The rhythms of daily life in rural Banda-Ahenkro—seat of the Banda paramount chieftaincy—are clearly orchestrated along lines of gender and age. Women rise early to kindle wood fires and cook the morning porridge; young girls sway beneath head-pans brimming with water on their way home from the boreholes; children move between home and the diesel-powered corn mills carrying loads of corn and corn flour; men move from compound to compound greeting neighbors and relatives, perhaps stopping by the paramount chief’s compound to greet the chief and discuss community affairs. As sunshine floods the village, men head to farm, to clear, to plant or harvest, depending on the season. Women too may begin their day by traveling to farm, spending the morning weeding, cultivating, and collecting firewood. They return to the village burdened with head-loads of firewood or foodstuffs in time to begin preparing the evening meal. Women who remain in the village spend their day processing foodstuffs: cleaning calabash seeds, winnowing cowpeas, pounding corn if they are unable to afford the services of corn mill, or sorting and tying tobacco leaves in preparation for drying. Many children over the age of six spend their mornings in local primary schools, though only a handful of the boys continue on to secondary schools. Girls over the age of about nine are busy with household chores and tending small children. Thus, men, women, and children appear to play scripted roles in the daily drama of household production and reproduction.
Archaeological analyses of gender and production often assume an isomorphism between such ethnographic scripts and archaeological patterns of household production and reproduction (see Parkington, Chapter 2, and Gifford-Gonzalez, Chapter 7). Yet historical anthropological studies demonstrate that productive relations may have changed considerably as societies were incorporated into a market economy (Etienne 1977; Kriger 1993: note 6). Roberts’s (1984) study of 19th-century textile production among the Maraka of Mali provides an example. Drawing on archival and oral historical sources, Roberts made a case for how increased market demand for indigo cloth altered gender-property relations among Maraka men and women, despite the fact that the technology and allocation of men’s and women’s tasks remained the same. Before the 19th century, production of indigo cloth took place within the household, drawing on the labor and property of husbands and wives. Husbands cultivated cotton, while wives grew indigo. Women cleaned, carded, and spun raw cotton, while weaving was a male domain. Cloth was consumed primarily within the household. As demand for dyed cloth increased among Islamic elites, indigo acquired a market value. Husbands began to use slaves to cultivate indigo to be sold at market, and deployed slaves in an expanded weaving enterprise. Despite their intensified involvement in spinning and dyeing, wives no longer controlled the source of dye, nor the finished product. The social relations of production were transformed again when Maraka slaves left their masters early in the colonial period and established themselves as independent craftspeople, undermining the control of elite males over production of indigo cloth. The textile industry experienced further changes as cheap (and less durable) imported cloth undermined, though did not altogether stop, local cloth production in this area (Roberts 1984:247; see also Etienne 1977; Kriger 1993).
Roberts’s case study has important implications for how archaeologists approach the study of gender in the past. When we emphasize household divisions in terms of gender-specific tasks, attention is diverted from how the labor process shapes the reciprocity of gendered tasks and property relations that, Roberts (1984:248) argued, changed with increased involvement in a market economy. The implication for those of us interested in modeling gender in the more distant past is that we must be attentive to how gender relations in recent societies—the source of our analogical models—were shaped by involvement in a market economy (Costin 1996:117). Inattention to this may lead us to assume a false isomorphism between contemporary and past gender relations. We can overcome this problem in two ways: first, rather than using analogical models as illustrative devices, we can adopt instead a comparative approach that is attentive to how archaeological evidence may diverge from ethnographic expectations (Stahl 1993; see Hall, Chapter 13); and second, we can study the effects of a market economy on gendered production and consumption, which requires a focus on historically recent societies. We know that the effects of this involvement were uneven. While local handicrafts disappeared in some areas, they persisted elsewhere, but often in altered form.1 The challenge is to examine how production and consumption intersect with gender from a processual, historical perspective. Not only will this help us to sort out patterns that might be unique to societies involved in a market economy, it can also help us understand gender relations as dynamic processes.
Our case study works toward the second of these goals and is set in the rural Banda area of west central Ghana, today more than 70 km distant from the nearest major market centers (Wenchi, Techiman, Bole, and Bonduku; Figure 11.1). In this paper we examine how an ethnographic model of men’s and women’s roles in the contemporary market economy compares to past patterns of gendered production. Using a direct historical approach (Stahl 1994a), we draw on archival, oral historical, and archaeological insights to examine how craft production changed in response to historically documented shifts in the regional and subcontinental political economy from the late 18th century to the present. This period witnessed dramatic shifts in subcontinental trade relations as the Atlantic trade eclipsed earlier trans-Saharan networks. Political dislocation and warfare disrupted the regional economy, especially at the close of the 19th century. Drawing on oral historical sources, we explore how gendered patterns of production may have been impacted by warfare and other political economic factors. Archaeological data are used to investigate changes in craft production, especially potting and weaving. Final sections explore the implications of assuming isomorphism in gendered patterns of production.
Figure 11. 1. Towns and villages mentioned in the text.
The Banda area lies immediately south of the Black Volta River and is centered on a range of razor-backed hills that rise dramatically out of the surrounding, low-rolling landscape. These hills, composed of resistant metamorphic rock, trend northeast-southwest and present a barrier to east-west movement for a distance of about 50 km south of the Black Volta. The area is connected to the network of paved roads by an untarred track that dead-ends at the Black Volta River. To the south it intersects with a gravel road that connects the market centers of Wenchi and Sampa (Figure 11.1). Occasional trucks travel to the area to transport foodstuffs to market or bring kerosene for sale. Locally produced foodstuffs (tomatoes, onions, groundnuts, and fish) and crafts (such as pots) dominate the local markets, and people must travel to market centers like Wenchi, 75 km distant, or Techiman, another 35 kilometers beyond Wenchi, to purchase cloth, clothing, and household equipment.
