals were introduced two millennia ago, the changes in grazing and water availability brought on by the contrasting floods and rains would have been important considerations for forager-herders as well as foragers. This would have been especially important if cattle were present since they require more water and grazing than game or fat-tailed sheep. One would therefore expect to find cattle-foragers more likely than sheep-foragers to concentrate near more permanent waterways. The earliest dates for both cattle and sheep presently come from the site of Toteng on the eastern edge of Lake Ngami where they occur in Later Stone Age contexts dating to the beginning of the first millennium.²³

One constraint for early herder-foragers would have been the incidence of tsetse fly, which is prevalent throughout much of the Okavango system. The exact boundaries of its distribution varied through time, however, and it may have been less of a problem along the Boteti River where the narrow gallery forests that follow the river (a location where tsetse could breed) quickly give way to open grasslands away from the river. Seasonal migrations of African buffalo (Syncerus caffer) and wildebeest (Connochaetes taurinus) from the Kalahari Desert to the Boteti River would have posed a threat to the establishment of cattle herds, however, as they act as a reservoir for diseases, such as malignant catarrhal fever, that are deadly to cattle.²⁴ The development of pro-active herding strategies that would avoid overlap between cattle and wildebeest at the river would have been important to the development of early herding economies such as that documented as early as the 7th century at the site of Kaitshàa, overlooking the southern margin of Sowa saltpan.²⁵ The faunal remains indicate a diet dominated by sheep and cattle, with less than 3 percent of all the identified bovids coming from wild species—none of them wildebeest.

Linguistics

Güldemann, Güldemann and Elderkin, and Ehret have recently proposed a dramatic revision of the so-called Khoisan language family, arguing that it is not a true family, but a composite of three distant or unrelated families—Khoe, Northern San, and Southern San. Proto-Khoe languages had a long history of separate development in East Africa before their arrival in the northern Kalahari approximately 2,500 years ago.²⁶ They propose that an early Proto-Khoe-Kwadi population originating in East Africa colonized southern Africa as small-stock forager-pastoralists almost a millennium before the arrival of Bantu-speaking agro-pastoralists. If this is correct, then not all the populations once lumped together under the general term “Khoisan” emerged from a common hunter-gatherer ancestral stock in southern Africa. Instead, there was a long period of separate development of Khoe languages in East Africa, and San languages in southern Africa, before the arrival of Proto-Khoe-Kwadi speakers around 500 bce. Güldemann further suggests that these immigrant Khoe-Kwadi forager-herders possessed a non-Khoisan genetic phenotype that progressively altered as they mixed with existing San populations in central Botswana, South Africa, and Namibia.

Genetics

Recent studies of East African Hadza and Sandawe populations suggest they derive a fraction of their ancestry from an admixture with a population related to the Southern African Khoisan, from whom they appear to have been separated for approximately thirty thousand years according to some statistical models.²⁷ The present-day Southern African Khoe and San fall into two genetic groups that reflect this different heritage and occupy loosely different regions of the northwestern and southeastern Kalahari. Additional Y-chromosomal evidence suggests the hypothesis that the spread of Khoe small-stock pastoralism resulted from a population diffusion from eastern to southern Africa approximately two thousand years ago.²⁸ Using a study of distinctive lactase persistence (LP) alleles that would allow for the digestion of milk into adulthood, Macholdt et al. found LP alleles were present in East African herder populations and Khoe populations in southern Africa, but occurred in significantly lower frequencies in Northern and Southern San.²⁹ These findings suggest that herding and milking had a long history among Khoe populations but not among Northern or Southern San. Finally, studies by Henn et al. found that the geographic distribution of the E3b1f haplogroup is also consistent with the hypothesis of an expansion of Khoe-speaking herder-foragers from Tanzania into southern-central Africa as a population distinct from the San.³⁰ This expansion was also independent of a later migration of Bantu-speaking peoples who followed a similar route from East Africa. Such genetic studies are still in their infancy, however, and many populations and regions in eastern and southern Africa have yet to be sampled. As a result, the regional cultural typologies based on ceramic styles produced by archaeologists, the language groupings proposed by linguists, and the statistical relationships proposed by geneticists are often difficult to reconcile into a consistent historical and geographic narrative.

