3—
Exclusion, Classification and Internal Colonialism: The Emergence of Ethnicity Among the Tsonga-Speakers of South Africa

Patrick Harries

Introduction

This chapter deals with the formation of ethnicity amongst the Tsonga-speaking people of the northern and eastern Transvaal. It argues that the notion of a 'Tsonga' ethnic group as defined by anthropologists at the turn of the century is of little objective value for it was more a product of their social and intellectual environment than an objective reality. While Tsonga-speakers were often excluded, as foreign immigrants, from local society, ethnicity as an expression of group consciousness emerged only with the development of a local petty bourgeoisie. This class was literate in a missionary-constructed lingua franca, later to evolve into the Tsonga language, and was equipped with a social and economic vision that extended well beyond the boundaries of the scattered precapitalist political unit of the chiefdom. A Tsonga ethnic consciousness initially arose out of the conflict between the petty bourgeoisie as the representatives of a new form of production and the traditional chiefs, as well as from the new identity forged by Tsonga-speaking industrial workers. This consciousness grew on the base of a modern economic infrastructure, and it should not be seen as a residual or transitional political identity that will eventually be transformed, by the same process of modernization, into a broader national identity. The roots of Tsonga ethnicity are not to be found in the primordial values associated with the precapitalist political systems and forms of production that early anthropologists categorized and classified by 'tribe.'

I shall argue that the degree to which an ethnic identity is adopted is dependent on the various class interests engendered by the historical regional division of labour or centre-periphery form of internal capitalism that has developed in southern Africa.^[1] People choose, adopt and emphasize cultural symbols that they believe to be signs of a shared historical Tsonga identity in order to benefit their class or regional interests.

Although whites had long employed popular stereotypes of the Tsonga-speaking population as a group, it was only in the 1950s-60s that the government actively intervened through its Bantustan policy to forge an ethnic alliance between chiefs and elements of the petty bourgeoisie and to impose a central homeland upon the disparate peoples who spoke Tsonga-related dialects.

A Tsonga identity has thus emerged as the product of a variety of forces. These include the European obsession with social classifications; a government policy that attempted to divide Africans along ethnic lines; and, as well, an awareness expressed by many Africans of the numerous benefits that accrued from the mobilization of people along ethnic lines. What must be questioned is the readiness with which historians have fused the two forms of ethnicity—early classificatory ethnicity and later politicized ethnicity—and have consequently extrapolated into preceding centuries the existence of ethnic groups, such as 'the Tsonga', imbued with a political and social unity that in reality only emerged in the twentieth century. Nor can Tsonga ethnicity be traced to an ethnos with the physiognomy of a soul. It is, rather, very much a human construct, a social product the fabrication of which over time is the subject of this essay.

The Migrations of Tsonga-speakers

In the early nineteenth century people who were later to be defined as Tsonga-speakers occupied the whole of Mozambique south of the Sabi river except for the enclaves of territory surrounding and immediately to the south of the town of Inhambane. For the major part of the century the chiefdoms to the north of Delagoa Bay fell under the hegemony of the Gaza Nguni; those to the west of the bay were heavily influenced by the Swazi; while the Zulu dominated the chiefdoms to the south of the bay. All three states incorporated, both politically and culturally, many of the people living on their borders who spoke language forms later to be classified as Tsonga.^[2] These people, particularly traders and hunters from the areas around Inhambane and Lourenco Marques, had for many years operated in the high and lowveld areas to the west of the Lebombo mountains. Here their commercial skills, rites, customs and organization bound them together and, together with their foreignness, singled them out as a distinctly separate group.^[3] But it was only in the second quarter of the nineteenth century that coastal peoples settled the area that was later to become the northern and eastern Transvaal in a purposeful way.

During the 1820s a number of Nguni refugee groups fleeing from the disturbances in Natal associated with the growth of the Zulu state passed through southern Mozambique. In the late 1830s one of these groups, led by Soshangane, took advantage of a decline in Zulu power following the battle of Blood River and reoccupied the fertile lower Limpopo. According to missionary historians who gathered evidence a half-century after the event, Soshangane's return to the south initiated a 'general exodus' of people living between the Nkomati and Limpopo rivers.^[4] The refugees travelled along the trade routes flanking two rivers, the Olifants and Limpopo-Levubu, that passed through the thinly populated lowveld and gave access to the healthier, well-watered areas to the west.^[5] The second major wave of immigrants entered the Transvaal as a result of the Gaza civil war and ecological upsets of 1858–62. Raids into the coastal and northern Delagoa Bay hinterland by the defeated heir to the Gaza throne and his Swazi allies continued into the 1870s, causing the movement of people into the Transvaal to continue unabated.^[6] The flow of war refugees was continually augmented by people fleeing from natural upheavals such as drought, famine and smallpox.

These immigrants settled under virtually every chief living on the escarpment and eastern highveld. Although they often shifted their political allegiances, they were welcomed by chiefs to whom they paid tribute in labour and in goods. They were attached to homesteads as individuals or, as small groups under their own

headmen, were scattered throughout the veld, colonizing those areas where human and animal diseases, poor soils or lack of water had previously restricted settlement. Coming from a different ecological area, East Coast immigrants introduced new types of food such as fowls, cassava, certain kinds of groundnuts, various grain and potato strains and, especially, maize. These new foodstuffs, together with their techniques of preparation and cooking, served as cultural markers that defined these displaced people, in the eyes of the autochthonous population, as outsiders. Chiefs competed with each other to attract these East Coast immigrants by offering them security and access to a means of production.

Probably the best known of these chiefs was João Albasini who at one time controlled a following of more than 5000 refugees in the Spelonken hills at the foot of the Zoutpansberg. Albasini, in contrast to many chiefs, allowed his followers to retain their clan-names, material culture and leaders without discriminating against them as foreigners. They were also offered the possibility of rapid advancement in his service as hunters, traders and raiders-cum-tax-collectors. His followers were a cultural and political conglomeration of people from various coastal chiefdoms. They had few links with other refugees drawn from the same region and were often embroiled in conflicts with these people, as well as with local chiefs who had given them refuge. Albasini presided over a shifting population the size of which depended upon his fluctuating fortunes. At the height of his power in the early 1860s he attracted large numbers of Venda-speaking refugees from north of the Levubu as well as Tsonga-speakers drawn both from the coast and from other local chiefdoms.^[7] As Albasini's power was destroyed by the decline of elephant hunting and slavery many of his followers deserted him for wealthier masters. Albasini, however, was merely one of many chiefs living in the Transvaal who gave asylum to large numbers of people uprooted from their homes east of the Lebombo mountains.