Despite its remote character today, Banda is thoroughly integrated into a market economy. The agricultural sector in villages east of the hills is geared to production beyond subsistence needs.2 The primary cash crops are yams, calabash (gourds), and more recently (since the mid-1980s) tobacco. Cassava (manioc) has become increasingly important over the last 20 years as both a staple subsistence item and a cash crop. Surplus beans, groundnuts, and condiments may be sent to market centers if planted in large quantities; however, they are more often sold in small portions at the weekly village markets. Cash is a prerequisite for household reproduction. Money is required for a variety of needs—to purchase seed stocks when the previous year’s stores run short; to hire seasonal agricultural labor (for example, to prepare yam mounds); to purchase commodities like cooking oil, sugar, Maggi cubes or soap, as well as condiments when household gardens run short; to purchase fish or meat; to buy kerosene or perhaps batteries; to have corn ground at the mill; to clothe family members; to pay school fees; to purchase the tools necessary for household reproduction (such as cook-ware and agricultural tools); and to buy materials, and perhaps to hire skilled labor, to build a house. Petty transactions in the local market are conducted primarily in cash, though pottery is sometimes exchanged through barter. Social payments too are monetized, with funeral obligations met with cash (see also Arhin 1995:98). In sum, as is the case throughout much of Africa today, it is “. . . impossible to be a social adult without the capacity to mobilize sums of money that are quite substantial relative to people’s incomes” (Guyer 1995:24). Some of these needs can be met without cash, as they are in difficult times (Dei 1988; Posnansky 1980, 1984; Stahl 1994b). Houses can be built using locally available materials (earthen walls and thatch roofs); however, people increasingly prefer “sandcrete” blocks, aluminum roofing sheets, and carpentered windows and doors. Corn can be pounded in wooden mortars rather than ground at the mill. And local oils (such as shea nut butter) can be substituted for bottled, manufactured varieties. Yet these manufactured goods, once luxuries, are today perceived as necessities (Arhin 1976/77:459–460).
As elsewhere in West Africa, money provides the sole means for acquiring cloth in Banda today. Cloth provides an important means by which individuals create and maintain status. Today, the prestige cloths in Banda are strip-woven kente cloths and manufactured wax prints. Kente, produced primarily by Akan weavers to the south using imported thread, is worn almost exclusively on ceremonial occasions. Wax prints are today manufactured in Ghana and Côte d’Ivoire; however, the most durable and expensive wax prints are imported (“Dutch” wax prints). They have virtually replaced locally made cloth (see Maier 1995:95).
Travelers in the 18th and 19th centuries moving throughout West Africa repeatedly commented on the quality of cloth produced throughout the subcontinent. Local cloth was woven on narrow looms. The strips were sewn together to create large cloths that might be tailored into garments (see Idiens 1980). Indigo-dyed cloth became an especially valued exchange commodity (Roberts 1984; and see Johnson [1980] on cloth strip currencies). Yet most cloth was apparently produced for household consumption (Etienne 1977). Although cloth played an important role in the European trade from an early period, European cloth was less durable than local cloth (Steiner 1985). As late as 1889, Freeman noted the lack of European cloth in the large Bonduku market. Although he observed “a few European articles for sale, fish-hooks and small fragments of looking-glass being the goods most in request . . . I did not see any European cotton goods exposed for sale, which is not a matter for surprise seeing that the native cotton cloths are greatly superior to those made in Europe, and are sold at very moderate prices” (Freeman 1967 [1898]:179). Contrary to the expectation of prestige goods models that stress the importance of exotic origins, locally made cloth was, and in many cases continues to be, the prestige dress of choice (Crossland 1989:52; Etienne 1977:56; Okeke 1980:113; Picton 1980:82, 84; Roberts 1984). Today, however, locally produced cloth is seen as more expensive than manufactured textiles.
Cloth production across a broad sweep of sub-Saharan West Africa drew on the reciprocal labor of men and women—often, but not exclusively, husbands and wives (e.g., Etienne 1977; Roberts 1984). Although responsibility for cultivation varied, women typically carded and spun cotton, while weaving was most often the province of men (see Kriger 1993; Okeke 1980:113; Picton 1980). Both Etienne (1977) and Roberts (1984) describe the role that cloth production played in creating mutual obligations between husbands and wives, signified by the importance of cloth in gift-giving. In Banda, a prospective husband was expected to give his bride a special cloth (nyankacha) composed of 12 strips of white woven cloth. The bridegroom and his friends gathered on the first day of the marriage rites to sew together strips of cloth provided by the bridegroom’s mother’s sisters. With the decline of local cloth production over the last 30 to 40 years, husbands are now expected to present their brides with gifts of manufactured cloth (i.e., wax prints). During the severe economic crisis of the early 1980s, young men in Banda were forced to postpone marriage because they were unable to afford the high cost of imported cloth, most of which was smuggled into the country.3 Cloth was also a means by which the bride’s mother displayed her wealth. Those who could afford to undertook a procession to the husband’s house, bearing household goods and cloth to be presented to the bride. The number and quality of pieces of cloth was linked to the mother’s wealth, and it might be years after the bijam (wedding) rites that a mother could afford to “send the daughter to the husband’s house.” Cruz interviewed women who “never went to the husband’s house,” implying that they never were able to afford the gifts of cloth and household goods. Thus gifts of cloth were, and are, central to the construction of social relations, with the crucial difference that the source of cloth has shifted from an item of household production to an item obtained through the market economy.
