From the early 1980s, Iron Age archaeologists in South Africa have been somewhat captive to a structuralist model for the interpretation of settlement space. This trend grew out of Adam Kuper’s book Wives for Cattle (1982), in which an analysis of bridewealth and marriage patterns for Southern Bantu-speakers isolated several binary oppositions that categorized people in space at several different scales. With this synthesis Iron Age archaeologists could infuse spatial organization with the determining logic of social and cosmological structure (for examples of this approach see Huffman 1982; Evers 1984; Pistorius 1992).
It is hardly surprising that gendered space figures prominently in the model, which became known as the Central Cattle Pattern (or the CCP). Consequently, one opposition orientates gender in space through a left/right distinction, while a center/surround opposition underpins prime male political and economic power, bounded by the domestic domain of women and children. Men accumulate and control cattle and exchange them as bridewealth with other men for wives and their productive and reproductive labor. As the medium through which all social, legal, political, and ritual transactions flow, cattle, their corrals, and the male court are located centrally in the homestead. This explicit power is juxtaposed with the outer residential circle of houses, in which women are more spatially constrained.
However, as has been repeatedly stated in the literature and throughout this volume, these models reify structure and impose an uncritical durability on meaning that compresses time and muffles the contextual detail of social relations. In their rather mechanical application, structuralist models distance people from social action, which becomes a set of rules within which people inflexibly live. While the Central Cattle Pattern is unquestionably gendered, it is as though men and women occupied two separate worlds, where, in an immutable framework of categories, people do not interact or contest power through the manipulation of their symbolic framework in historically distinctive ways.
The prime objective of this paper is to approach gender relations as a dynamic that is under perpetual negotiation, whereby “structures produce not rules but dispositions, and underlie not determinacy but strategy” (Miller 1995:103). Some control over economic and political conditions are obviously critical in order to discuss changes in gender relations as strategic dispositions within specific historical contexts. This is attempted through a closer examination of the Sotho/Tswana1 sequence in the western Highveld of South Africa from about AD 1400 (Figure 13.1). In contrast to static and ahistoric ethnographic models of Tswana society, there were relatively major political and economic shifts in this sequence that are recognized through significant changes in settlement organization and other categories of material culture. A central theme in correlating these material changes with gender interaction stems from widespread African beliefs in the socially negative force of pollution. These beliefs highlight dangers, usually to men, that result from improper sexual or social contact. As a result, there is always a concern to regulate social categories and the contexts in which interactions between them occur, particularly where women, through their sexual ambiguity, pose a threat to social and economic process. Anxieties about the maintenance of social order are materially expressed through the construction of boundaries that are symbolically displayed and patrolled. This paper attempts to recognize the physical and symbolic boundaries that regulated interaction between men and women and, in turn, see these as analogies for a wider social order (Douglas 1995:3).
The primary archaeological residue examined in this context is the “house”2 (Lane, Chapter 10). Among Tswana-speakers, the house is the dominant setting for the physical and symbolic mediation of day-to-day encounters between men and women. As Comaroff (1985:54) states, the house “invisibly tunes the minds and bodies of those who people them to their inner logic.” A second item of material culture used in this discussion is the ubiquitous artifact of Iron Age archaeology—the pot. It is surprising that while space became symbolic and gendered, albeit within a structuralist framework, Iron Age ceramic analysis continued to focus on mapping out large-scale relationships in time and space. Abstract ceramic classifications focused on migrations, streams, and traditions within the Early Iron Age and the nature of the Early Iron Age/Late Iron Age break (Phillipson 1977; Evers 1983; Mason 1983; Maggs 1984; Huffman 1989). At this scale of analysis the focus on large, regional linguistic and ethnic correlations completely submerges the more immediate social, economic, and ritual contexts of ceramic production, use, and discard. In this case, women, as both producers and significant users and manipulators of the most ubiquitous artifact in Late Iron Age archaeology, become essentially invisible. In contrast, it is instructive to note how the male transformation of iron ore and its accompanying symbolism has received much more attention in the literature. “Smiths constitute an ongoing obsession of social anthropology while potters appear only in footnotes” (Barley 1984:93 from Herbert 1993). As a homology for the human body and the medium in which food is transformed and served, pots provide a rich means of symbolic communication that is socially exploited as they are moved parallel to, and across, boundaries. Consequently, I examine more sociological approaches that examine the role of pottery as an active ingredient in the construction of social categories and social interaction (Braithwaite 1982; Welbourn 1984; David et al. 1988).
Figure 13.1. Map showing the area under discussion and the location of some Late Iron Age (Tswana) sites.
With this background in place, the paper proceeds in two parts. The first describes, contextualizes, and interprets spatial change and the associated gender relations, predominantly at the level of individual households, while the second adds and articulates pottery to this. In sum, I suggest that from AD 1700, Tswana spatial and symbolic order was tuned more to defining a gender hierarchy, whereas, in the early Tswana period, discernible material and symbolic boundaries suggest gender symmetry.
Ethnographers of the Southern Bantu draw sharp distinctions between the dispersed, exogamous households of Nguni-speakers and the endogamous towns of Sotho, and in particular, Tswana-speakers (Sansom 1974). From the ethnographic perspective, the large aggregated ward clusters (towns) that typify Tswana settlement patterns have been reified as an enduring cultural preference, as well as an adaptation to the relatively arid terrain over which Tswana-speakers have been historically settled. This comprises the inland plateau region to the north, south, and west of present-day Johannesburg, as well as eastern Botswana (Figure 13.1).
However, the pattern of town living as a defining characteristic of Tswana life breaks down under archaeological scrutiny. This shows that Sotho-speakers can be archaeologically identified from about AD 1400 on the Highveld, but that their initial villages were small and dispersed. The historically noted attribute of large ward clusters is a recent innovation, dating from the 18th century (Huffman 1986; Hall 1995). Clearly, the shift from dispersed homesteads to large aggregated ward clusters indicates considerable dynamism in political, economic, and social organization. These changes are most visible in the areas west of present-day Johannesburg, and it is in this area that attention is focused.