By the 1860s four small semi-independent clusters of East Coast refugees had begun to emerge in the Transvaal. The middle and lower Levubu river was largely settled by members of the Maluleke clan who were perennially under Gaza hegemony. The heterogeneous population in the Spelonken hills accepted the overlordship of Albasini and his chiefs. To the south the other major groups, consisting mainly of Baloyi and Nkuna clan members, lived in the Haernertsberg under various North Sotho-speaking chiefs. Smaller settlements, many of whose members were described as 'Hlangaan', developed to the south of the Olifants river around Pedi, immigrant Boer and Swazi communities.^[8] The discovery of alluvial goldfields in this area in the 1870s attracted large numbers of coastal immigrants who settled as labour tenants on company lands surrounding the mining villages. The final major wave of East Coast refugees, consisting of several thousand Gaza Nguni, settled in the eastern and northern Transvaal in 1897 following their defeat in the second Luso-Gaza war. In the twentieth century the movement of Mozambicans into the Transvaal, drawn by better living conditions, continued unabated despite a legislative attempt in 1913 to restrict this immigration.^[9]

By the early twentieth century East Coast immigrants were scattered throughout the northern and eastern Transvaal; many barriers divided them from each other and, although they tended to settle in low-lying areas and river valleys, no natural frontier separated them from their neighbours. Even where borders can be distinguished, immigrant communities were surrounded by wide belts of mixed settlement, and islands of linguistic minorities existed on both sides. In 1896 a northern Transvaal newspaper remarked of the immigrants from the East Coast that they were 'an admixture of refugees . . . they have no recognized king, but

subject themselves to the most wealthy, who assumes the leadership under the title of induna or headman'.^[10] It is ironic that the geographically diffuse and politically amorphous form of settlement practised by these nineteenth century immigrants was also recognized by N.J. van Warmelo, the government ethnologist who was later to play a leading role in their definition and delimitation as a 'population group' with its own 'homeland'. In 1937 he wrote that

the Tsonga-speaking refugees came over the border [sic ] in small parties and settled down wherever they could. Very often they became the subjects of Sotho and Venda chiefs and, though the tendency to reassemble and live together was there, they usually failed to muster sufficient strength to form tribes of any importance.^[11]

Even to use the term 'Tsonga-speaking' with reference to the nineteenth century is misleading as it invokes an erroneous linguistic unity. No single language linked the early East Coast refugees who were settled throughout the northern and eastern Transvaal. Instead, they spoke a rich variety of language forms that reflected their diverse geographical origins. The codification and categorization of Tsonga as a language was only undertaken at the end of the nineteenth century by missionaries who, trained to categorize and classify national groups and characteristics according to language, reified this linguistic category into an ethnic group—into a 'tribe'.

Missionaries and the Definition of the 'Tsonga Tribe'

During the nineteenth century a number of popular stereotypes were applied in a rough way to those people living in, or originating from, the region east of the Lebombos and north of Zululand. The term 'Tonga' entered the English language as a borrowing from the Zulu who used it in the nineteenth century to refer to all the conquered peoples of the coastal areas north of Zululand.^[12] It was used in a similarly pejorative way by several chiefdoms in southern Mozambique, at least one of which was later to be classified as Tsonga-speaking.^[13] There are two hypotheses as to how the Nguni arrived at the term 'Tonga'. The first is that a phonetic soundshift caused the Tsonga word /ronga/, which denotes 'easterners', to be pronounced /tonga/ in Zulu. The second is that the original inhabitants of the East Coast were called 'Tonga' before waves of Bantu-speaking immigrants entered the area some time before the sixteenth century.^[14] Despite the imprecision with which the Zulu used the term, the Natal colonists adopted the word as a general term applied to those people living along the coast north of the Zulu border regardless of their actual linguistic or cultural affiliations. Because of the derogatory overtones with which the term was imbued, however, it was never used by the people to whom it was applied. It was only in the twentieth century that linguists expurgated the abusive connotations from the term, initially by introducing an aspirated /h/ and finally by adopting the term 'Tsonga'.^[15]

The Portuguese at Lourenco Marques distinguished between intrusive Vatua or Nguni-speakers and the indigenous Landims, a practice that has led some historians to view the terms 'Tsonga' and 'Landim' as synonymous.^[16] In the Transvaal, East Coast immigrants were given various general labels by the people with whom they came into contact, such as 'Knobnoses', given to them by Europeans because of the distinctive nasal cicatrization practised by some of the coastal peoples, and 'Gwambas', a term applied by indigenous Africans living in the Zoutpansberg and Spelonken hills. This label, initially derived from the name of a chief near Inhambane whose followers had traded in the northern Transvaal,

had evolved into a synonym for 'easterner'.^[17] On the diamond and gold fields, 'East Coasters' were also called 'Shangaans', a term that, correctly used, should be applied to those people who adopted the material culture of the Gaza Nguni chief Soshangane. Thus, in present-day South Africa, only the descendants of the Gaza Nguni immigrants who entered the eastern Transvaal after the second Luso-Gaza war of 1897, are officially classified 'Shangaan' or Tshangana' and in this way distinguished from the descendants of earlier immigrants, the Tsonga, who were in most cases never under Gaza rule. Yet the word 'Shangaan' has become an all-embracing term used to refer to the Tsonga-speaking peoples of southeast Africa and, in a more general way, to all Mozambicans employed on the South African mines. It is obvious that during the nineteenth century these terms were used in a generic and popular way to embrace diverse peoples and chiefdoms with no common name and, as ethnographic terms, they are of very little value.

A far more scholarly attempt to delineate an ethnic group emerged at the end of the nineteenth century out of a heated debate within the Swiss mission over the linguistic relationship between their followers in the Transvaal Spelonken and those living on the East Coast.^[18] When the Swiss missionaries arrived in the Spelonken in 1873 they agreed to restrict their activities to the area south of the Levubu river so as not to compete with the German missionaries who were already conversant with Pedi (North Sotho) to the far south and the Venda-based dialects north of the river. One of the first steps taken by the Swiss missionaries to ease their work was to reduce the various dialects spoken by the East Coast refugees amongst whom they lived to a single written language. This resulted in the publication in the 1880s and early 1890s of a language primer and several religious works in a lingua franca which the missionaries named 'Gwamba'. In compiling this language the missionaries had been strongly influenced by the northern Delagoa Bay dialect spoken by their assistants and by most of the immigrants in the Spelonken.^[19] Consequently when missionaries were sent to evangelize the linguistically-related people living near Lourenco Marques they were greatly disadvantaged in their competition with local Wesleyan evangelists because of the foreignness of their northern dialect.

To avoid being viewed as foreigners from the north, or 'Karangas,' Henri A. Junod, one of the missionaries in the Lourenco Marques area, started to codify another local lingua franca which he called 'Ronga'. In 1894 Junod produced a basic Ronga reader and over the following three years he completed a grammar and collection of folktales and an extensive ethnographic work on what he then referred to as 'the Ronga people'.^[20] It was this division of the 'Ronga', both linguistically and socially, from the 'Gwamba' in the Spelonken area that sparked off the debate within the Swiss mission over how to categorize the people among whom they worked.^[21]

Henri Berthoud, who was the mission's leading expert in 'Gwamba' and who, as a major explorer, was familiar with many of the language variations of southeast Africa, argued from a pragmatic perspective that a single language with a common grammar and orthography would reduce the mission's printing costs. He also opposed the adoption of a further grammar and orthography as he feared, with good reason, that the creation of two written languages would divide the followers of the mission. Berthoud hoped that the written Gwamba language would unify, in much the same way as Jacobine French, High German, or Castilian had in Europe linked large numbers of people who shared, however distantly, a linguistic relationship. He argued that Junod's classification of the Ronga and other peoples was as arbitrary as the mission's earlier categorization of

Gwamba. He was particularly opposed to Junod's claims to 'scientific' criteria; terms like 'Ronga' meaning 'easterners', together with other linguistic sub-groups such as 'Djonga' ('southerners') and 'Nwalungu' ('northerners') were no more precise than the single term 'Gwamba'. Junod's terms were sometimes used as points of reference by outsiders wanting to identify the people of a specific geographical area or by rootless refugees wishing to establish a claim to an area of origin; they did not refer to bounded groups with a common social organization or material culture. As Berthoud stated, 'the Ronga do not form a specific tribe and their name is a geographical designation rather than an ethnographic or linguistic one.'^[22] What defined 'Ronga' as a language was the coastal missionaries' desire for a local written language as a means of spreading the Gospel. The debate between Junod and Berthoud ended with the latter's premature death in 1904.