Local spinning and weaving declined within the last two generations. In 1931, a touring District Commissioner (Russell 1931a,b) noted that weaving was common around Banda-Ahenkro and Adadiem (a village on the west side of the Banda hills; Figure 11.1). Documentary sources suggest that cotton was grown only sporadically on the eastern side of the Banda hills, and then only for household consumption. Raw cotton was, however, abundant in the markets at Sampa and Bonduku (Chief Commissioner 1926; Russell 1931b:4, 6). The tools required for producing thread and cloth were simple, and their acquisition was easy; thus, according to oral sources, virtually every woman was a spinner, and every man a weaver. As elsewhere, cloth was produced primarily for the household, though women recounted to Cruz that they occasionally sold their thread to dyers and weavers who produced for the market. While dyeing appears to have been a specialized activity of Muslim men on the western side of the hills, some elders in villages on the east side of the hills recall that dyeing was a household activity (Stahl, interview at Bui, 6/25/89). Middle-aged and old women still have spinning equipment in their houses today and display with pride their spinning skills. Spindle whorls are still made in the area today, as a specialized product of Muslim men in the village of Kokua near the border with Côte d’Ivoire (Crossland 1989:52, 78–79, Fig. 23). Peddlers purchase the whorls at the Kokua market, reselling them at surrounding markets (Crossland 1989:79). Interviewees in villages throughout the Banda area reported that they most often obtained their spindle whorls from Kokua, though one old man in Makala reported that local women occasionally made their own (Stahl, interview at Makala, 6/30/89). While it is clear that growing involvement in the market economy removed textile production from the household, the effects on male–female property relations are less clear. Cash became the medium through which cloth was accessed, but Banda women may have been less dependent on men to access cash than in other case studies (e.g., Etienne 1977; Roberts 1984). Today some Banda women engage in cash crop production independent of their husbands (see below). Pottery production too provided a means by which women could access money and goods.
The potting craft was eroded, though not eliminated, as the Banda area became increasingly enmeshed in a market economy. Today metal and plastic vessels offer alternatives to locally produced earthenware pots. Yet pottery continues to be made and used in the area, and provides a means by which women in villages on the western side of the Banda hills access cash today. Although local potting has diminished in the face of competition from metal vessels—first imported, and more recently made in Ghana (at Tema)—pottery is still produced in three villages on the west side of the Banda hills (Adadiem, Bondakile, and Dorbour). Women in Bondakile produce on a large scale and supply a number of surrounding markets (Crossland 1989:51–82). Potters in Adadiem specialized in the production of large vessels, such as those used for water storage, but demand for their product has dropped off considerably as people increasingly use metal or plastic basins and drums. In Dorbour, where Cruz undertook a three-month study of contemporary potting (Cruz 1996, n.d.) women produce a variety of vessel forms, though on a smaller scale than at Bondakile. Potting is complementary to farming, and its proceeds provide women with a supplementary cash income. In fact, the pressures of monetization seem to be the primary motivation for women taking up potting today; more young women in Dorbour are learning potting than previously, and several older women who formerly spun cotton as their primary craft activity are today learning potting. In one case, a young widow recently began to perfect potting skills she learned as a child in order to provide for herself and her children. Other women who lack potting skills acquire unfired pots from potters, which they subsequently fire and sell for a small profit.
Potters in Dorbour obtain their clay near the Palati River where several clay pits have been dug. They dig their own clay, often with the help of children or their male relatives. Women may not enter the pit while menstruating. Men may dig clay to sell to potters who are unable to do it themselves. By contrast, women in Adadiem are not allowed into the clay pit. The source of clay here was opened relatively recently (in the last 10 to 15 years) and is reportedly inferior to their previous source, now exhausted. Women in Adadiem told Cruz that the clay in the old source disappeared when women were forced to begin digging the clay as men increasingly migrated to cocoa-farming areas of the south. Women in Bondakile are also dependent on male relatives to dig the clay (Crossland 1989:55). Although the case is less marked in Dorbour, potters in all three villages rely on the labor of relations, most often men, to acquire raw materials for potting. The stories told by Adadiem potters suggest that this is a point of contention between men and women—male out-migration limited women’s access to their male relative’s labor, diminishing women’s ability to generate a product that was their property.4 While cash generated by male labor on southern cocoa farms may have flowed into the household, women had more direct control over cash and goods acquired through potting.
While the labor of men and children is crucial in acquiring raw materials, pots are made exclusively by women, who may be helped by a daughter who is learning the art. Women fashion pots in their household courtyards, working on six to eight pots at once over the course of three days. Although a potter may make a wide variety of pot types, she works on one type at a time. The product is homogeneous in size and decorative treatment. In Dorbour, pots are made by modeling, pulling the clay upward using a rounded piece of calabash or a metal lid as a turntable. The potter fashions the rim and pot body on the first day; once these are dry, the pot is turned upside down and the base added on the second day, sometimes using a wooden paddle to shape the added clay. On the third day, after the base is dry, the interior is scraped to thin the vessel walls, and the body decorated using a corncob. The rim and shoulder areas are burnished with a pebble (see Crossland 1989:56–59; Frank 1993:387–390).
Firing is a collective activity, for it involves a rapid succession of activities that must be completed while the pots are hot. Yet each woman’s pots are kept separate during the firing process. Pots are piled on a bed of small pieces of wood; additional branches or bark from larger trees are placed vertically on the sides of the pile. The entire firing process takes an hour to an hour and a half, depending on the size of the pile and the number of vessels; however, the firing itself takes only 30 to 40 minutes. Once the fire has died down, the pots are removed and may be blackened with dry groundnut shells or grass and dipped in a solution of pounded bark and water. The solution gives a shiny finish to the vessels and reduces their porosity. Only water storage vessels, which rely on the porosity of the pot to cool the water, remain unfinished (see Cruz n.d. for a more detailed account of contemporary potting; also Crossland 1989; Frank 1993).
The range of contemporary pot forms has been affected by competition with manufactured alternatives. Demand for large storage vessels and cooking pots that were a mainstay of Adadiem potters has been eroded by the growing availability of more durable metal and plastic forms. Potters in Dorbour make primarily small cooking and water storage pots, still valued for use on farms where people do not want to leave more expensive metal substitutes untended. Small ceramic pots are also used in most households for boiling herbs and barks used as medicine. Competition with manufactured alternatives has eliminated the production of some forms altogether (as in large men’s and small women’s eating bowls).