Here, the precolonial Tswana sequence can be subdivided into three phases, using either changes to settlement size and the presence or absence of stone walls to demarcate internal space, or the boundary between a homestead and the bush beyond. The start of the first phase is identified through a significant break in ceramic style from the preceding Eiland phase of the Early Iron Age. Evers (1983) correctly identifies an abrupt break between the final 11th- to 13th-century phase of the Early Iron Age and the following Sotho/Tswana sequence (Evers 1981). Radiocarbon dates from Icon and other sites show that early Sotho-speakers first appeared between AD 1300 and 1400 south of the Limpopo River, and farther to the southwest from about AD 1500 (Figure 13.1). Southern Tanzania is a possible point of Sotho origins (Huffman 1989, 1996c).
The first phase is poorly recorded, even though archaeological visibility is high. Mason has excavated several sites at Olifantspoort and elsewhere in the Magaliesberg, and at Rooiberg (Mason 1986: Fig. 1). Other data has been collected by Hanisch (1979), Hall (1981), and Boeyens (1997), among others. An understanding of settlement organization is limited to the location of hut floors and to detail on these floors. Villages are small, with maximum dimensions of about 100 m squared, and comprise loose circles of huts around a central cattle corral (Figure 13.2). The suggestion is that these were relatively short, single-component occupations (Hanisch 1979; Mason 1986).
Figure 13.2. Site plan of a first-phase homestead and a composite plan of a first-phase floor.
Apart from the position of individual huts there is little other spatial evidence. There are no indications that individual households were separated by walls from their neighbors, nor is there any indication if or how homestead perimeters were marked off from the surrounding bush. Cattle must have been penned within pole or brush fences. Poor preservation limits clarity on midden location, though there is some suggestion from Irrigasie that a single homestead had several separate middens and that these may have been located behind huts (Lathey 1995). At Rooiberg small stone circles are interpreted by Mason (1986) as grain bin bases. These are located either right next to hut floors, or to the side and front of huts. Overall, the evidence indicates that first phase homesteads were uniformly small and dispersed, and probably part of small-scale chiefdoms with a political hierarchy of two levels, and certainly no more than three (see Huffman 1986).
The lack of spatial detail around huts is in sharp contrast to the wealth of detail preserved on the hut floors (Figure 13.2). This is particularly so for Mason’s excavations of first-phase sites in the Olifantspoort area, particularly site 29/72 (Mason 1986:241 ff.), and Boeyens’s excavations further west at Rietfontein (Boeyens 1997). The large amounts of collapsed pole-impressed wall daga lying on these floors indicates cone on cylinder structures.
Although there are no visible doorways onto the floors, these can be established through the positions of the raised platforms that were built over approximately one-third of the floor surface (Figure 13.2). This raised platform, or apse, expresses a southern Bantu organizational principle based on a front/back, public/private, and secular/sacred opposition (Kuper 1982). The apse defines the ritually charged and private rear of the hut that is often associated with ancestors, and the entrance to the hut is positioned in opposition to it. The apse can be used to store male and female goods, such as ancestral spears or private storage vessels (for Nguni-speakers, see Raum 1973:157). The apse, and the sharp curb at its front, marks a boundary between private/sacred and public/secular.
Other features associated with the apse or immediately in front embellish this theme, particularly in relation to the transformation of food. Pot dimples occur on the apse immediately back from the curb edge, as well as below the curb to the front. As vessels that store, transfer, and transform the raw ingredients and carry the cooked, it is appropriate that pottery straddles this zone. All floors at Olifantspoort preserved upward of seven pots, which include storage jars and a high number of serving bowls. A significant characteristic of these vessels is their elaborate decoration, an issue that is taken up in more detail below.
Equally important is the position immediately in front of the apse of a sunken wooden mortar that occurs on the left side (as viewed from the front) and a fire bowl and associated cooking stones, usually three, in the center. On several floors this arrangement is associated with a slab “table” (Mason 1986:243). The mortars comprise a shallow 0.25 m diameter dish, which slopes down to the rim of a 0.15 m diameter wooden mortar, which is sunk a further 0.15 m below the level of the floor and wedged in place with stones. The mortars are made from Acacia sp., a hard wood that is entirely appropriate for grinding down sorghum and millet (Mandy Esterhuysen: personal communication). At Rooiberg, first-phase floors also preserve fire bowls and permanently fixed lower grindstones (Mason 1986: 299). The presence of carbonized sorghum and bone, including a number of cattle molars, with pots and a fire bowl emphasizes food preparation and consumption inside the hut.
The position of the mortar on the left and, on two floors, a bench on the right introduces another categorization of space that alludes specifically to a gendered division of labor. A return to Southern Bantu ethnography indicates that a left/right distinction often allocates position to women and men respectively (Kuper 1982). The repeated pattern on these first-phase floors of left-side mortars and a link with cereal processing and women is too strong to ignore. Furthermore, the recessed nature of the mortar and the kneeling or bent posture implicated in its use contrasts with the raised clay seats built on the right-hand side that draw attention to male space where a formally seated posture emphasizes a more vertical orientation. This juxtaposition of posture is similar to the example Lane draws on from Guyers’s study of the Beti in Cameroon (Chapter 10). On some floors, a distinct space either behind the seats or on the right side on floors without seats preserves a range of debris, including iron artifacts.