After spending several years in the Transvaal, Junod published in 1907 an Elementary Grammar of the Thonga/Shangaan Language, which marked the abandonment of the geographically and politically imprecise term 'Gwamba'. Berthoud's defeat was honourably recognized when the following year his supporters published his posthumous Shangaan Grammar. Within a few years distinct Ronga and Thonga/Shangaan (formerly 'Gwamba') written languages had been established on the basis of separate grammars and orthographies. Ronga came to dominate southern Mozambique and Thonga/Shangaan the northern Transvaal and central-southern Mozambique. Meanwhile, American missionaries working in the Inhambane area delineated a third related language which they named 'Tswa'. Despite the defeat of the movement calling for one written Tsonga dialect, the ideal of a single unifying language is still expressed in missionary circles.^[23]

Junod and the other missionaries interpreted the African world through the prism of a specific intellectual system or structure of knowledge. This demanded a classification of the myriad of new details emerging as much from the invention of the microscope and telescope as from the discoveries of explorers. Without the orderly structuring of detail there could be no clarity and no understanding, a factor that was reflected as much in Junod's entomological studies as in his desire for social classification. Junod was particularly influenced by nineteenth century concepts of nationalism and the central role of language in the classification of national groups and characteristics. This is evident in his strong criticism of Dudley Kidd's The Essential Kaffir (London, 1904), which grouped all Africans together, failing to distinguish between them on linguistic or any other grounds. To Junod, Kidd's extreme generalizations, made on the basis of race alone, were both confusing and unscientific. Junod believed that to give a scientific basis to ethnographic observation required the creation of a taxonomy of languages and related social customs. This idea was derived from the work of contemporary European classicists who sought the roots of nationalism in their continent in the languages and social customs of the early European tribes.

Encouraged by evolutionism and by Sir James Frazer, the classicist-turned-anthropologist, Junod used the same schema to make sense of the complex and confusing African world into which he had plunged. But to make African societies fit the European pattern, he resorted to pseudo-history by hypothesizing that at some time in the distant past, migrants originating from different areas had imposed themselves on an earlier proto-Thonga people and had adopted their language. Working from this shaky and, at best, historically speculative premise, Junod saw language as the common thread holding together the Thonga as a 'tribe' or 'nation'. By 1905 he ascribed the 'recognition of the Thonga as a tribe' largely to the work of the Swiss mission.^[24] Seven years later in his two-volume Life

of a South African Tribe, which he published in English in order to reach an influential audience, he divided the Thonga into the 'northern clans', who spoke Thonga/Shangaan, and the Mozambican Ronga who occupied the area south of the Nkomati river. Yet Junod had never been to the southern half of what he defined as the Ronga area nor had he visited the extensive Thonga/Shangaan area north of the Nkomati river or that of the 'Tswa' to the west and north of Inhambane. But because he automatically associated language with culture he unconsciously imbued all the people who spoke these two artificially defined languages with distinctive social customs and traits. His distinction between the Ronga and Thonga/Shangaan was not always clear and, consequently, many historians and others have conflated them into a single Tsonga ethnic group.

Junod's ethnic taxonomy was further reinforced by his interest in folklore, proverbs and folk songs. Contemporary folklorists were deeply involved in European nationalist movements, the roots of whose consciousness they sought, in opposition to the nationalism imposed from above by the philosophers of the Enlightenment, in the cultural expressions of those elements of the nation least influenced by cosmopolitan worldliness—the peasantry. Junod saw songs, tales and proverbs as an expression of the bounded consciousness of the Thonga and as a means of getting at what he referred to as 'the soul of the tribe' or the pure, original culture untouched by waves of immigration or foreign contact.^[25] It was partly his interest in folklore and related particularistic cultures that drove him to oppose the pragmatic centralism of Henri Berthoud's desire to create a single, unifying Gwamba or Thonga/Shangaan language. Because of his positivist approach, Junod, unlike Berthoud, failed to see that his linguistic and related social divisions were human constructs that were in no way scientifically objective. Unlike microbes or river mouths, the Ronga and Thonga/Shangaan languages were not awaiting discovery; they were very much the invention of European scholars and, perhaps even more so, of their African assistants. The linguistic borders determined by the Europeans conformed to a certain preconception of what they expected to find. This is perhaps best illustrated by analyzing the discourse of Henri Berthoud's older brother, Paul, who in 1884 wrote with reference to the Gwamba that:

As a rule a large tribe has not as such, any proper and general name. But the tribe being divided into a certain number of clans, each one of those smaller communities goes by its proper name; wherefore it is incumbent on the foreigners, either black or white, to apply a generic name to all the people and clans which belong to the same tribe. The propriety then, of such a generic name, lies in its being related to the social character of the tribe, and in its being taken from the tribe's own language.^[26]

This was very much the form of ethno-linguistic classification adopted by Junod, who wrote that 'there is no true national unity amongst the Thongas. They are hardly conscious that they form a definite nation, and therefore they possess no common name for it . . .' and that 'there is no feeling of national unity in the tribe as a whole; its unity consists only in a language and in certain customs which are common to all the clans'.^[27]

Contemporary observers in the Transvaal accentuated, concretized and reinforced these forms of social classification on the basis of created linguistic distinctions when they noted that Tsonga-speaking immigrants as outsiders were often different from local people in the manner of their speech, clothing, house form and settlement pattern, as well as the food they ate and the way in which they

prepared it.^[28] They were even described as being physically different from the people amongst whom they settled.^[29] Yet these differences were generally symbols of exclusion rather than cohesion for they were differences often shared with many non-Tsonga speakers.

In fact, there was no 'pure' Tsonga culture that could be regarded as a uniform or static entity, for Tsonga-speaking immigrants in the Transvaal did not come from a common cultural pool. The material culture expressed by an individual was not static or 'traditional', nor was it bound by linguistic affiliations. The cultural markers exhibited by Tsonga-speakers such as diet, tools, clothing, custom and language were, moreover, marked by continual adaptations to changing social and environmental situations. The division of the people of southern Mozambique into various linguistic sub-groups was therefore totally arbitrary and nowhere did they present a common bounded and static linguistic entity.

East Coast traders, hunters and later waves of refugees who entered the Transvaal at different times and from different areas brought with them elements of various material cultures which were, because of their foreign origin, distinguishable from local cultures. But here again, these were factors of exclusion rather than cohesion and the line became blurred as Tsonga-speaking immigrants adapted themselves to their new surroundings. Long distance migrations demanded that fish- or beef-eaters who moved westwards into dry, riverless or tsetse-ridden areas of the Lowveld were obliged to adapt their diet and production strategies to the new environment. In an attempt to assimilate to local norms, some Tsonga-speaking immigrants attended initiation lodges run by host chiefdoms while others adopted local totems. Some continued to practise circumcision which on the East Coast had largely been abandoned by those chiefdoms dominated by Nguni-speakers. Their music was influenced to differing degrees by the Pedi and by people today classified as Venda, Lovedu, Chopi and Ndau, many of whose instruments they have adopted. Many Tsonga-speakers were incorporated, through the ideology of kinship, into host clans in the Transvaal, alongside whose members they constituted a single production unit. This process of individual assimilation was so advanced that it led one anthropologist to speculate that an entire Venda-speaking clan had its roots east of the Lebombos, while another believed that the Tsonga-speaking Baloyi clan had once spoken a Shona-based dialect. An ecologically symbiotic relationship also existed between Tsonga-speaking agriculturalists who colonized malarial and tsetse infested river valleys and plains and the cattle-keepers living above the valley.^[30]

As long as male immigrants stressed their independence by clinging to foreign customs, especially those that were related to sexuality, such as puberty rites and marriage patterns, intermarriage with local people was generally precluded. As outsiders who practised 'barbarous' customs and spoke the local language badly, they were considered inferiors and classified as such by being labelled 'Tonga'. For immigrants to benefit fully from the patronage of members of the host clan they had to resort to fictive filiation and suffer the exploitation as junior members that this often implied. When immigrants arrived in a group under their own chief, however, their position was far stronger and they were more likely to maintain their distinctive material culture.