Potters have responded to declining market demand through innovations in style and marketing strategies. New forms have been added to the ceramic repertoire (for example, bowls with interior striations used for grinding vegetables, modeled after an Akan vessel form; see Crossland 1989:72–73). Younger potters in Dorbour experiment with new decorative treatments with an eye to increasing consumer appeal. Cruz’s data also suggest innovations in marketing. Women on both sides of the Banda hills recall that women formerly traveled to the potting centers to acquire ceramic vessels. Formerly, parties of young women were sent to the potting villages by the mother of the bride to acquire the complement of pots that a mother supplied for her newlywed daughter. Other times women bought pots in excess of their own needs. They sold the surplus pots in villages and markets on the east side of the hills as well at more distant market centers (such as Wenchi and Techiman). With declining demand over the past several decades, potters themselves began to transport their wares to villages on the east side of hills or to the large market at Bonduku in Côte d’Ivoire. Potters from Dorbour or their female relatives head-load the pots a distance of roughly 20 km on footpaths through the hills to eastern villages where they sell their wares at small weekly markets at Banda-Ahenkro and Sase, and by traveling door-to-door. They stay with members of their extended family until they have sold their entire load.
Often times, attempts to model production in the past (whether concerned with issues of gender or not) treat the type of ethnographic baseline outlined above as a relatively direct model for past production (e.g., Crossland and Posnansky 1978). The result is a “mapping on” (Wylie 1985:94; 1988) of ethnographic detail to archaeological contexts, which limits our ability to examine change over time. Here, we use our ethnographic baseline as a comparative model (Stahl 1993:250–252) to identify patterns of similarity and difference between production and consumption, past and present. This enables us to understand how local craft production was affected by changes in the broader political economic context, which we briefly describe in the next section.
The Banda area has a long history of involvement in a broader regional and subcontinental political economy. The area immediately north of the forest was home to a series of important transit markets from the 13th century (Arhin 1979:1–17) where goods from the north (salt, cloth, copper alloys, and slaves) were exchanged for forest products (kola nut, gold, and slaves). One of the earliest entrepôts was Begho, located ca. 35 km south of Banda-Ahenkro near the contemporary village of Hani. Forest gold and kola were funneled through Begho to the termini of the trans-Saharan trade (Posnansky 1987; Wilks 1982a,b). Begho, with its links to the Mediterranean economy, was eclipsed by the growing Atlantic economy and concomitant development of states in forested regions to the south from the 18th century. Asante, the most powerful of these, expanded its control over most of present-day Ghana during the first half of the 18th century (Wilks 1975; 1993). Banda was forcibly incorporated into the Asante confederacy in the dry season of 1773/74 (Yarak 1979) and remained under Asante hegemony until the British invaded Kumase in 1896.
As an inner province of Asante (see Stahl 1991), Banda was required to supply soldiers to serve in Asante wars. Oral historical and archival sources chronicle the frequency of war, especially during the second half of the 19th century, a time when British interests were becoming increasingly entrenched along the Gold Coast. The oral traditions of Banda and surrounding peoples are replete with references to warfare (Table 11.1). Episodes of warfare, though perhaps periodic and exaggerated by oral accounts (Ameyaw 1965), must have disrupted the daily routines of household production and reproduction, including craft production. Casualties were often heavy, leaving households with inadequate resources for household reproduction (Stahl n.d.). Some families augmented their numbers by purchasing captives, who were subsequently adopted (Stahl and Anane 1989:28). The accounts of wars associated with Banda’s incorporation (1773/74) into Asante are particularly vivid (Ameyaw 1965). Banda won an initial encounter, but Asante forces regrouped and attacked again, two miles south of present-day Banda-Ahenkro. When it became apparent that defeat was inevitable, old men, women, and children fled into the mountains and took refuge in the caves above Banda-Ahenkro. Banda forces retreated into the mountains, positioning themselves on the crest of one of the hills, while the Asante occupied the towns below. Food was in short supply, and famine ensued in the hideout, ultimately forcing Banda peoples to surrender. During later conflicts, Banda peoples reportedly abandoned the area altogether. Banda traditions (Ameyaw 1965:7–8) describe sojourns in Bona to the northwest, Gyaman to the west, and Longoro/Nkoranza to the east/northeast at different points during the 19th century (Table 11.1). Scenarios of village abandonment are corroborated by observations of the first British representatives in the area. Lonsdale (1883) observed ruined villages all along the road from Wenchi to beyond Menji on his march to Bonduku, a time when Banda peoples reportedly had abandoned the area because of hostilities with Gyaman.
Asante’s northern provinces were an object of British territorial ambitions in the closing decades of the 19th century. Banda officials signed a treaty with the British in 1894 (Ferguson 1894) as part of a wider effort to encircle Asante with groups loyal to the British, at the same time staving off competition from rival European powers (see Arhin 1974). When the British emissary first contacted Banda peoples, they were living in refuge at the village of Lawra or Bue (Bui), along the Black Volta River, in the aftermath of hostilities with Nkoranza (Table 11.1). Banda peoples were also under pressure from the mounted troops of the Imam Samori. Armed with guns, Samori’s forces forged an empire through wars of conquest (1861–1898) that included areas from northern Sierra Leone to northern Ghana (Holden 1970; Muhammed 1977). Forced eastward under growing pressure from the French, Samori became a factor in Gold Coast politics during the 1890s, when he shifted his base of operations to Bonduku, capital of Gyaman. His forces pursued a scorched earth policy in western Gonja (immediately north of Banda; Figure 11.1), burning villages, confiscating crops, and enslaving captives (Haight 1981; Northcott 1899: 16). Samori’s troops reportedly stayed north of the Volta River, leaving Banda peoples at Lawra in peace. Nonetheless, Banda supplied foodstuffs to Samori, and a large number of Banda people on the north side of the river were taken captive (Fell 1913). Only after Samori was driven from the area by British-led troops did Banda people resettle their old villages.