What is known of the socioeconomic and linguistic relationships between Khoe and Bantu speakers on the Makgadikgadi pans is equally complex. In this area, interaction between early Khoe- and Bantu-speaking peoples resulted in a more hybrid merging of Bantu and Khoe. Alan Barnard describes it thus:

Most of the Eastern and Northern Khoe Bushmen are fully integrated into the economic milieu of the wider, Bantu-speaking society. Many especially in the north, are dark skinned, and in appearance as well as genetic makeup, similar to the black populations which surround them. The genetic evidence suggests discrete “black” and “red” populations (as they occasionally designate themselves), rather than a gradual distinction between the two (Nurse and Jenkins 1977: 4; Chasko et al. 1979). … Still this does not explain how they came to speak the languages they do. Westphal (1963: 260) cites legends among Bantu speakers of “small groups of primitives” who lived by hunting and gathering, lacked chiefs, and even fire. Cashdan (1986: 152–3) suggests that such groups may also be ancestors of present day Khoe-speaking Bushmen of northern and eastern Botswana, excluding the G//ana who have migrated eastwards from the Central Kalahari Game Reserve.³¹

New studies provide more detailed genetic evidence for mate exchange in the genetic inheritance of local Shua (Khoe) residents who live around the Makgadikgadi salt pans. Pickrell et al. estimate that 70 percent of current Shua ancestry is Bantu, suggesting that twenty-two or twenty-three generations would be required to produce this outcome.³² They place the beginning of intermating between the two groups at about 550–690 years ago (with twenty-five- and thirty-year generations, respectively), which translates in calendar years to the years 1320–1460. Current Shua-Khoe have just 10 percent Khoisan, paternal Y-chromosomes, and 40 percent Khoisan, maternal mtDNA.³³ This indicates a high degree of hypergamy in which Shua women were disproportionably finding partners among Bantu-speaking men. This pattern is evident in other hunter-gatherer/farmer relations elsewhere, including Sotho-Tswana and Nguni speakers much further south. In the later case, the adoption of “clicks” from Khoi or San languages is also evident. Contemporary ethnic distinctions between “black” and “red” populations should therefore be viewed for what they are—contemporary cultural constructions of identity rather than scientific indicators of long-term separation between historically discrete populations. Dating genetic models of hypergamy is also subject to many assumptions that may affect its accuracy, particularly over relatively short periods such as the last two millennia. While archaeological evidence can provide an additional perspective on these processes and may help to contextualize and date them, this also involves assumptions about the genetic and linguistic identities that archaeologists associate with particular material remains, ceramic styles, and economic typologies.

Attributions of forager or farmer economies do not follow once-accepted categorizations that contrasted LSA foragers and ceramic-using forager-herders as speakers of Khoisan languages with farmers speaking Bantu languages and practicing economies dominated by farming and herding, ceramics and metal. The contemporary linguistic and genetic findings from the Makgadikgadi indicate that the identity of prehistoric farmers and foragers—and their interaction—requires a more multidimensional and multivariate approach; it cannot be simply read from counts of ceramics, stone tool typologies, or percentages of domesticated and wild faunal in archaeological assemblages.