It is through comparison with the spatial and material patterns recovered from AD 1700 third-phase Tswana ward clusters that the material described for the first-phase floors becomes significant. Without preempting this comparison, I point out at this stage that first-phase floors preserve configurations of features and material that express concerns about classification and the mediation of categories across boundaries. The close association between cereal mortars, fire bowls, male space, and pottery with the sacred back is entirely appropriate. Among the Tswana (Comaroff 1985), the preparation and cooking of food is a final step in the transformation of rank nature that must be subject to ritual sanction lest the potentially polluting essence of “without” and “beyond” disturb the order of “within.” Women, who undertake most of the agricultural work, are the main medium through which pollution can be transferred. Furthermore, there is the well-documented notion that sexual pollution by women can contaminate food and endanger men. Among the Bemba, for example, menstruating women do not cook, and a Lele man can lose his virility if he eats food prepared by his wife who has neglected her postintercourse cleansing (Douglas 1995:152, 156).
These ethnographic examples are meant to illustrate some general notions about gendered labor, pollution, and the safe transformation of food that is facilitated through the close juxtaposition of boundaries on the first-phase floors. Whatever the exact meaning of these mediations, I strongly emphasize their compact nature within the compressed space of the hut interior. This last point is important in relation to the comparison with the third-phase floors given below, because, though boundaries as oppositions have been emphasized, their immediate spatial proximity on these first-phase floors suggests a close knit symmetry in the integration of social categories. There is little evidence for spatial differentiation outside the huts, but on the basis of the “house” as a microcosm of the larger system, I suggest that gendered loci interface directly, without any physical or architectural intervention. Consequently, as discussed below, there is a greater role for the mediation of boundaries through mobile symbols, particularly pottery.
The second phase in the sequence starts between the early to mid-17th century. The distinction is based on the appearance of archaeologically more resolved spatial markers in the form of low stone walls. These define homestead boundaries, central cattle byres, and stone wall rays at right angles to boundary walls, which must have physically separated some individual households within homesteads (Taylor 1979:11). Sites are, however, still small. It is during this phase that Sotho-speakers expand into the southern grasslands of the Free State south of the Vaal River (Maggs 1976: Fig. 1). At a crude level, this expansion and possible population increase may be attributed to the onset of a warmer and wetter pulse between AD 1500 and 1675 (Tyson and Lindesay 1992; Huffman 1996).
The question of why site boundaries and corrals were more permanently and visibly marked with stone at this time has not been adequately addressed. Environmental explanations focused on grassland occupation and a lack of wood are simplistic, because this stone wall phase also occurs in well-wooded regions. More likely is that stone walls elaborate a landscape of increasing permanence and spatial claim, and fissioning communities have to intensify the negotiation of place as spatial proximity shrinks between homesteads. Equally, the physicality of walls solidify and embellish boundaries that must correlate with changing emphases in social relationships. Boundary walls elaborate a distinction between inside/outside, culture/nature, and order/chaos; and the parallel internal definition of central byres and domestic margin must, in part, inscribe the processes by which the outside is socially harnessed and transformed.
It is in the third phase on the western Highveld, starting in the early part of the 18th century, that the spatial complexity that starts to develop in the second phase is fully realized. Major settlement expansion takes place, whereby some chiefdoms come to reside in a single place through the aggregation of homesteads into large ward clusters. In the early part of the 19th century European travelers visited some of these large Tswana towns. For example, Campbell (1822, 1:181) estimated between 10,000 and 12,000 people residing in one BaRolong town, and higher estimates were made for the BaHurutshe capital of Kaditshwene (Figure 13.1). In this phase it appears that homestead fusion, rather than fission, becomes the norm. Figure 13.3 illustrates site 20/71, which is a relatively small BaKwena third-phase settlement in the Olifant-spoort area of the western Magaliesberg. This completely excavated site (Mason 1986; see also Pistorius 1992) provides a comparison with the first phase. Needless to say, Tswana ethnography is directly relevant to this phase.
Figure 13.3. Site plan of a third-phase settlement and a composite plan of a third-phase household.
In contrast to first-phase sites, there is considerably more spatial detail in the third phase (Figure 13.3). Extensive stone walling demarcates the characteristically scalloped town perimeter, in which each individual scallop defines the back wall of a household’s back courtyard. Large central cattle byres, cattle tracks, and some walls between the central byres and household circle make up the rest of the stone walling.
Within an individual household, space is divided between a hut with front and back verandahs under the eaves of an extended cone. The hut door always faces inward toward the town center. There is a front courtyard, which may be separated from the center by either a stone wall or pole fence (see Hammond-Tooke 1993:10) and a back courtyard behind the hut. To one side of the hut is a small separate stone enclosure, which may have been roofed. In some cases there may be lateral walls extending from the hut to the points of the boundary scallop that separates the front and back courtyard.
Compared to the compact arrangement of mortar, fire bowl, apse, and seat on first-phase floors, the Olifantspoort 20/71 floors are mostly featureless. Most huts have no apse and no internal fire bowl; none have clay seats. The density of other material on the floors is low and there is little evidence for food preparation within the hut. These activities have been segmented and allocated to discrete spatial loci. Grain bin bases indicate that cereal is stored in the back courtyard behind the hut. This is the private space of the household, and in particular, the domain of women. Lower grindstones are found permanently wedged in place under the eaves of the back verandah. The small separate enclosure to the side of the hut was clearly used for cooking. These enclosures preserve elaborate mosaic hearths and firestones, slab “tables,” lower and upper grindstones and pots. The front verandah often has a hearth built at the junction of the verandah and hut wall. While some pottery is found on hut floors, many vessels are associated with the front and back verandahs and the cooking hut. Some bowls were set into plaster on the front verandah.