Table 11.1. Conflicts Involving Banda.
The British were slow to establish an administrative presence in areas north of Asante. Although Asante’s northern provinces were considered part of the colonial realm from 1897, effective administration did not begin for almost 20 years. Colonial officials were concerned with increasing the flow of raw materials to the metropoles (such as industrial oils, cocoa, and cotton) and the consumption of finished industrial products in the colonies (see Constantine 1984). Monetization of the local economy was a prerequisite to successful penetration of the market economy and was encouraged by leveling taxes in new currencies (Guyer 1995; Hopkins 1970). Officials also endeavored to extract labor from northern areas of the colonies to work the mineral and timber reserves that were generally concentrated in the south (e.g., Bassett 1995; Grier 1981). Yet the effects of these regional efforts varied. The limited documentary evidence suggests that Banda was not subject to sustained efforts to impose cash crop production (see neighboring regions of Côte d’Ivoire, Bassett 1995; see also Isaacman and Roberts 1995:9), nor to the same labor extraction techniques as areas to the north (see Grier 1981). District officials toured the area only sporadically.5 Nevertheless Banda farmers were expected to send foodstuffs to the market at Kintampo, then the District Headquarters, though they did not always comply (Roy 1902). And the chieftaincy was required to provide carriers to serve district officials in numbers ranging from 42 to 80 per month. The Banda chief complained to the District officer in 1901 that his “men were away a month at the time carrying for the government, that the farms suffered in consequence of their absence, and that they themselves returned very thin and pulled down, unable to do a day’s work for a considerable time afterwards, and when they were fairly fit and recovered they were sent off again” (Walker 1901). A 1926 report (Chief Commissioner 1926) noted that Banda had fulfilled its obligations in supplying food to colonial troops garrisoned in the area and was being called on to supply 250 carriers to accompany the troops when they decamped. The Chief Commissioner instructed the District Commissioner “not to make any fuss” should the Banda chief refuse to comply: “simply inform him that if the carriers are not forthcoming the troops cannot go, and he must carry on with the provision of food. The odds are that there will be more carriers than required” (Chief Commissioner 1926). Although the demands seem to have been sporadic, they surely impacted the availability of labor and agricultural surplus. Finally, some Banda men migrated south to work the cocoa farms around Sunyani, with consequences for household labor allocation.
Throughout the period in question Banda was linked to the Atlantic economy, its political economic life dominated first by Asante, and later by the British. Ultimately, products introduced through the Atlantic trade undermined local craft production. While potting and cloth production survived well into the 20th century, we turn now to considering how production strategies may have been altered by broader political economic circumstances, paying special attention to the issue of how periods of warfare and dislocation, as well as pressures of monetization, may have affected craft production.
Gold Coast colonial officials showed special concern for the problem of “sanitation.” This encompassed a variety of practices, from refuse disposal, the disposition of human waste, and burial practices to “village planning” (and see Comaroff and Comaroff 1992:40–42 on the “body work” of colonial production). Village planning schemes involved relocation to nearby sites where villages could be laid out on a grid pattern (Gold Coast Colony 1910:14; 1918:20; Stahl 1994a). Local district commissioners were very successful at convincing Banda villagers to rebuild their villages (Stahl n.d.), resulting in a series of archaeological sites abandoned around 1920. We have conducted three seasons of archaeological investigations at one such site—Makala Kataa. Locals report that the site was abandoned when a British official, locally known as the “breaker of walls,” convinced Makala peoples that their houses were built too close to one another and represented a fire hazard (Stahl 1994a). When Stahl first began working in the area, several older residents of Makala recalled living on the site as small children. This village was probably established after 1896 when the British pushed Samori’s troops to the north and represents an occupation of roughly two decades. We refer to this area of the village as Makala Phase 1, or late Makala. An earlier village occupation occurred to the southwest, which we call Makala Phase 2, or early Makala. Based on dateable imports and thermoluminescence dates, this village was occupied during the late 18th and early 19th centuries.
During the course of three field seasons we have tested both occupation and midden contexts in both areas of the site. At early Makala, we have excavated 156 m2 with a volume of 132 m3. We excavated a comparable area at late Makala (155 m2), but with a considerably smaller volume (90 m3) given the shallow nature of the deposits in this area. We isolated architectural features in both areas, and these, combined with the character of midden deposits and material culture, suggest differences in length of occupation, and therefore the “permanence” of settlement, as well as in processes of site abandonment. The evidence for these differences will be detailed elsewhere; however, we briefly outline them here since they have important implications for the nature of production and consumption.
The early 19th-century occupation at early Makala has an air of permanence about it—the houses were substantial, probably coursed earthen-walled structures (McIntosh 1976), with individual structures joined in compounds. We have evidence for refurbishing and rebuilding in the form of superimposed floors and layers of slurry comparable to the plaster used to finish floors and porches today. The height of the midden and the density of refuse suggests a sizable occupation for a considerable period (on the order of decades). Yet evidence suggests that early Makala was abandoned rapidly. A kitchen area associated with one of the residential mounds at early Makala is especially revealing in this regard. Here we found a number of whole ceramic vessels left behind in useable condition, along with numerous grindstones and hearthstones. These are objects that would be salvaged in most cases (as illustrated by one of our workmen taking the hearthstones home to be used by his wife at the end of the field season!). When viewed in light of the oral histories that stress the frequency of warfare in the area, a scenario of rapid abandonment is plausible (see Cameron and Tomka [1993] on the signatures of abandonment).