Divuyu

The ceramics recovered from the 7th–9th-century site of Divuyu are among the earliest recovered in the sandveld west of the Okavango and belong to an Iron Age tradition with origins to the north among Western Bantu speakers in central Angola.³⁹ Excavations at the site, which is situated on top of the “female hill” at Tsodilo, seventy kilometers west of the Okavango, recovered quantities of carbonized mongongo nuts (Schinziophyton rautaneii), an important wild plant food in the diet of local foragers, as well as farmers and small-stock herders.⁴⁰ While no other plant remains were recovered, a stable isotopic analysis of the single human skeleton found at the site produced a δ‎¹³C value (8.8 ‰), indicating a diet consisting almost entirely of domesticated plants.⁴¹ A mix of hunted game and domestic sheep and goats supplemented by small quantities of fish, birds, and tortoise provided animal protein.⁴² Fishing has a history at Tsodilo that extends back into the Middle Stone Age at White Paintings Shelter.⁴³ At Divuyu, there was an extensive use of jewelry and metal tools, including chisels that may have been used to mine or process specular hematite, which in historic times was widely traded. It was ground into a glittering blue powder and sprinkled on the hair and body as a cosmetic.⁴⁴ Very few lithics were recovered, suggesting there was little interaction with stone-using hunters at this time.⁴⁵

Nqoma

The 8th–12th-century site of Nqoma, on a lower plateau of the female hill, produced a similar mix of domestic and wild animals along with clumps of carbonized sorghum (seed samples identified by J. M. J. de Wet, University of Illinois, in a personal communication to Denbow).⁴⁶ Cattle dominated the animal sector of the economy, followed by small-stock and hunted game. Carbon isotopic analyses of three human skeletons from the site produced δ‎¹³C values (−10.5‰, −9.3‰, and −9.3‰) that indicate diets dominated by domesticated grains.⁴⁷ Even more iron and copper jewelry was recovered at Nqoma than Divuyu, along with significant numbers of lithics that included formal segments and scrapers. The contrast with Divuyu suggests increased interaction with LSA hunters who may have brought game, hides, or foodstuffs to the site. LSA populations may have also been pressed into service as labor for the extensive specularite mines at Tsodilo. The archaeological evidence thus suggests that while some peoples of Later Stone Age heritage were attracted to and perhaps incorporated within these settlements, others lived in neighboring rock shelters with less access to metal, ceramics, and domesticated plants and animals.⁴⁸ Such spatial and resource separation suggests social stratification.

Matlapaneng

At the 9th–12th-century site of Matlapaneng outside Maun on the southern edge of the Okavango, carbonized remains of domesticated sorghum, pearl millet, a cucurbit seed, and wild Grewia seeds were recovered using flotation methods.⁴⁹ Here, cattle remains made up 31 percent of the faunal assemblage, followed by 23 percent goats/sheep (caprines); the remaining fauna came from a variety of wild species.⁵⁰ Interestingly, no fish remains were recovered despite good preservation and the site’s location on the banks of the Thamalakane River. The ceramics from Matlapaneng link it culturally with both Nqoma in the Tsodilo Hills, where small quantities of fish remains have been recovered, and with Kalundu Tradition sites in the Victoria Falls region of southwestern Zambia, where fish remains are absent or rare.⁵¹ The narrow occupation layer at Matlapaneng, which is overlain by approximately thirty-five centimeters of sterile sand, does not indicate successive occupations but rather a multiethnic occupation of Bantu speakers drawn from multiple locations west and east of the Okavango.

Stone points manufactured from local silcrete recovered at Matlapaneng, along with chert and quartzite cores, segments, scrapers, and flakes, indicate that stone supplemented metal tools.⁵² The variety of lithics suggests many of these tools were made at the site. From the lithics alone, however, it cannot be determined whether they were made by a distinct subset of the population, perhaps Khoe- or San-speaking foragers, or whether they were made by Bantu-speaking agro-pastoralists familiar with LSA lithic technology. Perhaps they were even the product of a hybrid population made up of people drawn from Bantu and Khoe or San stock.

Xaro1

The Xaro1 site is situated on the banks of the Okavango River at the northern edge of the Okavango Delta. It has an upper component containing Hambukushu pottery dated to the late 18th or early 19th century. Below this is a sterile layer of blocky clay that most likely derives from a period when the site was flooded and covered in papyrus mat. Beneath this layer lay charcoal-tempered ceramics identical to those from Divuyu, along with others that are stylistically less distinctive. The wet condition of delta soils was not favorable for the preservation of iron. Only seven lithics were recovered, however, suggesting that iron or perhaps wood was favored for tools.