This summary of household space at a third-phase 18th century Tswana town indicates a greater degree of spatial detail compared to a first-phase households. The first-phase floors are multipurpose spaces for sleeping, for cereal storage, for the preparation and cooking of food, and for consuming this food. On the first-phase floors there was little physical separation of gendered activity. Following Kent’s (1990a) general principle of increased spatial segmentation with greater sociopolitical complexity, the third-phase households exhibit more functionally discrete spaces that are conceptually and physically separated. Food storage, preparation, and cooking has been fragmented and “bounded” in dedicated spaces, and the main hut seems to be restricted for sleeping. The private/back, secular/front distinction has been elaborated in the back courtyard and verandah and front courtyard and verandah. The back courtyard is the last space before one encounters the bush beyond the homestead and it is appropriately female space, where cereal is stored and ground down. The cooking hut is further to the front, and the consumption of food, especially if guests were visiting, would take place in the front courtyard. It is in this space where men and women, and the products of their labor, meet. Cereal, the product of female labor, moves from the homestead margin of the back courtyard to the front, in contrast to domesticated meat, the concern of men, which would move to the front courtyard from the homestead center.
The discussion suggests that men and women were more spatially proximate in the first phase, in contrast to the third phase, where gendered work becomes spatially fragmented into physically defined, activity-specific loci. At the scale of the whole homestead, the implication is that the domestic margin increasingly segments to accommodate female labor and, at the same time, spatially isolates this activity from men when they are in the household. This is an example of a general principle that is summarized by Guyer (1991:261), namely that “female and low-status workers tend to engage in higher-frequency tasks carried out within narrow spatial confines, a pattern that male and high-status workers try to avoid.” These spatial distinctions underpin a gendered division of labor that becomes more rigidly defined and separate between the first and third phases. I suggest that in the third phase the spatial configuration underpin hierarchy and control, in contrast to the symmetry in the first phase. Syntheses of Tswana oral records animate this archaeology in terms of a specific set of historical circumstance that highlight male anxieties over female labor and the security of the agricultural base. This historical background is only briefly described, and a full discussion can be found in Huffman (1986), Hall (1995), Manson (1995), and Parsons (1995).
The third-phase settlement attributes of large aggregations and a tendency for settlements to be located on broken and higher ground, indicates that defense was one determining factor (Huffman 1986; Hall 1995). Oral records indicate that the 18th and early 19th centuries were a period of increasing military competition between Tswana chiefdoms that eventually led to extensive destruction from the 1820s (see Hamilton 1995 for a wide range of views on this Difaqane/Mfecane period). Several causal factors contributed to these Tswana wars in which the Hurutshe senior chiefdom increasingly lost power to the BaKwena and Pedi in the east, the Tshidi-Rolong to the west, and the Ngwaketse to the northwest (Manson 1995:354). Some factors are interrelated and others independently coincident.
One cause of aggression focused on land and grazing. Higher rainfall through the 18th century resulted in consistently good food supplies and, consequently, demographic increase and pressure. With limited space to expand, competition over critical agropastoral resources steadily rose. The introduction of maize from the east coast at some time during the 18th century also added to agricultural production, which had significant demographic repercussions over the whole summer rainfall area where maize was viable (Marks 1967). Cattle raiding figured prominently in these wars and, as suggested by Manson (1995) and Parsons (1995), underpinned the concern of men to create wealth in cattle in order to accumulate wives through bridewealth and so maintain or increase a pool of female labor. Oral traditions suggest that at times women were forcefully obtained as an additional aspect of cattle raiding (Manson 1995). Lastly, there was also intense competition over trade in skins, fur, feathers, metal, and ivory, and access to the routes along which these goods were passed. In the late 18th century this was exacerbated by the growing market demands from the encroaching colonial frontiers from the east coast and the Cape to the south (Manson 1995).
Overall, these factors intensified competitive entrepreneurship between rival Tswana chiefs. Fusion into large aggregated towns was a necessary outcome of historical circumstance, while greater political centralization also controlled the natural tendency for individual homesteads to fission. Although indications are that consistently good agricultural surpluses were produced through the 18th century, droughts at the end of this period and early in the 19th century (Hall 1976) undermined agriculture and contributed further to military tension, aggregation, and, above all, the need to regulate and control agriculture (Huffman 1996:59). It is within this historical context that the spatial segmentation within the third-phase settlements and individual households takes on meaning. The spatial elaboration of the domestic domain suggests more complexity in the management, control, and organization of female labor that relates to the larger-scale political and economic structures that were dependent on it.
That this structural complexity related to male control and male anxiety over the agricultural base can be elaborated through other aspects of Tswana social organization evident in early traveler records and Tswana ethnography. For example, men controlled the ritual that sanctioned female labor and the agricultural cycle. Field allocation, land preparation, planting, harvest, and first fruits were all stages in the cycle that could only proceed through the ritual sanction of the chief and household heads (Sansom 1974:147). Above all, the chiefs’ ultimate control lay in rain-making (Schapera 1971). Furthermore, seasonal taboos ritually outlawed women from digging pot clay, making pots, or cutting thatch during the agricultural cycle (Schapera 1971:93). This seasonal restriction on potting underpins a fear of drought generated by the pollution of heat and its effect on agricultural fertility through the fired transformation of clay. While the ethnography does indicate that women had some control over agricultural produce, this power could not be formally accrued because of the relatively immediate consumption of food, in contrast to the male accumulation of cattle (Kinsman 1983). Female agricultural production, therefore, was under the ultimate control of men. In the circumstances of the late 18th century, ideologies of economic value were ritually underwritten and “powerfully legitimized structures of domination and inequality” (Comaroff 1985:67).
The segmentation of domestic space through physical and architectural boundaries increasingly isolated women during the third phase. At the center of this domain lay the hut, an explicit metaphor for the female body that symbolizes the centrality of biological and agricultural fertility for social continuity. This metaphor is explicitly rendered by the Tswana phrase go tsena mo tlung (“enter the house”), which refers to sexual relations within marriage (Comaroff 1985:56) (Lane, Chapter 10). However, the division of labor over the construction of the hut and the symbolic elaboration around the doorway allude to unity. Men work with wood and in the vertical by constructing the frame of the hut (Vogel 1983), and they must have also constructed the wooden sliding doors at Olifantspoort. The symbolic inference is that men control fertility through their construction of the frame, and particularly through their construction of the door. However, these doors, without exception, slide open to the left, female side, in and along a recessed clay groove (for details see Mason 1986:395). This symbolic combination of the recessed and the vertical is elaborated on some of the door curbs at Olifantspoort through the insertion of rows of acacia thorns set into the clay. Trees with thorns invariably connote maleness (Schapera 1971:92). Furthermore, women are integral to completing the process of construction by closing off the hut by plastering the walls and floors and thatching the roof. In so doing, women continue their connection with the earth.