The cultural boundaries first defined and established by missionary anthropologists at the turn of the century have been extrapolated back into the past by historians who see 'the Tsonga' as a primeval ethnic group occupying a large part of south-east Africa. This static approach to the concept of ethnicity has led to a great deal of historical imprecision such as the extension of the

geographically limited observations of shipwrecked Portuguese mariners to embrace all 'the Tsonga' or to see 'the Tsonga' in nineteenth century Mozambique as being dominated by the Gaza Nguni.^[31] The use by historians of generic terms such as 'Gwambas' and 'Knobnoses' is similarly imprecise.^[32]

What lies at the base of this willingness to accept such classifications is the attempt by Europeans in the nineteenth century to order the African world in their own image. Because they were unable to break from their ideological heritage, Europeans implicitly believed their concept of ethnicity to be the natural order and not merely one convention amongst others used to make sense of the world. Caught within this mental structure, Europeans applied to Africans their own system of ethnic classification and accepted without question that Africans should use the same distinctions and concepts. Thus to Henri Berthoud, 'the Gwamba is to other tribes the same as the Jew is to European nations'.^[33] In this way Berthoud rightly indicated the exclusion of the 'Gwamba' because of their foreign rites and customs but then imbued an extremely diverse and fragmented conglomeration of refugees with all the political and social rites and customs of Jewry. The missionaries used their European conceptual framework to classify the two groups in terms of extreme opposites; the difference between the Venda and Tsonga was synonymous with that between Germans and French or Spartans and Athenians. Consequently, when disputes arose between Tsonga and North Sotho or Venda-speaking chiefdoms, these were interpreted not as clashes between chiefdoms in the way that intra-Tsonga disputes were, but as 'race conflicts'.^[34]

The African Mode of Self-Identification

In reality, however, the Africans conceptualized the structure and order of their world in an entirely different way. The basis of African political and social life was the chiefdom. This grouped together members of the same productive unit and was dominated by the members of one clan. Membership of the clan was expressed through the use of a common patronymic, or shibongo, through which an individual identified himself as a member of his clan leader's house. Outsiders who professed fealty to the clan leader or chief defined themselves as being 'from the land of their host clan while the latter's unifying ideology of agnatic descent provided for their gradual incorporation through the adoption of the clan patronymic. Junod referred to the chiefdom, or tiko, as 'the true national unit' in which political identity was rooted.^[35]

Various symbols bound the members of the chiefdom together and distinguished one chiefdom from another. Foremost amongst these was the institution of chieftaincy, for the chief, as the believed direct descendant of the founding ancestors and as the senior member of the kin group, was the embodiment of clan unity and the centre of its corporate identity. He administered a form of justice that was entirely based on the moral community of the clan and the chiefdom, protected the army with his war medicines, interceded with the clan ancestors and generally regulated production strategies. The chief gave to his followers a sense of belonging and unity by using symbols of office that were believed to invest him with special powers and by organizing various rites that were limited to clan members, such as first fruit ceremonies and entry to the age regiments. The cohesion of the clan and the differences between clans were accentuated by marriage patterns which stressed clan endogamy, by the accreditation to each clan of a separate area of origin and migration and by particularities of dialect.

There were a number of popular controls over the succession to office of the chief. These consisted of various legal precedents which provided for the exclusion from the chieftainship of the eldest son of the chief's first wife. Thus the councillors could declare that for moral reasons he was unfit to rule; a son born of a junior wife, but still of royal blood, could be chosen; or the chieftainship could pass to the line of the regent, who was generally the eldest brother of the deceased chief. Consequently, the inheritance of the chieftainship was not automatic and ascribed; it was processual and was dependent upon the support of the councillors and army whose backing was, in turn, influenced by the power and popularity of the individual competitors for office. An unpopular chief was continually threatened by the segmentation of a part of his following.^[36]

Tsonga-speaking chiefdoms in the Transvaal remained small and independent of one another and manifested no tendency to grow through conquest. But the roots of ethnic consciousness cutting across the divisions of clan and chiefdom may be discerned by the beginning of the twentieth century. Clan endogamy had broken down entirely in areas of the Transvaal like the Spelonken where large numbers of refugees had gathered.^[37] Men were meeting and working on the mines and plantations as 'Shangaans' and 'Tongas' and literacy in Thonga/Shangaan, although limited to a small number of people, provided Christians and traders from different areas with a common means of communication. But the vast majority of the population remained illiterate and the individual's world remained largely a small personal one, limited to the chiefdom with whose members he or she shared symbols and rituals that gave meaning to their lives. Poor communications and a limited area of social and economic exchange further restricted the development of a political consciousness extending beyond the clan and chiefdom. What defined the Tsonga in the final instance were their neighbours. The Tsonga/Knobnoses/Gwamba/Shangaans, as their various neighbours called them, only took on or adopted an ethnic identity later in the twentieth century. This new identity emerged as a result of the politicization of the old classificatory ethnicity—a politicization that was the product of the new economic infrastructure introduced by capitalist development.

Africans and the Land

As the Afrikaner Transvaal state expanded to the northeast during the 1880s, its policies discouraged the formation of large scale chiefdoms by Tsonga-speaking immigrants. A common practice was for white farmers and missionaries to create petty chiefs from amongst the Africans living on their estates. Traditional chiefs tended to live scattered throughout the northern and eastern Transvaal beyond the borders of white-occupied farms.^[38] When Albasini died in 1881, his chiefs were obliged to recognize the suzerainty of the Transvaal state by purchasing their titles from the local Native Commissioner. As chiefs who were unwilling to pay were promptly deposed, the Native Commissioner was soon elected over Albasini's son as chief of 'the Magwambas' living in the Spelonken.^[39] This policy of segregating Tsonga-speaking chiefs from their Venda-speaking counterparts and of neutralizing the Albasinis as a unifying force was continued into the twentieth century.^[40] In many ways, this official view was an extension of local mission policy which opposed the existence of powerful chiefs who could restrict Christian proselytizing.

More importantly, the missions had divided the population of the northern Transvaal along linguistic and geographical lines. Tsonga and Venda speakers were ministered to by, respectively, Swiss Presbyterian and German Lutheran

missionaries who recognized the Levubu as the border separating their parishes.^[41] This division of mission work along linguistic lines left substantial numbers of Tsonga-speakers, living north of the Levubu, in the Venda constituency of the Lutherans while Venda-speakers living south of the river were considered to be in the Tsonga area of the Swiss mission. These geographic and linguistic divisions, initially introduced for pragmatic reasons, were later to become of great political importance as French and German-speaking missionaries sublimated their own nationalistic feelings into an attempt to protect 'their' African people. The Berlin mission was to become the de facto Venda church while the Swiss mission was even at one stage to change its name to that of the Tsonga Presbyterian Church. But just as mission and government interests were both opposed to the formation of large chiefdoms, Republican land policy also actively broke up and scattered existing chiefdoms.