Despite the abundance of “Iron Age” ceramics, no remains of domesticated animals other than dogs were recovered from the lower horizon.⁵³ Identified game included zebra (Equus quagga), river otter (Lutra maculicollis), and a variety of medium to large bovids. Overall, however, the faunal was dominated by fish remains, predominately catfish (Clarias/Synodontis). Many Chelonia sp. carapaces were also common and, given the locale, these most likely came from terrapin rather than land tortoise. Many of the animal bones exhibited evidence of burning, which suggests roasting over an open fire rather than boiling in pots. Many Chelonia carapaces were burned on the outside, indicating they were cooked in the shell. None of the fish bone showed evidence of burning, and it is likely that fish were either boiled in pots or roasted over low heat. No plant remains were preserved, but stable carbon isotope analyses of two adult human skeletons, one definitely associated with a decorated potsherd belonging to the Divuyu tradition, produced δ‎¹³C values (−16.6‰ and −16.9‰), indicating a diet in which C4 domesticated plants were present, but played only a minor role.⁵⁴

Qogana

The site of Qogana is situated on the edge of a lagoon on the northeastern side of the permanent swamp, almost due east of Xaro1. Two two-by-two-meter test pits were excavated at the site, one in a probable occupation area where ceramics and the remains of a clay-plastered reed structure were uncovered. Fragments from a similar burned reed structure were also exposed at Nqoma. Such structures are still commonly seen around the Okavango today. The second test pit, fifty meters away from the first, revealed a bone-filled refuse midden. Although a calibrated 9th-century radiocarbon date aligns temporally with the date suggested by the Divuyu ceramics from Xaro1, the ceramics from Qogana are different. They are charcoal-tempered like most other ceramics from this period in northern Botswana, but most vessels were globular rather than necked jars with rolled, combstamped rims similar to those recovered in the lower levels of Serondela east of the Okavango near the Chobe-Zambezi confluence. Comparable levels at Serondela also date to the 9th century but included features normally associated with a farming village such as figurines of domestic cattle, iron fragments, and burned chunks from pole-impressed clay or daga structures.⁵⁵

There is considerable overlap in the material culture at all these sites: charcoal-tempered ceramics, clay-plastered housing, and the use of metal tools, which, at Nqoma and Matlapaneng, were supplemented by stone tools. Yet at the same time, there is considerable variety in economic adaptations, which range from small-stock and cattle-based agro-pastoralism at Divuyu, Nqoma, Matlapaneng, and Serondella, to the predominately foraging and fishing economy at Xaro1, and a hunting and fishing economy at Qogana. The similarities in ceramic styles indicate some form of connection or communication existed between these communities, but the varying economies make it uncertain whether they represent early Khoe adaptations to the Delta similar to those of the present-day Buga-khoe and Dxericu, or a subset of an early Bantu-speaking population that shifted to fishing and hunting, or hybrid communities of both Khoe and Bantu stock. In any of these cases, the variety of adaptations suggests active processes of ethnogenesis as communities responded to differing micro-environmental habitats by developing specialized adaptations—adaptations still found among present-day Yeyi and Khoe (Kxoe) in the Delta.⁵⁶

The language of contemporary Yeyi peoples is similarly intermediate, with 10–15 percent of their lexicon containing clicks that were likely adopted from, or inspired by, neighboring Khoe (Kxoe) languages.⁵⁷ Different from the Makgadikgadi pans, in the Okavango, the socio-linguistic dynamics of contact and interaction appears to have resulted in overlapping, but to some degree socially separate, farmer, forager, and fisher adaptations. Unfortunately, genetic information of these Delta populations comparable to what scholars know of the Makgadigkadi pans is not yet available. Such information could help illuminate the early social dynamics of the Okavango region.