So far the articulation of the third-phase archaeology, its accompanying historical context, and its ethnography has emphasized male control. Still further detail from Olifantspoort and other similar sites indicates that male labor also shifted and escalated during the third phase. At Olifantspoort hide and metal work are prominent toward the end of the occupation (Mason 1986), and at Mabyanamatshwaana (Pistorius 1995:57), the “large-scale introduction of iron forging” also intensified at approximately the same time. This intensification is entirely consistent with the historical context of greater trade in these commodities. These male crafts were also spatially elaborated in discrete areas of the settlement. Returning to Figure 13.3, the enclosures marked 2 and 3 were specifically for hide and metal forging, respectively (Mason 1986). Enclosure 2 was littered with large kidney-shaped hornfels flakes, with severe attrition on the working surfaces. Microwear analysis indicates a hide-working function (Binneman 1987). Other aggregated third-phase sites in the western Highveld preserve abundant evidence for similar activities (Mason 1969). Hide working, wood working, and metal working were male activities that equated with “refined social accomplishment” (Comaroff 1985:70). This contrasts with the low-value, mundane work of women, constrained as it was to the domestic periphery. However, these activities were spatially linked to the chief (Figure 13.3, enclosure 1) and the court. This close connection to the chief indicates control over prestige commodities and implies that men also occupied positions of power and subordination within their own ranks.
The parallel intensification and segmentation of both female and male activities in the third phase raises the question of time allocation to tasks. It is possible that toward the end of the 18th century women had to invest considerably more time on agriculture than was the case during the first-phase settlements. Even if maize was viable on the western Highveld, and even if it did boost the amount of cereal harvested, the logistics of agriculture in the context of large, aggregated towns may have been considerably more difficult for women at this time. High-density aggregated living must have reduced the availability of suitable agricultural land that was conveniently close to towns and, coupled with the environmental imperative for dispersed fields, required considerable travel time to those fields, or even permanent residence there when weeding and bird scaring became critical as the harvest approached (Schapera 1971; Comaroff 1985). Campbell (1822, I:181) recorded that the fields of one BaRolong settlement were up to 20 miles (32 km) distant from the town. There is no question that the agricultural cycle for women was long and strenuous, and that this was exacerbated through other work routines (Kinsman 1983; Eldredge 1993:109).
Another factor to consider is that men may have increasingly withdrawn their contribution to agricultural labor as their own priorities shifted to a greater preoccupation with cattle, raiding, defense, and intensifying the acquisition of prestige commodities for trade. Furthermore, as discussed by Guyer for the southern Cameroon (1991:260–261), farming traditional cereals, such as sorghum and millet, is deeply sacred and ritualized and their production is shared between men and women in a sequential set of tasks, and is therefore not the sole prerogative of women. It is with the introduction of New World staples, such as maize and cassava, that the ethnographic picture of almost exclusive female farming assumes prominence in Africa, and though an important food, there is a negative attitude toward maize. Richards noted that preparing land for maize “was considered hard and unromantic work by the Bemba, quite unlike millet cultivation” (Richards 1939:304 from Guyer 1991). The rejection of maize beer in favor of millet beer for ritual occasions also underscores that maize is cognitively marginal compared to the traditional cereals. If maize, however, was as important to some Southern Bantu toward the end of the 18th century as has been suggested (Marks 1967; Huffman 1996), then studies such as Guyer’s alert us that its introduction in the third phase from about 1750 must have had implications for the organization of labor.
It seems overly simplistic, however, that the secular status of maize is only a function of its relatively shallow time depth within southern Bantu agriculture. If it was important as a staple food, why is that not reflected in its integration into local cosmologies? There are many examples that show how valued innovations are rapidly embedded. Although speculative, it is possible that men, as the pivots of ritual sanction, actively marginalized maize in ritual because it was so important. The inverse proposition that actual importance is not reflected in ritual refocuses attention on the male control of labor and on male anxieties about the real power of women, as well as the critical importance of agriculture in the aggregated and aggressively competitive context of the 18th century.
The arguments so far have been based on a consideration of spatial organization and change between the first and third phases. I have inferred from the increase in the compartmentalization of space a concomitant segmentation and control of female labor that possibly reflects the development of asymmetrical power relations with men. I now add to and elaborate these arguments through a consideration of the ceramic sequence.
It has been noted that pottery from the first and second phases of the Sotho sequence is elaborately and richly decorated (Mason 1986:723; Hanisch 1979; Hall 1981, 1985). In some assemblages, over 50 percent of the vessels are decorated with combinations of textured designs that cover large areas. Decoration is combined with a rich use of red ochre and black graphite coloring. In comparison, pottery from the aggregated third phase is decoratively bland. Motifs are simple and the majority of decorated vessels have only a single band of nicking on the rim. This general trend is widespread and is not specific to one cultural group.
To formally demonstrate this difference in decorative intensity between the AD 1400 to 1700 first and second phases, and the post—AD 1700 third phase, I reanalyzed all reasonably sized published collections. This analysis follows the multivariate system advocated by Huffman (1980), which combines profile with decorative position to produce multivariate classes. Class definitions are given in Table 13.1, their distribution through the sequence are given in Table 13.2, and Figures 13.4 and 13.5 illustrate pottery from each phase.