Late Makala differs in several ways. The early 20th-century houses at Late Makala have a less permanent quality about them. There is less overburden on floors, suggesting that walls may have been constructed from less durable wattle and daub. Patterned post holes at the base of our units are consistent with this insight. Further, the mounds appear to represent isolated structures, not joined in compounds. Contemporary Banda peoples build this type of structure today on farms—in contrast to coursed earth structures, pole and daga buildings can be raised rapidly and require less material, though they are less permanent. This is consistent with the historic scenario that late Makala was founded after British forces eliminated the threat of Samori (that is, after 1896); this followed a decade of considerable dislocation, with Banda people moving about, seeking refuge in several places (Table 11.1). On their return to Makala, people may have constructed shelters relatively rapidly, intending to build more permanent structures in the event that peaceful conditions prevailed. Thus, the visit by the “breaker of walls” may have offered a convenient opportunity for rebuilding (Stahl n.d.). The evidence of abandonment differs as well. There was little in the way of useable material culture left at late Makala, consistent with people relocating a short distance from the site, allowing them to salvage useful items as needed (e.g., Lightfoot 1993; Tomka 1993).
How was production affected by periods of political economic upheaval that find expression in both the oral historical and archaeological records? By comparing material remains from these two temporally controlled contexts with our contemporary baseline, we are able to gain insight into changing patterns of ceramic production and consumption, which provide us a springboard for examining the implications for gender and the labor process.
The early and late ceramic assemblages at Makala Kataa show considerable overlap in vessel form and decorative treatment; however, preliminary stylistic analysis documented increased homogeneity in decorative treatment through time (Stahl 1994a:197–199). This seemed to signal changes in production during the course of the 19th century. Large, abandoned clay pits on the eastern side of the hills attest more widespread production in the past, and the Makala data suggested that potting may have become more geographically restricted during the course of the 19th century. Neutron activation analysis (see Arnold et al. 1991; Blackman et al. 1989, 1993), which provides insights into the geochemical signatures of clays and their sources, has helped sort this pattern out more clearly. Cruz analyzed 160 samples that included clay samples from functioning and abandoned clay pits, both east and west of the hills; contemporary ceramics from Dorbour, Adadiem, and Bondakile; archaeological ceramics from the early and late occupations at Makala Kataa; and archaeological ceramics from historic middens (1930–1960s) in Banda-Ahenkro, excavated by Andrew Black, a doctoral candidate at the State University of New York, Binghamton. Only rims were submitted for analysis, and the sample of archaeological ceramics was stratified by vessel form—large bowls, small bowls, large jars, and small jars (Table 11.2).
Table 11.2. Source of Neutron Activation Samples.
Neutron activation analysis differentiated three groups among analyzed clays and ceramics based on geochemical signatures (Cruz 1996). Clay samples collected from abandoned pits on the east side of the hills (Bui, Bungasi, and Sabiye) fell into one group (East side group). Clay samples and modern pots from Dorbour and Adadiem clustered into what Cruz has called the West side group. Bondakile clay and pottery formed a distinctive group. These clusters provided a baseline against which to compare the archaeological ceramics.
With the exception of two bowls, all vessels (n = 37) from early Makala (Phase 2) belonged to the West side or Bondakile groups, both on the west side of the Banda hills (Table 11.3). Chemical signatures suggest that bowls came disproportionately from Bondakile (n = 14), while the West side group yielded mostly jars (n = 20). By contrast, vessels from late Makala (Phase 1) were concentrated primarily in the East side group (n = 35), with the exception of large jars (n = 4) and large bowls (n = 4), which were associated with the West side group. No vessels in this sample originated in Bondakile. The small sample from historic middens at Ahenkro was divided between the East (n = 5) and West (n = 4) side groups. Thus, neutron activation analysis suggests changes in where villagers on the east side of the hills obtained their ceramics, and perhaps by extension, changes in sites of production (for details see Cruz [1996, n.d.]).
Table 11.3. Results of Neutron Activation Analysis. Source of vessels by provenience and vessel type.
The significance of these results is brought into focus when viewed against the political economic context sketched above. The early occupation at Makala Kataa dates to a the period after Asante had asserted its authority over Banda. The character of the archaeological deposits suggests a relatively long-lived occupation. The neutron activation data demonstrates that women in this period obtained their pots almost exclusively from producers on the western side of the Banda hills, suggesting first, geographically restricted production; second, production by women in villages west of the hills at levels beyond household needs; and third, a degree of regional stability. During periods of warfare, oral sources suggest that travel and market activity were restricted, especially travel through the sparsely inhabited hills. But the period of relative quiescence represented by early Makala appears to have ended abruptly, and historical sources point to warfare as a likely cause. While it is not possible to pinpoint precisely the length of hiatus between the early and late occupations, the neutron activation data suggest a change in where women acquired pottery. Now pots came primarily, though not exclusively, from sources east of the Banda hills. The size of the abandoned clay pits near Bui and Sabiye suggest production on a considerable scale. This does not mean that potters on the west side of the hills ceased production; they may well have continued to supply surrounding villages. However, only very specialized forms (large bowls and jars) found their way across the Banda hills. Women on the east side of the hills focused on production of smaller forms.6
In sum, we see a shift from a geographically restricted (that is, early Makala) to a more dispersed (late Makala and historic Ahenkro) pattern of production through the course of the 19th and 20th centuries.7 The challenge is to identify why this shift occurred. One contributing factor may have been the unstable political economic conditions that prevailed at the end of the 19th century. Banda was involved in several wars from the mid-19th century, forcing Banda people to flee the area (Table 11.1). This quite obviously disrupted existing regional networks of craft production and trade. With the reoccupation of the former villages after British pacification (late Makala) the threat of warfare diminished, but for a people who had faced the threat of slave-raiding from Samori’s troops and had experienced repeated episodes of warfare, women may have been disinclined to travel to moderately distant potting centers to obtain household necessities. Another possibility is that Banda women acquired new skills during their sojourns in other areas. Alternatively, oral histories suggest a significant influx of captives and refugees into Banda during the second half of the 19th century (Stahl 1991, in press), who may have included potters who took up their art in their new homes (see also Frank 1993).