Kaitshàa: 650–1000 CE

The site of Kaitshàa, a name meaning “vulture water” in the local Khoe dialect, lies on a hundred-meter-high escarpment overlooking the southern edge of Sowa salt pan. A small test excavation at the site in 1996 dated the settlement to early in the second millennium but found little evidence for connections with long-distance networks.⁵⁸ Later excavations in 2011 using finer sieves focused on uncovering an earlier component that was overlooked in the earlier excavation. These excavations revealed extensive Zhizo deposits dating to between 650 and 900 in some parts of the site. More surprising, however, was the evidence for early long-distance trade provided by the discovery of more than two hundred glass beads from the early levels. These indicated the settlement participated in long-distance East Coast trade networks at an unexpectedly early date.⁵⁹

Several of the beads have a glass chemistry that placed them in a rare bead grouping (Chibuene) that pre-dates 700. These Chibuene series beads have only been found at four sites: Chibuene on the Mocambique coast, Kaitshàa and Thabadimasego on the Makgadikgadi pans, and Nqoma almost 1,800 kilometers from the Indian Ocean in northwestern Botswana.⁶⁰ The early beads attest to early intracontinental trade in salt and specularite that linked these settlements in the far interior of southern Africa with luxury trade networks that reached the Indian Ocean. The speed at which early farming communities mapped onto important natural resources across this vast and relatively unpopulated region suggests they took advantage of preexisting Khoe and San exchange systems that would have served to consolidate rights to resources as well as to ensure the long-term livelihood and well-being of small populations in an uncertain and varied environment. Non-luxury products were likely included in these exchanges, but they are difficult to identify except in unusual circumstances such as the discovery of water-restricted sitatunga from pre-1000 levels at Bosutswe, more than 150 kilometers from any habitat where they could have lived.⁶¹

The present genetic and linguistic makeup of Makgadikgadi populations suggests that early interaction between indigenous Khoe foragers and herders and Bantu-speaking agro-pastoralists resulted in mixed populations that predate the present divisions into distinct Khoe, Tswana, and Kalanga communities. The socio-linguistic context of this ethnogenesis must have been very different from the more recent past, as it favored not only intermarriage but also the ascendance of Khoe over Bantu languages.

Even so, the faunal remains indicate that the early domestic economy was focused almost exclusively on domesticated animals, with identified small stock and cattle outnumbering game (impala and buffalo) by fifty to one.⁶² This is surprising given the large herds of zebra, wildebeest, and buffalo that still graze the Makgadikgadi grasslands, but it is perhaps indicative of the need to separate cattle from wild disease vectors such as wildebeest. The economic incentives of the salt trade that brought exotic goods to the region could also have been a factor that directed labor supplies away from subsistence hunting—at least in the period before 1000. Chickens, originally a Southeast Asian domesticate that likely accompanied the glass bead trade from the coast, are also present. Oval grinding stones suggest that sorghum or millet were grown, but no carbonized seeds were recovered. Unfortunately, the single human skeleton recovered in 1996 has been lost, so carbon isotopes cannot be used to determine the proportion of wild or domesticated plants in the diet.

Kaitshàa: 1000–1200 CE

The upper levels of the Kaitshàa dating between 1000 and 1200 represent another inflection point in the history of the site. The ceramics from these levels belong to the Leopard’s Kopje rather than Zhizo tradition. Despite evidence for continuing linkages with the East Coast trade, only four glass trade beads were recovered from these levels.⁶³ Salt was undoubtedly still harvested from the pans, but few trade beads were received in exchange. On the other hand, almost a hundred ivory fragments were recovered from the early second-millennium levels (compared with just two bangle fragments from the Zhizo component). The increase in ivory at Kaitshàa, along with the discovery of a cache of nine ivory bangles at the nearby site of Mosu 1 dated to the 10th–13th centuries, demonstrates an increase in the importance of ivory in long-distance trade from the region.⁶⁴ However, the local return, at least as measured in numbers of luxury glass beads or marine shells, was very little when compared to the earlier Zhizo component. Sheep and cattle continued to dominate the Leopard’s Kopje levels, but the remains of only fifteen wild animals, primarily impala and buffalo, were identified. Oval grindstones suggest domesticated crops were grown.