Fifteen ceramic classes account for all the ceramic stylistic variability in the sequence, and it is usual for first- and second-phase assemblages to have over 50 percent of these classes. Of note in these assemblages is the relatively high number of bowl classes, and the fact that from sites with good samples, well over 50 percent of the bowls are decorated (Mason 1986). Third-phase assemblages have much less stylistic complexity, and a maximum of five classes occurred in only one assemblage. Bowls occur in low frequencies and decoration is minimal. It should be noted that the placement of ochre and graphite was not used in the definition of classes, but is nevertheless a highly visible and extensive attribute of first- and second-phase ceramics, though infrequent on third-phase pots.
This pattern of declining decorative intensity, especially in the post—AD 1700 aggregated phase is clear. There is a simple correlation between first- and second-phase small and dispersed homesteads and high decoration, while low decorative intensity correlates with large third-phase aggregated ward clusters. More specifically, in the first phase, large numbers of pots were recovered from within the hut, while vessels were more scattered around third-phase households. Interpretation of this ceramic trend focuses, first, on the symbolic role of pottery and decoration in mediating boundaries between men and women, and, second, on a brief consideration of changes in the organization of ceramic production.
The starting point for a symbolic interpretation comes from a principle developed from ethnoarchaeological work on African pottery. For the Azande, Braithwaite (1982) has shown that pottery used in spaces where men and women interact tends to be decorated, while vessels that are used in private contexts are not. She suggests that decoration “functions” as a low-key ritual and symbolic marker in contexts where breaches in the social order do, or potentially can, occur. The higher intensity of decoration in areas where male and female encounters take place expresses a concern about the overlap of boundaries, and consequently, those boundaries have to be defined. In general, elaborate decoration correlates with high vessel visibility because it communicates information about social categories, roles, and status.
Table 13.1. Definitions for the fifteen Moloko ceramic classes.
Class 1 |
recurved jar decorated with a single band immediately below the rim |
Class 2 |
recurved jar decorated in one zone on the neck |
Class 3 |
recurved jar decorated with a band immediately below the rim and a band low on the neck |
Class 4 |
recurved jar decorated with a band immediately below the rim, a band on the neck, and a band of decoration extending below the neck onto the body |
Class 5 |
recurved jar with multiple bands extending to the base of the neck, a band of arcades on the shoulder, extending onto the body, and isolated triangles low on the body |
Class 6 |
open bowl with a single band of decoration immediately below the rim |
Class 7 |
open bowl with a band of decoration immediately below the rim and a band on top of the rim |
Class 8 |
open bowl with decoration in one zone on the body |
Class 9 |
open bowl with a band of decoration immediately below the rim and a band on the body |
Class 10 |
open bowl with decoration on top of the rim, a band of decoration immediately below the rim, and a band on the body |
Class 11 |
open bowl with multiple bands on the body with a band of chevrons or arcades extending very low on the body |
Class 12 |
open bowl with a wide band of decoration below the rim and which is decorated in the same zone internally, or has alternating bands of graphite and ochre all the way to the bottom |
Class 13 |
constricted pot with a wide band of decoration below the rim |
Class 14 |
constricted pot with a band of decoration immediately below the rim, followed by a wide band of decoration down onto the upper body |
Class 15 |
constricted pot with a single band of decoration immediately below the rim |
The ethnography underscores the suitability of pottery to “patrol” boundaries because of the fundamental notion that pots are conceptually equivalent to people (David et al. 1988). It is through this link that pots can be actively manipulated in a wide range of social and ritual contexts in order to communicate meaning about “everyday social life,” where “differences of form and decoration help create categories of time and persons” (Barley 1994:116; and see Herbert 1993:200 ff. for a review of African potting). Among the eastern-Tswana Pedi, this homology is clear. Pottery production is analogous to procreation because successful manufacture requires that “fertile” clay (women) and strong water (semen/male) need only lie together for one night to be fruitful (Krause 1985:68). Further afield, the Thonga liken a child to a cooling pot after firing, and both must be carefully weaned to maize and water (Junod 1912:100). Shona pottery, “the women’s weapons,” are symbols of female reproductive organs and can be manipulated to communicate a range of intent, including sexual rejection of her husband through the inversion of her pots (Aschwanden 1982:199).
Table 13.2. Distribution of ceramic classes through the Moloko.
Figure 13.4. Pottery classes from first- (A) and second- (B) phase sites. Numbers refer to classes, which are defined in Table 13.1.
Furthermore, the motifs used on Tswana pottery are described in terms of female clothing and allude to female fertility (Figures 13.4 and 13.5). Triangles and arcades are described as the tips of the notched rear skirt (ntepa) of an initiated women (Vogel 1983), while the chevron is likened to the vulva (Evers and Huffman 1988:740). Other terms for these motifs include the swallow (peolane), the moon (kgwedi), and the snail (kgopa), and through their association with rain and agriculture, also refer to women (Krause 1985). The moon and snail are depicted as full circles, and often decorate the foreskirt (thetho) of initiated or married women, which are made and decorated by men (Vogel 1983). These same motifs adorn a wide range of other artifacts, including beadwork, wooden porridge and meat dishes, drums, and mural art. Pottery is also decorated with the red-white-black color triad, which is an explicit symbol in rites of transformation through potentially harmful periods of liminality. White, therefore, purifies, cools, and cleanses, while red intensifies and consolidates, and black strengthens and protects (Hammond-Tooke 1981:137).
Figure 13.5. Pottery classes from third-phase sites. Numbers refer to classes, which are defined in Table 13.1.