Whatever the scenario, monetization and an expanding market economy probably played a role in women’s decisions to take up potting in villages on the eastern side of the hills, and just as likely played a role in its rapid decline in the second half of the 20th century as earthenware vessels came under increasing competition from metal forms. By the early 20th century, people were increasingly enmeshed in a market economy, as suggested by the plethora of petty trade goods including glass, imported beads, and pipes at late Makala (Stahl 1994a). We might anticipate altered productive strategies as men and women struggled to gain access to the currencies required to purchase these luxuries that ultimately came to be defined as necessities (Arhin 1976/77). By the mid-20th century, potters on the west side of the hills were again marketing their wares to consumers on the east of the hills, as illustrated by the Ahenkro midden data. Ultimately, women on the east side of the hills gave up potting altogether, while those on the west side persisted in their craft. If we take clues from Cruz’s ethnographic data, women in the western villages altered not just their production strategies, but also their marketing strategies, to ensure a broader market for their products. This was a crucial change, for it shifted the burden of transporting pots (a not inconsiderable one) from the consumer to the producer. When women in villages east of the hills gave up potting, they gave up one means of accessing cash or goods obtained through barter. In some cases this may have increased women’s dependence on their husbands and male relations to access the cash required to purchase clothing and household goods (see Etienne 1977; Roberts 1984), though some Banda women responded by taking up cash crop production (see below).
Our archaeological data are less robust when it comes to textile production. The only material trace we have of the apparently ubiquitous cloth industry is spindle whorls. We recovered these in small quantities from both early (n = 9) and late Makala (n = 9). Because spindle whorls are obviously curated, we might anticipate that they would occur in small quantities in archaeological context; however, their limited numbers makes it unlikely that market production of cloth was undertaken at either locale. An intriguing contrast is provided by our excavations at the earlier site of Kuulo Kataa (AD 1300–1650), which is contemporary with the site of Begho (35 km to the south) where Posnansky (1976:48) reported finding a “fairly large number of spindle whorls” from all levels in each locality. By contrast, excavations at Kuulo Kataa (covering an area of roughly 90 m2) have yielded a single fragmentary spindle whorl from the upper 10 cm of deposit. While negative evidence is problematic, it raises the issue of whether ethnographic models that stress the role of domestic spinning and weaving in establishing and maintaining household relations have great time depth. Cotton cloth may have been a prestige item restricted to a small segment of the population during this earlier period (that is, before the 18th century), with other textiles (such as bark cloth) being more widespread. Rather than an item of household production, cotton cloth may have been produced by specialists in market centers like Begho.
A final category of craft that disappears altogether from the archaeological record at the end of the 19th-century is locally made smoking pipes. Local pipes are short, fat-stemmed affairs that were smoked by placing a reed into the end of the pipe stem. These pipes were ubiquitous at early Makala (n = 173). Stylistically they are quite variable; however, many had “quatrefoil” bases or stems common on 19th-century pipes in Ghana (see Stahl 1992:126–128 for a summary of Ghana pipe styles). Many were finely crafted—highly burnished, finely incised, with well-executed, highly individualized decorative treatment. While we await the results of neutron activation analysis on a sample of pipes, for now we are unsure of where they were made. By the later occupation, local pipes virtually disappeared and were replaced by imported European ball clay pipes. Whereas we recovered only three fragments of imported pipes from early Makala (all from surface or near surface contexts), there were 128 fragments of imported pipes from late Makala, and only 5 local pipe fragments. Here is a clear sign of involvement in the market economy, with equally clear effects on a local craft. This is a dramatic example of an object that shifts from a site of local to foreign production. Any suggestion as to whether pipes were produced by men and/or women is pure speculation, though the fact that spindle whorls are today produced by Muslim men suggests that men may have been involved.
As archaeologists work to engender a more distant past, there is a temptation to treat ethnographic scripts of gendered activity as isomorphic with the past. Some believe there is safety in numbers—if we can document similar patterns of gendered production in many societies (the criterion of ubiquity; Stahl 1993:249), then we can be relatively certain that similar patterns pertained in the past (e.g., Hayden’s comparative ethnography, 1992a:34–37). But normative ethnographic accounts of what men and women do in the present overlook at least two important factors: individuality and history. While it is clear that growing involvement in a market economy and pressures of monetization forced Banda men and women to alter their productive strategies, individual choices differed (and see Kriger 1993). Thus, while some women in villages west of the hills chose to pursue potting as a strategy for generating cash, others became involved in cash-crop production. Cash-cropping is often perceived as a man’s domain; however, there are no proscriptions against Banda women establishing farms, and indeed, some Banda women have become successful cash-crop farmers. For example, beginning in the mid-1980s, the Pioneer Tobacco Company began to encourage large-scale production of tobacco by supplying would-be tobacco farmers with seedlings and the supplies needed to build drying barns. Men were heavily invested in yams and calabash as cash crops and were reluctant to switch to tobacco. Women were first to take up the Pioneer scheme; when it became clear after one or two harvests that there was money to be made, men became involved in large numbers. Still, in 1995, the farmer with the largest harvest was a woman.8 Nonetheless, women were dependent on the labor of others—often men and children—and accessing this labor was likely a source of tension as suggested by the stories of Adadiem potters. But not all women engaged in potting or cash-crop production, and thus they became dependent on their husbands and male relations to access clothing and tools of domestic production (such as cooking utensils). In sum, individual strategies varied.