The dramatic decrease in imported glass beads in the later levels at Kaitshàa suggest, in regional terms, that the inhabitants of the site were now at an economic and social disadvantage. This change correlates with the rise of more centralized polities to the east at Bosutswe, Toutswe, and K2. These emerging chiefdoms dominated the political landscape and the luxury trade across southern Africa at the beginning of the second millennium. It is not possible to determine whether the change in the terms of trade at Kaitshàa was translated into greater social or ethnic differentiation between emerging Khoe and Bantu identities on the Makgadikgadi. The construction of a stone wall to block access to the headland from the surrounding plain, however, suggests that defense was now an important consideration as Leopard’s Kopje peoples settled the region and perhaps entered into new social and economic relationships with their forager or herder-forager neighbors.

Tora Nju: 1300–1400 CE

As power was consolidated at Mapungubwe and then Great Zimbabwe between 1200 and 1500, a number of stone-walled enclosures were built around the edge of the Makgadikgadi at Tora Nju, Tlapana, Mosu 1, and Lekhubu or Gaing-o.⁶⁵ Stylistically, these ruins resemble the residences of small-scale regional authorities of the Zimbabwe culture. Two of largest ruins, however—Tora Nju and Gaing-o—have Khoe names and are today associated by local people with Khoe, not Bantu, ownership or ritual practices.⁶⁶ Indeed, the only name for the Tora Nju ruin is Khoe, which translates as “house of a god.”

The glass beads, ceramics, and a single radiocarbon date from Tora Nju indicate an occupation between 1400 and 1600. While this suggests the ruin was built by incoming Kalanga peoples as part of an extension of Zimbabwe hegemony into the region from the east, both Khoe and Kalanga speakers associate it solely with Khoe ownership. While little archaeological work has been carried out, the spacing of stone ruins around the eastern and southern edges of Sowa Pan suggest they represent another inflection point in the region’s history, one in which more centralized authority was imposed.

The ruins might represent a period when more contemporary distinctions between Khoe and Bantu cultural identities were emerging. Livingstone reports separate Kalanga, Tswana, and Khoe villages, with cultural and linguistic distinctions among them. However, in the mid-19th century, the cattle-keeping Khoe were not painted as a subservient class working for their Bantu-speaking neighbors, but as independent herders in their own right. He also makes it clear that other Khoe communities were competent elephant hunters, carrying on a tradition that may extend back to the beginning of the second millennium at Kaitshàa. The early historical record, though spotty, suggests there were many microeconomic adaptations at play in a regional, mosaic economy where ethnic distinctions only loosely paralleled economic or linguistic categories.

By the end of the 19th century, a far different picture is presented. As European demand for ivory increased, Tswana overlords—acting as intermediaries with European traders—came to control the trade, if not the actual elephant hunting. Whatever social status the Khoe language and cultural heritage might have been afforded in the past, it was now demeaned as the Khoe were reduced to a subservient class. As they were disenfranchised from their herds, their languages also fell into severe decline and many are now extinct.⁶⁷ As domestic livestock herds grew in the early 20th century they competed with wild game for grazing, water, and scarce labor supplies, the latter often drawn from Khoe or San sources. The long history of interaction, intermarriage, and ethnogenesis that had pertained in earlier centuries was forgotten. It was replaced with a truncated, false history designed to legitimize an apartheid-like dominion of Bantu farmers over Khoe foragers. In the absence of archaeological, linguistic, or genetic evidence to the contrary, this picture was extended uncritically into the deep past.