The conditions under which Africans held land in the northern Transvaal changed markedly during the last two decades of the nineteenth century. In the 1870s most Tsonga-speakers lived beyond the reach of any white settlers and even in those districts where the latter owned farms, such as the Spelonken, Africans could live and grow crops almost wherever they wished, including on white-occupied land.^[42] At the end of the 1870s some of the larger white landowners in the Spelonken shared their farms with several thousand Africans. By 1888 it was estimated that some 12, 500 East Coast immigrant families lived on ten white farms in the Spelonken.^[43] It was only in the late 1880s that white settlers started to arrive in the northern Transvaal in appreciable numbers. These were largely landless bywoners who, in exchange for military service, were provided with small 'occupation farms'. But this movement into the northern Transvaal of small-scale, poorly-financed farmers was accompanied by what one missionary referred to as the 'Irish-style eviction' of white tenant farmers.^[44] This was so because, as the Witwatersrand gold discoveries pushed up the price of land and drew Africans more deeply into the money economy, landowners started to turn off their estates bywoners who had been occupying large tracts of land and began to levy direct cash rents from the resident African population. Occupation farmers also exercised a tenuous hold over their newly surveyed farms as these were already densely settled and attempts to remove their African occupants were met with protracted resistance in the form of incendiarism, cattle-stabbing, settlement on fields ploughed by the white farmer, and the destruction of improvements to his farm.^[45] It was these forms of resistance that forced the Republican government to create rural locations, or reserves, where chiefs could continue to exercise a certain degree of independence and thus maintain a hold over, and be responsible for, their followers.

Although locations were first envisioned in 1853, legislated for in 1876 and entrenched in the Pretoria Convention of 1881, it was only in 1892 that a 'Knobnose location' was delineated in recognition of services rendered to the state by Albasini's government auxiliaries. Despite its name, this reserve was settled by a conglomeration of Venda and Tsonga-speaking chiefs and commoners living in the Spelonken and, as the location was unhealthy and deficient in water, most people preferred to squat as tenants on surrounding white-owned farms. To the south, many Tsonga-speakers were included in Modjadji's location and in only one instance was a Tsonga-speaking chief, Muhlaba of the Nkuna, assigned his own location. This was a malarial stretch of land considered by the local native commissioner to be 'quite unsuitable for white people', and the Nkuna refused to give up their tenancy of the neighbouring Harmony Proprietary Company's farms.

Most Tsonga-speakers lived on land that had not been inspected or surveyed for private farms and hence was termed 'state land'. However, by the end of the nineteenth century, Africans were steadily drifting on to white-owned farms. This movement was encouraged by a discriminatory tax system which penalized Africans living in rural locations or on government land with heavy taxes relative to those living as tenants on white-owned land, while those in active service on white-occupied farms paid least.^[46] The sale of state land also caused many Africans to settle on white-owned land. Many were drawn by the fertility and better access to markets of European-owned farms. This movement was facilitated by the large scale sale of occupation farms to land companies and local speculators. The undercapitalized, if not impoverished, occupation farmers were unable to raise capital because the terms of their tenancies precluded the mortgaging of their farms, and they typically lived in 'mud cabins that would disgrace a Connemara squatter or a Skye crofter'.^[47] They were broken by the almost continual commando service demanded by a decade of wars mounted by Pretoria against the northern Transvaal chiefdoms. With their capital invested in livestock and without government aid, they were unable to withstand the effects of the extended drought and the rinderpest epizoötic of the mid-1890s. Many abandoned their lands and turned to transport-riding, hunting, woodcutting and salt-extraction, although even these traditional resorts of the poor had been made increasingly difficult by government concessions and regulations. In 1896 it was estimated that 29 out of the 30 white families in the Lowveld were starving and had been reduced to living off locusts and honey.^[48] The slide of the white community of the northern Transvaal into impoverishment was to continue well into the twentieth century.

Absentee landlords, often mining companies prospecting for minerals, were only too willing to encourage the settlement on their lands of Africans who would undertake bush clearance and pay them rent and grazing fees. Many Africans preferred to live on land owned by the state or absentee landlords, where taxes were lower than in the reserves where, if they paid rent, it was in cash rather than labour and where existing forms of social control and production could be maintained. Others moved from chief to chief or farm to farm in an attempt to better their living conditions. This meant that white farmers had to compete for labour not only with each other and with land companies but also with chiefs living on state and private land and in the reserves. Because of this competition, the labour extracted by white farmers from their tenants could not exceed the combined monetary value of the rents and taxes paid by tenants living beyond the borders of white-occupied farms. Similarly, because of the private reserves that existed on estates owned by land speculators and the state, white farmers were obliged to reserve large parts of their farms for tenants who paid them rents in both labour and money.

The Republican anti-squatter laws of 1887 and 1895 were legislated in order to force African tenants off 'private reserves' so as to spread the labour more equitably and control competition between white farmers. But these laws had the opposite effect for they caused large numbers of Africans in the northeastern Transvaal to move into the Zoutpansberg mountains, which remained largely independent of white rule until 1898, or on to the malarial lands of the Lowveld. Until southern Mozambique was finally conquered by the Portuguese in 1897, Tsonga-speakers also showed a tendency to drift back to their kinsmen east of the Lebombos. All these areas were free of the anti-squatter laws, but, even more importantly, they were relatively inaccessible to tax collectors. The emigration of these loyal Africans deprived government officials of taxes and military allies,

constituted a loss of labour and rents to fanners and closed a healthy production market to white traders.^[49] Hence the implementation of the anti-squatter laws, and even the extraction of taxes, was erratic.

Following the Anglo-Boer War, demands for the implementation of anti-squatter laws increased as the price of land soared and as the labour needs of wealthy white farmers rose with their transition from stock to arable farming. These farmers were opposed to the existence of government reserves which provided Africans with valuable farm land and which pushed up the cost of farm labour by providing Africans with an independent means of existence. At best, government and private reserves were viewed by white farmers as labour pools for mining capitalists. But the British administration in the Transvaal, in its support for mining capital, extended the reserves and made little attempt to evict 'squatters' who paid taxes and rents and who sold a considerable amount of both food and labour to the mines. By 1906 in the Spelonken alone there were over 40, 000 Africans living on land that was owned but not occupied by whites.^[50]

In 1908 the first post-war Responsible Government, which represented wealthy farming interests, moved a year after its election to force African peasants into relationships of labour tenancy on white-occupied farms. A bill was tabled in the legislative assembly with the express purpose of removing up to 300,000 squatters throughout the Transvaal. According to the founder of the Swiss mission in the Spelonken this was 'the most tyrannical law that has ever existed in a Republican [sic ] country, a law that would dismember tribes and clans and disperse thousands of families'.^[51] When the British High Commissioner, Lord Selborne, blocked the passage of the bill, however, the government replied by applying the old Republican anti-squatter laws. This resulted in many Africans in the northern Transvaal moving not on to white-occupied farms, but rather into the rural locations. Between 1908 and 1910 the number of Africans in northern Transvaal locations almost doubled from 52,500 to 101,700; those living on unsurveyed Crown land dropped from 109,000 to 90,000; and those on white-owned land, although still the majority, decreased from 175,800 to 168,000.^[52] A common African reaction to the anti-squatter laws and the increasingly overcrowded locations was for families to club together and purchase land, initially held in 'trust' but after 1905 in freehold. Between 1910 and 1912 Africans in the northern Transvaal purchased more than 16,000 morgen of land worth over £15,000 and by 1913 they held a total of 71,500 morgen in freehold.^[53]

The Natives Land Act of that year was a compromise between mining and landed capitalist interests. It promised on the one hand to extend the rural locations as labour reserves for the mines, while on the other hand it promised, first, to provide farmers with labour, by acting against rent tenancies and, second, to prohibit Africans from owning land outside areas 'scheduled' for their occupation. Land bought by a combination of more than six Africans had to be purchased on a tribal basis and held by the Minister of Native Affairs for the tribe concerned. In later years, the term 'tribe' became a synonym for African purchasers of land in scheduled areas; as one northern Transvaal attorney stated in 1930, 'a Tribe is a syndicate of ten to fifteen families which buys land and elects a chief and petty chief.^[54] The Land Act also encouraged labour tenancy by proposing a graduated tax, in effect an annually increasing fine, on those landowners who accepted rents from Africans in cash or kind. But this section of the act could not be implemented until sufficient land had been released to cater for those rent paying 'squatters' who refused to become labour tenants. For two decades after the Land Act Africans were to retain a precarious hold on their land through the rent tenancy or 'Kaffir farming' system.