History matters too, as does the broader regional political economic context within which household production and reproduction takes place. As Wolf (1982) argued, involvement in the world economy transformed the societies that anthropologists study, sometimes in patterned ways. The implication is that contemporary patterns may be the result of relatively recent historical periods, limiting the applicability of a comparative ethnographic approach to the study of a distant past. The ethnographic model of potting in Banda tells only part of the story. The archaeological data makes clear that potting—in contrast to textile manufacture—was geared to production beyond the needs of the household during the first half of the 19th century. Women (some of them, anyway) in villages west of the hills produced pots in excess of their household needs, which were consumed by women in villages east of the hills. Yet this pattern broke down late in the 19th century—and potting became more geographically dispersed. Now women in villages on the east side of the hills obtained only specialized forms from the western potting villages, other forms were now produced on the east side of the hills. This “choice” may have been made out of exigency—insecurities created by warfare and slave-raiding may have made regional trade a risky business, forcing women to adopt alternative strategies. In the long run, potters east of the hills gave up their craft, perhaps because their products were inferior, and in the face of declining demand, they could no longer compete with potters from western villages, who were now taking upon themselves the burden of transporting the pots. Changes in the sites of cloth and pipe production were more dramatic, shifting outside the local area altogether and fueling in part the need to generate cash.
As we have tried to demonstrate here, the observation that women make pots can be broken down into more interesting questions, such as under what circumstances do women choose to invest their labor in the manufacture of pots? And why were women content to rely on extra-household labor to produce a crucial tool of household production (pots), while production of textiles appears to have been a household activity? And why did potting persist while textile production declined? We suspect that the answers have something to do with the different roles of material culture in the social world (Stahl n.d.)—textiles were invested with more than the simple function of clothing bodies. They were, and are, important in creating and maintaining social relations (and therefore in social reproduction), as well as in creating and maintaining prestige, while pots in this case seem not to have been so invested (see Hall, Chapter 11; on clothing see Hendrickson 1996; see Costin 1996 for a case where control of textile production shifted to the state).
The value of the ethnographic model in this study is as a comparative model against which to assess patterns of similarity and difference in historic and archaeological sources. The dissimilarities highlight for us how the past may have differed from the present, and in turn cast present circumstances in greater relief—this exercise in thinking about the gendered nature of craft production in the past has made us think in new ways about production in the present, spurring us to ask new questions that our present data are sometimes unable to address. While one of us (Stahl) was initially skeptical about the value of adopting a gender focus, this has proved to be a rewarding avenue of inquiry because of the way it has humanized our understanding of the political economy. We come away with more questions than we can answer, but more engaged in the study than ever.
The Banda Research Project has been supported by numerous agencies and individuals. Archaeological investigations at Makala Kataa were funded by the Wenner-Gren Foundation for Anthropological Research (1989; G-5133); the National Geographic Society (1990; Grant #4313–90); and the National Science Foundation (1994; SBR-9410726). Historical and ethnographic research was supported by grants from the British Academy (1986) and the National Science Foundation (1994–1995; SBR-9410726). The Neutron Activation Analysis was supported by National Science Foundation funds through the Archaeometry Laboratory at the University of Missouri Research Reactor and Sigma Xi funds awarded to Maria das Dores Cruz. Hector Neff Michael Glascock, and the staff of the Research Reactor gave critical help in the NAA analysis of Banda ceramics. Author Cruz further acknowledges the support of the Fulbright-IIE/Luso-American Cultural Commission and the Louise Woods Memorial Scholarship Fund.
The research was licensed by the Ghana Museums and Monuments Board, and we thank the Acting Director, Dr. I. Debrah, for his longstanding support and assistance. We are grateful to the staff of the Ghana National Museum and the Department of Archaeology, University of Ghana, Legon, for their kind assistance. Special thanks to Professor J. Anquandah, former Head of Department at Legon and now Dean of Social Sciences. Our work would not be possible without the support and patience of the Banda Traditional Council, including the paramount chief, Tolee Kofi Dwuru III, and his elders, Tolee Kojo Donkor and the women and elders of Dorbour, and the Banda people. Special thanks to the families in whose midst we have lived—first James Anane and his wife, Adjua, and more recently Sampson Attah and his wife, Afiriye.
Conversations with colleagues and students have influenced our thoughts on gender and political economy. Susan Pollock encouraged author Stahl to take gender seriously, and gender studies conducted by Binghamton students, notably Jennifer Astone and Catherine Dolan, demonstrated the value of a gender approach to her. We thank Susan Kent for the invitation to contribute. Susan Kent, Diane Gifford-Gonzalez, and Susan Pollock gave helpful feedback on earlier drafts. As always, thanks to Peter Stahl and members of the Banda Research Project for valuable input along the way.
1. For examples in textile production see Bassett 1995; Maier 1995; Roberts 1992; on metallurgy see David and Robertson 1996; Goucher and Herbert 1996; Rowlands 1989; Schmidt 1996a,b; Warnier and Fowler 1979.
2. Involvement in the cash crop sector varies somewhat, with villages along the Banda-Menji road more obviously involved in producing crops for market. Nonetheless, trucks travel to even the remote villages to buy surplus yams and calabash during peak harvest periods.
3. Smuggling was a common activity in border areas like Banda. Wax prints were purchased in Côte d’Ivoire and head-loaded into Ghana on foot paths. The sparsely inhabited Banda hills were ideally suited for this clandestine trade.
4. See Moore and Vaughan (1994), who examine the contesting claims of men and women with regard to agricultural labor in northern Zambia.
5. In 1931, Russell reported that seven or eight years had elapsed since the last tour of inspection of some Banda villages (Russell 1931b).
6. It may be that clays on the east side of the hills were unsuitable for production of large vessels, for the underlying geology east and west of the hills differs (Bates 1962; Kitson 1924:40–41; Gay 1956:2–5). This remains speculative without further testing of clays. However, Cruz’s informants stressed the importance of expertise in producing large vessels.
7. It is interesting to speculate on whether this corresponded with a shift in the nature of production. In the case of early Makala, villages west of the hills were clearly producing beyond the needs of their households. As production became more dispersed, and women east of the hills became involved in potting during the occupation of late Makala, were more women involved in potting (now for household consumption)? Or were the pots coming from eastern villages produced by a small number of specialists producing well beyond their household needs?
8. We have few details on the role of Banda women in cash crop production, a topic that will be the focus of future research.