In general, decoration “discourses” on life states, roles, and status. Through its public display and manipulation, decorated pottery draws attention to the sexual power and ambivalence of women, but at the same time it also marks off that ambivalence. In the Tswana world the sexual ambiguity of women and, as Comaroff states, their “lack of closure” (1985:71) is an explicit concern for men that often manifests itself as the undesirable state of heat (go fisa), or pollution, that disrupts social order. While women are the prime purveyors of polluting heat, it is generally not a condition that is consciously encouraged or passed on malevolently. It is something that inheres frequently in women because of the perpetual cycle of liminality they experience. Consequently, women with “hot blood” could pollute men through sexual intercourse. Similarly, women who are pregnant, menstruating, or in childbirth are “hot”. Women who have suffered a miscarriage can be “hot” for a year after the event (Vogel 1983:395).
As already noted, the potential of women to pollute creates obvious spatial constraints on them. An obvious example is their general confinement to the domestic periphery, and the wild, liminal area beyond the boundary of the homestead, but away from the “cool” judicial domain of the male court and the associated central cattle byres. There are also taboos against their spatial proximity to the male world of iron smelting (see Herbert 1993). Many other cases could be cited. Temporal constraints also affect women. The taboo against making pots during summer has already been mentioned. As a consequence, women “required confinement, they could not extend themselves in physical or social space” (Comaroff 1985:71).
This background provides rich interpretive potential for the disparity between the decorated first- and second-phase pottery and the mundane third-phase pottery. In the relative spatial intimacy of first-phase settlements, symbolic boundaries assume more importance than in third-phase settlements, where physical boundaries controlled, directed, and regulated encounters between men and women. Vessel visibility is high in the first phase, but with spatial segmentation in the third phase, vessel visibility declines. A further impetus to this trend is the possibility, as noted in the ethnography of the Pedi (Mönnig 1967) and the Venda (Stayt 1968), that men eat alone.
A closer examination of changes in both the decorative intensity and the frequency of bowls through the sequence suggests an additional factor in vessel visibility. As noted (Tables 13.1 and 13.2), out of the 15 classes that account for all stylistic variability, 6 are open bowls. In first- and second-phase assemblages, there are at least 5 bowl classes represented, in contrast to third-phase assemblages, where only 1 class occurs. Furthermore, in some first-phase collections, bowls account for over 50 percent of total vessel counts, while in the third phase, this figure drops below 10 percent (Hanisch 1979; Hall 1981; Mason 1986). Historically, bowls were used for serving food, particularly porridge (Quin 1959), and there is no reason why this function was different in the past. In terms of the symbolic mediation discussed above, the transformation of cereal through cooking and its transferral across boundaries between women and men, suggest why the vessel of transferral carries a high symbolic load. This is certainly the case in the first phase, where the spatial evidence indicates that cooking, serving, and eating took place within one confined space. The sharp decline in bowl frequency and the near absence of decoration in the third-phase assemblages implies the opposite. Spatial compartmentalization indicates that the dynamics of food transfer changed. The isolation of food preparation and cooking in discrete spaces is one factor in this. Another is that men stopped using clay bowls to transfer food because they began to manufacture their own wooden bowls and meat trays. As noted above, Tswana men work wood and, specifically, carve wooden porridge bowls and meat dishes (Vogel 1983). The implication is that in the third phase, men were making the wooden porridge bowls and meat trays on which they were served. The use by men of their own personal porridge bowls is also noted in the ethnography of the Pedi (Mönnig 1967). The low frequency of clay bowls in the third phase possibly reflects their substitution by archaeologically less visible wooden bowls. Men, consequently, asserted some control over the transfer of food. Furthermore, ethnographic examples of wooden porridge bowls and meat trays show that men decorated them with the same symbol set found on pottery. Men manipulated material culture that symbolically categorized the fertility of women, and in so doing, actively defined boundaries between them.
In general, the manipulation of space and the possible male intervention in the nature of food transfer constitute strategies developed to order social interaction in the high-density living conditions of large third-phase settlements. A brief digression into the ethnography of Tswana marriage patterns and witchcraft beliefs provides examples of other social strategies that embellish this point.
Tswana marriage is unique among southern Bantu-speakers, with its emphasis on cross-cousin partners. As pointed out by many ethnographers, endogamous marriage (Sansom 1972; Preston-Whyte 1974; Hammond-Tooke 1981; Kuper 1982) creates a high degree of introversion in Tswana kinship. With little ideological constraint over marriage partners, kinship is confounded and creates overlaps in kinship categories so that agnates may also be affines. This kin ambiguity, particularly among women, further emphasizes a concern for boundary definition. In contrast to the strongly exogamous marriage rules among the dispersed homesteads of Ngunispeakers, where there is no ambiguity about the category “women”, Tswana wives are not necessarily outsiders, and there is less distinction between “our women” and “their women” (Sansom 1972:205; Hammond-Tooke 1981). Consequently, it is possible that the Tswana emphasis on endogamy developed as aggregation proceeded, and therefore is a relatively recent development. Furthermore, the preference in elite marriage is for a cross-cousin on the agnatic side. Again, this preference may have been emphasized more as aggregation progressed as a strategy by elite men to recirculate cattle within their own line, thereby centralizing the medium of political power.
Turning to witchcraft, Sansom’s (1972) discussion of the Pedi suggests that public denunciation seldom occurs. In this regard, there is “a congruence between the lack of ideological limitation on the identity of either a marriage partner or a witch” (Sansom 1972:205). A category of “witch” cannot be defined through clear-cut kin categories or from gender, because these categories are ambiguous. Everyone potentially could be a witch, but to publicly denounce would be disruptive for aggregated town living. In contrast with the nature of witchcraft accusations in small dispersed settlements, the notion of malevolent and personally directed affliction is not emphasized among the Tswana, and female pollution, for example, is not maliciously or consciously imparted. It is impersonal, and its ritual purification is communal. I have mentioned these aspects of social organization because, if contextually specific to the third phase, they add circumstantial significance to the interpretations made above.