In purely economic terms it continued to pay Africans to remain independent producers on company or Crown lands. Here they intermittently paid rent and grazing fees, whereas under a labour tenancy relationship the family head or his sons were required to work for three months each year without pay.^[55] Many white farmers automatically entered into rent paying tenancies with the residents on their farms for, as one northern Transvaal chief stated, 'when a white man buys a farm he finds trees, bushes and natives on that farm'.^[56] A farmer who did not have the capital needed to exploit his land directly would rent out one section and reserve another part for his labour tenants. The persistence of 'Kaffir farming' in the northern Transvaal almost two decades after the passage of the Natives Land Act implies that labour tenancy agreements continued to favour African workers. If the latter felt that the terms of their tenancies were turning against them, they would frequently desert their employers by moving to rented land. They also exercised the more radical alternative of moving on to government land, reserves or mission farms, or of purchasing farms within scheduled areas.^[57]

The preceding paragraphs have stressed that the geographical movement and dispersion of Tsonga-speakers in the Transvaal initiated by the early migrations were increased in the early decades of the twentieth century by the impact of capitalist development. Tsonga-speakers lived scattered throughout the northern and eastern Transvaal and they exhibited little ethnic consciousness, no desire to form centralized political units, and they made no claims to an historical homeland. In 1935 the government ethnologist explained that 'the Tsonga in the Transvaal are, with some exceptions, not organized into tribes at all, but represent a large formless population, the make-up of which almost defies analysis'.^[58] Missionaries, farmers, the Native Affairs department and anthropologists all remarked on the mobility and progressive industry of Tsonga-speakers who lived in decentralized political communities with shallow historical roots and readily adhered to new customs and values.^[59] Local people tended to regard the high achievement motivation of these coastal parvenus with a mixture of hostility and resentment, and Tsonga-speakers in the north-eastern Transvaal soon acquired a stereotype of being pushy and aggressive. At the same time, the undermining of the old forms of political consciousness and patronage centred around the chief allowed a greatly extended and more flexible sphere of social and economic exchange. Out of this fluid situation arose a new class of emergent petty bourgeois farmers who, equipped with vernacular literacy, were to lay the basis for the emergence of a Tsonga ethnic consciousness.

The Erosion of the African Position

As early as the turn of the century, it was noted that African producers in the northern Transvaal annually supplied Pietersburg and Pretoria with 'thousands of bags' of grain and that African maize production in the Zoutpansberg exceeded production in other areas of the Transvaal where Africans dominated the cereals market. The local newspapers frequently reported in the following vein:

The Kafirs grow enormous and increasing quantities of mealies [maize]; quantities so much in excess of their own requirements that the district supplies more of this indispensable article of food for native labourers on the Rand fields than any other part of South Africa.^[60]

African production of cereals for the market was encouraged by both traders and the mines. Nor was the state willing to act against Africans who provided an important source of government revenue; in the years immediately following the

Anglo-Boer War the direct taxes paid by northern Transvaal Africans to the government more than quintupled to £140,000^[61] It is clear that a relatively prosperous, if small, class of African farmers was emerging at the expense of their peers. Evidence for this lies in the purchase of land by individuals who themselves took on rent paying tenants. In 1911 there were 2000 'Shangaans' living on an African-owned farm in the eastern Transvaal and, five years later, there were some 10,500 Africans living on land held in freehold by Africans in the northern Transvaal.^[62] Some of these farmers commanded an annual income of £500 and virtually all had adopted the plough which, together with draught oxen and wagons used for marketing purposes, required a considerable capital investment. Some market-orientated cattle farmers had herds of up to 300 head. Thus by 1930 a number of African farmers had emerged who were able to rent out land and annually market several hundred bags of grain as well as fairly substantial numbers of cattle.^[63] In evidence given to the Natives Economic Commission of that year, it was stated that in the northern Transvaal over the previous forty years, ' . . . [African] marketed produce has increased. This increase is considerably greater than the increase in population.'^[64] According to another witness, 'You will find to-day that [the Africans] have raised tens of thousands of bags of Kaffir corn purely for market purposes and the greater portion of that money which they get for their corn is to pay for land and to buy land.'^[65] But the growth of this African petty bourgeoisie was abruptly truncated in the 1930s as the government intervened in the northern Transvaal to halt the growing poor white problem.

The destruction of northern Transvaal farms by the British during the Anglo-Boer War had pushed increasing numbers of already poor Afrikaans farmers into a marginal existence. In many instances landowners found it more profitable to enter into tenancy relationships with Africans rather than politically more powerful Afrikaner peasants or bywoners .^[66] Although large numbers of whites lived in conditions of extreme poverty in the northern Transvaal, they received little sympathy from the government and, considered 'indolent, lazy and indigent', were treated as a social rather than an economic problem.^[67] The government did however make available a large number of small farms on long lease and with the option of easy purchase in the poorly watered northern districts. But this merely compounded the problem, for by the early 1930s these uneconomic cattle farms had become desperately overgrazed and were occupied by large numbers of settlers subsisting largely on game and maize meal.^[68] It was only in the 1930s, as Afrikaner nationalist politicians under the leadership of D.F. Malan sought to mobilize political support along ethnic lines through the building of new class alliances, that the central government took steps to solve the poor white problem in the rural northern Transvaal.

In the early 1930s, when depression and drought threatened to overwhelm the farming sector, the government supplied white farmers with cattle feed and financial aid. Similar aid was not extended to African farmers whose ability to market their grain crops was severely impaired by the high mortality amongst draught oxen and donkeys.^[69] Most importantly, large-scale government-sponsored land settlement schemes were introduced to relieve the pressure on the overgrazed northern cattle farms. But the settlement of poor whites on over 36,000 morgen of irrigable land, particularly along the upper Levubu, required the removal of thousands of African tenant farmers.^[70] At the same time tenancy relations started to turn against African squatters as the price of land climbed, from an average of 27s per morgen in 1918 to 34s in 1933 with the government paying as much as 180s for irrigable land on the upper Levubu in

price of land rose, white farmers decreased the amount of land available to tenants and limited their rights to graze livestock. Grass burning was restricted and fencing reduced a tenant's rights to commonage. Threatened by a re-emergence of anti-squatter legislation and by the effects of the Depression, speculators started to sell farms that, in many cases, were occupied by over 1000 African tenants. As the the number of 'private reserves' declined and farmers increasingly directly exploited their lands, the amount of labour they demanded increased, and by the the 1930s it was expected that a worker's wives and children would be included in any labour tenancy agreement. The transformation in the 1930s of a large part of the African peasantry into a landless proletariat was movingly captured by a mission-supported African newspaper published in the Spelonken whose editors remarked in 1932:

We are gradually being dispossessed of the land which we and our ancestors, from time immemorial, occupied. Daily we see big parties emigrating from their old homes (because the farmer has bought the farms and requires them to work) to places they might live in security and with freedom. But alas! such a place is nowhere! They may perhaps go to the locations but they will experience in the course of time that they are in no better position as the locations are congested and barren of vegetation.^[71]

Three years later a similar editorial stated that:

Things are rapidly changing and many landowners [have come to] live on their farms where they earn their livelihood through farming. As a result many natives are now turned out from such land to give room to cattle crops, mealie lands and tree plantations.^[72]

As wealthy farmers restricted the amount of land available to African tenants, the latter were obliged to spend more time in search of urban employment which made them and the whole labour tenancy system 'unreliable' in the eyes of the farmers. But most farmers could not afford to pay wages and were locked into the labour tenancy system; some 60 per cent of Highveld farmers were estimated to run Lowveld labour farms and even the Member of Parliament for the Zoutpansberg district maintained three 'squatter' farms. Thus when Chapter Four of the Natives Trust and Land Act of 1936 required farmers to pay a licence fee for each African tenant termed a squatter who performed less than six months' labour service annually, the Zoutpansberg Farmers Union demanded that the chapter be made nonapplicable to their district. If more than three months' labour service were demanded, it was estimated that 25,000 labour tenants would desert their farms.

The crisis of the 1930s showed that the chiefs had little power to protect their followers living outside the reserves, while within the reserves farming was becoming less viable because of the influx of people from the surrounding white farms. Chieftaincy as an institution had been eroded and people were looking for leadership to other political institutions, particularly those led by the rising petty bourgeoisie.

The Waning of Chiefly Power

In frontier areas like the Zoutpansberg, where the ratio of blacks to whites in the first decade of the twentieth century was estimated at 100:1 and where a police force of fifty had to cope with a population of over 300,000, chiefs had of necessity performed the role of paid civil servants.^[73] They were obliged to help

collect taxes and supply labour for public works and farms and prevent what the government declared to be poaching, the destruction of state forests and the consumption of illicit liquor. Native Commissioners were unanimously opposed to the detribalization process as the chiefs 'were of great assistance in maintaining law and order'.^[74] The War Office was particularly mindful that if the authority of the chiefs were to collapse it would be replaced by a wider and more unified political consciousness. As early as 1905 it warned that the breakdown of the chieftaincy system:

does not seem altogether desirable, for a general fusion of the hitherto antagonistic tribes would then be possible and this would constitute a far greater danger to the white community than is to be apprehended from any of the present tribes.^[75]

This perspective dovetailed with that of ethnologists and evolutionist anthropologists such as Henri Junod who feared that urbanization and the loss of chiefly control would lead to the 'demoralization' and 'degeneration' (i.e. proletarianization) of the African population. Junod's work ossified Tsonga-speakers in a pristinely primitive tribal world. It was an unsophisticated and 'natural' world which needed to be protected by being kept apart.^[76] Emerging from the same mould, the young liberal segregationist Edgar Brookes supported the creation of reserves in which Africans could 'develop along their own lines and under their own chiefs'.^[77] At a time when 60 per cent of the African population lived as tenants on white farms or in the cities of South Africa, it was nonetheless commonly believed that Africans lived in 'traditional tribes'! This historically static view was perhaps best expressed in a handbook sponsored in 1934 by the South African inter-university committee for African Studies, Isaac Schapera's seminal collection, The Bantu Speaking Tribes of South Africa . Although this book claimed and probably achieved the status of 'a manual of South African ethnography', the introduction frankly stated that 'the greater part [92.5 per cent] of the book is devoted to an account of the Bantu as they were before affected by the intrusion of white civilization'. Like Schapera, Junod hoped that his work would influence native administrators to understand the exoticisms of tribal life.

At the ideological level the Native Affairs Department was strongly influenced by this strain of anthropology. Members of the Department established compilations of 'traditional laws' by drawing borders that were ethnically conceived around regularities of rite and custom. At the economic level they became increasingly aware of the need to conserve within the reserves the elements of non-capitalist society that bore a large part of the costs of the reproduction of the urban labour force. For, as capitalist development undermined and transformed 'squatting' and the old forms of production in the reserves, it also disintegrated the system of exploitation feeding the growth of industrial capitalism. To check this process, various laws were passed in the 1920s in an attempt to bolster the powers of the chiefs and preserve 'the tribes'.^[78] In many cases, this amounted to creating chiefdoms where none had previously existed.

The 1936 Natives Trust and Land Act 'released' large areas of land in the northern Transvaal for African settlement. All 'scheduled' and 'released' land was henceforth to be purchased on a tribal basis and a Trust fund was established 'to acquire land for and on behalf of specific tribes in order to provide necessary extensions to the tribal locations'. People forced off white farms by the anti-squatter section of the act would be settled under chiefs in these areas. Chiefs

were also given the power to levy special taxes on their followers for the purchase of tribal land. They remained a central element in Native Administration: in 1938 the Native Affairs Commission recognized that

hereditary chiefs with their headmen are the instruments through which native administration works. Without their assistance it would be very much more difficult and very much more expensive to maintain the customary law and order and respect for authority which characterizes the Bantu rural population.^[79]

But attempts to bolster the power of the chiefs were not merely aimed at strengthening the Native Affairs system; they were also, perhaps primarily, aimed at supporting the chiefs whose political power was increasingly threatened by the rising African petty bourgeoisie. As early as 1920, the year of the Witwatersrand mineworkers' strike, the Native Recruiting Corporation of the Chamber of Mines had expressed the fear that unless conditions on the mines were ameliorated,

the different tribes will become more and more in sympathy with one another, with a growing disregard of loyalty to their respective tribal chiefs and a fusion of common interests under the guidance of the educated classes of natives irrespective of tribe or place of origin will result.^[80]

Land alienation, together with tenant and freehold forms of African land tenure had undermined the chiefs' major source of political power: their ability to control the distribution of land. Opposition from white farmers to the sale of land released by the 1936 act continued to deprive the chiefs of any real power. As the Native Affairs Commission complained in 1938, 'the authority of the chiefs and respect for tribal institutions is under continual attack owing to the landless condition of the head of the tribe. This . . . militates against the maintenance of that necessary tribal unity and control which it is the policy of the state to foster.' The popularity of the chiefs had also declined: much of the democratic element in chieftaincy as an institution had disappeared when the size of the chiefdom was petrified and chiefs became civil servants appointed by and responsible to the Native Affairs Department rather than to their own followers.

Liberals like Brookes and Schapera soon came to realize that the reserves only catered for a minority of the African population and that they were unable to perform their protective function; that chieftaincy as an institution had been transformed; and that the chiefs were no longer the sole political representatives of the African population. It was with Junod in mind that Brookes, having broken from his earlier segregationist thinking, wrote in 1934:

The influence of the old school which regarded the tribal Native as the only phenomenon of study, has been great. To those responsible for legislation and administration, it appears as the orthodox school, with the right to monopolize the term 'scientific. The glorification of tribalism and of all the customs that stand behind many of the provisions of the Native Administration Act of 1927 by which the chief has been made an important part of the administrative machinery . . . tends to assimilate all Natives to the position of tribal Natives in the Reserves. Others may be the majority but they are an embarrassing phenomenon . . . they do not live as the social anthropologists think they ought to live . . . do not think on the lines which the Native Affairs Department considers suitable for Natives. They obstinately refuse to develop on their own lines.^[81]

A new school of social anthropology, sensitive to the question of social change, saw major transformations taking place in the ethnic consciousness of people

living in the northeastern Transvaal. In 1938 Eileen Krige reported that

peaceful conditions, closer contacts between members of different tribes in the service of the white man, subjection to a uniform system of European law and administration which calls forth similar responses among different tribes, all are tending towards gradual elimination of marked tribal differences and greater uniformity in tribal customs and conditions.^[82]

The Consolidation of a Tsonga/Shangaan Ethnic Awareness

As the chiefs lost their power of protection and patronage, the chiefdom and clan declined as a focus of political consciousness. Within the clan new political institutions emerged that paralleled those of the chief. These included firmly hierarchical women's organizations associated with epidemic forms of spirit possession and politically structured competitive dance groups.^[83] It was the rising petty bourgeoisie that benefited most from the undermining by capitalism of the old mode of production, from the tenancy and freehold land tenure systems, and from the breakdown of the powers of the backward-looking chiefs. As the chiefdom deteriorated as the centre of economic life and political identification, a new ethnic culture, engendered and encouraged by the petty bourgeoisie, started to play an increasingly integrative role in rural society.