One last factor that requires brief mention in relation to the ceramic trend is possible changes to the organization of production. The combination of archaeological and historical evidence points to male intensification of craft production during the third phase that is tied to the centralization of chiefly power, premised on an increasingly specialized economy through trading opportunities. Although a parallel shift in pottery production would be inconsistent with the argument that women were steadily constrained physically and socially, it is possible that women collaborated among themselves in changing production. Discussion of this is circumstantial, however, and requires archaeological evidence on the actual locales of ceramic production and studies of regional trace elements (see Hall and Grant 1995). It is, however, worth considering relationships between changes in the time allocation to tasks such as potting, and the influence this can have on the nature of production and the character of the pottery (Costin and Hagstrum 1995 give a general summary of labor investment and other factors).
Another way of describing the third-phase assemblages is that they are relatively standardized in terms of the repetitive nature of motifs and their limited simplicity (Figures 13.4 and 13.5). Standardization relates to more efficient labor and time costs. As mentioned, the trend toward simpler and blander pottery on the western Highveld crosscuts cultural boundaries. If women generally had to reapportion time allocation to tasks in the third phase, or use time more efficiently in fulfilling other tasks, then simplification in ceramic production may have contributed one saving. It is also possible that the factor of time investment could have shifted the scale of production from the individual or homestead level to more specialized workgroups that supplied pottery over a much larger area or region. In contrast to first-phase pottery, it is logical to infer less labor investment for the third-phase vessels. There are also other hidden costs in ceramic production, such as the collection and preparation of ochre and graphite (Eldredge 1993:107) for decoration, which is such a prominent feature of first-phase assemblages but is conspicuously absent on third-phase vessels. It is also of interest that ceramic changes at Great Zimbabwe perhaps provide a comparison. Evers and Huffman (1988:740) note that pottery production appears to have become narrowly standardized and also stylistically limited in terms of shape and decoration. While each case must be individually investigated, correlations between ceramic standardization and changes in social organization suggest that there may be some common process at work.
A further aspect of the Tswana social system that could also have impacted ceramic production is the flexibility and fluidity of personnel in the ward system of aggregated towns. Although a village or ward may originate as an agnatic group, “unrelated individuals or groups may be added to the kgoro, making it a heterogeneous group surrounding the homogeneous core” (Mönnig 1978:219). This receptivity to outsiders must also have developed during the 18th century as a strategy that encouraged dislocated people to centralize with other chiefs. Fluidity in personnel of different cultural backgrounds could have loosened the need for ethnic expression through material culture (Hodder 1979), and standardization and simplification of style could have been one other strategy for “getting along” in aggregated towns. These ideas about the relationship between women, their labor time, and the nature of ceramic production are speculative, but indicate that a closer examination of this dynamic is warranted.
In this paper I have tried to rework old data with new questions and other perspectives on what this data could mean in terms of social and, specifically, gender correlates. I have “fragmented” the Tswana sequence into a series of historically discrete parts, each of which has a contextually specific set of implications for the nature of gender interactions. The outcome has been an interpretation that recognizes two periods with different emphases in the way gender relations were managed.
The first is characterized by small dispersed homesteads, in which gender interactions were separated, bounded, and defined through symbolic mediation. Although spatial evidence for this early phase of the sequence is poor, I have inferred that the spatial or architectural segmentation of gender and gendered activity was relatively low. I have also suggested that economic and social power was more symmetrical. In sharp contrast, the evidence from the third phase, from AD 1700, suggests changes in the form of social interactions. Spatial segmentation is elaborated through physical boundaries and, as a consequence, there is less uncertainty and more predictability about the nature of encounters. It is suggested that the value of pottery as a medium that facilitates social distinction and interaction declines as spatial separation increases. The circumscription and constraint of women within the domestic zone suggests that there was a more rigid division of labor, and more structural control over productive and reproductive processes that lay at the core of social security. In contrast to the first phase, third-phase gender relations may be described more in terms of hierarchy and asymmetry, where there was more direct male control.
It is clear that the interpretations are more detailed for the third phase than for the first. The variable quality of the data contributes to this, but another critical influence has been a growing wariness of extrapolating Tswana social and gender relations beyond the historical circumstances of its generation (see Lane 1996). Gender matters aside, working with this sequence has drawn attention to the fact that the details of Tswana ethnography are analogically appropriate for the recent period, from the early part of the 18th century. In this regard, I have drawn attention to aspects of Tswana social form that are most probably responses to the economic and political circumstances that prevailed and intensified at that time. For the first phase there are general structural features that are common within Southern Bantu ethnography, but, as outlined at the start, imputing detailed meaning through the ethnography would be misleading. The way these principles are contextually managed can be very different. Indeed, the privilege of time means that the archaeology can provide an independent commentary on the ethnography by contributing toward an understanding of its generative context, and its appropriate temporal use.
At the start of this paper I emphasized the house as the unit of analysis, one that provided a setting that actively “tuned” people to a social logic. It needs to be reemphasized that this logic can be many-sided, depending on whose minds and bodies are being tuned. For men, the domain of women may be seen as dangerous, and it is, therefore, constrained, bounded, and marginalized because it can upset the social order at the center of their power. The intention of this paper, however, has not been to portray active men and passive women. Boundaries obviously acknowledge that the power to impose structure recognizes equally the potential of abutting powers to alter that structure. For women, the house and its symbolic boundaries flaunt productive and reproductive power as reminders of their fundamental importance for social continuity. The house and its symbolic boundaries, therefore, mediate many sides of social power, and the tension between them always holds the potential for change.
1. Sotho is a term for one of the four main languages spoken in South Africa. Tswana is a variation of Sotho. Although these terms have their origins in the colonial categorization of indigenous society, I still use them as convenient labels for the period from AD 1400, when the first “Sotho”-speakers can be archaeologically identified, and after which there is undoubted continuity.
2. In this paper “house” refers to all the component parts of a domestic unit, comprising courtyards, storage areas, cooking areas, sleeping areas, and so on. “Hut” is used simply as a convenient label to distinguish the roofed area of a “house.”