The petty bourgeoisie emerged as an alternative source of political leadership to that of the chiefs through organizations such as the Native Vigilance Association. This political movement was formed in Pietersburg in the early twentieth century and campaigned for African rights through its newspaper, written in North Sotho, Tsonga and English, Leihlo lo Babatsho ('The Native Eye'). Many of the same issues, particularly regarding segregation and land alienation, were later taken up by the northern Transvaal branch of the African National Congress as well as by groups associated with the Communist Party and the northern Transvaal section of the Joint Council System. Many were also to find a political voice through membership of Mission and Ethiopianist churches. But while the petty bourgeoisie expressed itself on one level in national terms, at the regional level it began to foster and articulate an ethnic consciousness.

At the centre of this movement was the Swiss Mission which controlled Lemana college, the only senior educational institution for Africans in the northeastern Transvaal. The annual synod of the Swiss mission was also the only institution linking literate Tsonga-speakers throughout the Transvaal. The political ideology of the mission was strongly influenced by Henri Junod's concept of Tsonga-speakers as a 'tribe' and by the self-help schemes of Booker T. Washington, which had so many parallels with the successful Afrikaner nationalist movement.

The ethnic consciousness expressed by the Tsonga-speaking petty bourgeoisie tended to be a defensive reaction to the politicization of ethnicity. This may be traced to the actions of men like John Dube and Pixley Seme, who sought to mobilize rural support on an ethnic basis after having been ousted in 1917 from the leadership of the South African Native National Congress by the Rand-dominated Transvaalers. The assertiveness expressed by numerically larger and politically more centralized and confident ethnic groups such as the Swazi and Zulu surged forward in the late 1920s and early 1930s. Encouraged by white segregationists, this ethnic assertiveness found expression at all levels of African society and threatened to marginalize disparate and unorganized peoples such as

those considered to be 'Tsongas'. Ethnicity had to be mobilized to maintain the balance of power within the African nationalist movement. The rise of ethnic awareness also indicated a shift in the awareness of the petty bourgeoisie away from politics at the national level, with its more abstract concern for civil rights, towards local issues. These had become of crucial importance because of the extreme social and economic dislocation in the rural areas caused by proletarianization and betterment schemes. Tsonga-speakers also laboured under the very real fear, first expressed at the time of the land commissions, that, as immigrants into the northern Transvaal, they did not have a secure historical tenure to their land. They were subject to attacks such as that made by a Venda headman when addressing the Eastern Transvaal Natives' Land Commission in 1916:

You must take no notice of these Shangaans. They are no good. We are Bawendas here. These Shangaans came to the country. . . . You must remove the Shangaans. There will not be enough room [for us both]. Take the Shangaans away.^[84]

Tsonga-speakers were not the product of a military tradition and had no paramount chief to represent their interests, factors which made them inferior to the more politically centralized ethnic groups in the eyes of many whites and blacks. Consequently, Tsonga-speakers were divided by the Native Affairs Department into administrative districts dominated by Venda or Northern Sotho chiefs, while within the black urban areas they were marginalized as a minority group. The reasons for the emergence amongst the petty bourgeoisie of a Tsonga ethnicity were well expressed by the mission-educated editors of the African newspaper in the Spelonken. In 1934 they gave three reasons for the existence of their newspaper:

We want to arise . . . nearly all our brothers have risen. The Zulus have their national paper, the Xosas have their own National paper . . . the different Bantu tribes are getting up. Our second aim is to create what we call the 'Shangaans National Pride'. Too often we hear of people, Shangaans included who, when amongst nations or tribes other than theirs, are afraid of calling themselves after their own tribe. Some even go to the length of changing their names for fear they will be laughed at. A third aim is to present Shangaan idioms and the language of today.^[85]

The following year the editor urged 'The Shangaan nation . . . [to] learn to respect one of its own, otherwise we shall be no people. The world will laugh at us and everyone will do as he pleases with us.' In another editorial they significantly stressed that 'the increase of land-buyers among us means the increase of investment of wealth for our race'. It also meant an increasing independence of the chiefs and a bolstering of the economic power of the petty bourgeoisie.

Appeals by the petty bourgeoisie to an ethnicity that stretched well beyond the borders of the individual chiefdoms gave them a source of political power, at the regional level, that was not controlled by the chiefs. The rise of an ethnic consciousness also tended to obscure the class divisions accelerated by the economic crisis of the 1930s, for as the ability of the petty bourgeoisie to purchase land was restricted and their productive capacity undermined, they turned to rentier incomes and invested their profits in education. But although ethnicity tended to be the product of a petty bourgeoisie literate in Tsonga and equipped with a sufficiently wide social and economic horizon to be able to incorporate all Tsonga-speakers, it also had roots at a more popular level.

From the late 1850s 'amaTonga' and 'Shangaan' workers were recruited and travelled in batches, consisting of men from the same home areas, along recognized routes to areas of employment scattered throughout South Africa. By the late 1870s a form of nascent worker consciousness linked these Tsonga-speaking communities in much the same way as the 'tramping systems' of England or the tour de campagnonnage of France.^[86] It was in the schools on and around the mines that many 'Shangaan' men first acquired a basic literacy in the Tsonga lingua franca of the Swiss mission. On the mines, 'Shangaans', because of the distance and the impoverished nature of the area from which they came, were prepared to do dangerous and heavy manual work and consequently took on the occupational stereotype of underground labour.^[87] Working in ethnic teams at the rock face bred solidarity. Miners were housed in ethnically segregated rooms and barracks in the mine compounds and their representatives and policemen were appointed on an ethnic basis.^[88] In a harsh and often hostile world, ethnicity took on some of the functions of the extended family. From the mining capitalists' perspective, the tribal structuring of the compounds cushioned the impact of industrialization by retaining the 'moral restraints and standards of tribal life'.^[89] This allowed the mines to argue that tribal (migrant) workers were too unsophisticated to appreciate trade unions and, more importantly, caused working class consciousness to be divided along ethnic lines. As early as 1913, a government report considered that 'inter-tribal jealousies have always rendered it possible, in the last resort, to protect Europeans by utilizing one tribe against another', a view that is still today held by many mine employers.^[90]

These divisions were compounded by the traditional desire of men from the same home area to work and be housed together. Thus Tsonga-speakers tended to gather on certain mines and the home friend network, with its sense of continuity for the migrant, has always been a centre of social activity.^[91] Ethnic competition has been encouraged by the organization of recreational activities on the mines. From the initiation of tribal dancing, dance teams and their supporters were defined and separated along ethnic lines.^[92] As many dance movements were brought from the home area, they were easily distinguishable from 'foreign' dance styles and further entrenched these differences by parodying and stereotyping their competitors. In this way, an ethnic identity has been concretized and, as a Chamber of Mines pamphlet stated in 1947, 'competition between the tribes is encouraged.'^[93]

Ethnic competition was also fostered in various other ways, such as the installation of noticeboards which urge workers to vie with each other along ethnic lines over such issues as room and barrack cleanliness and absenteeism. Promotion has often been dependent on ethnic patronage. The competition between groups of workers on the mines was commonly expressed in songs such as:

Take off your [dancing costume] skins!
There is no relish left, you Shangaans!
It has been eaten by the Sotho and Xhosa,
And we will not get it.^[94]

This emerging ethnicity had a contradictory effect on the development of a working class consciousness. While it divided and weakened the labour movement as a whole, ethnicity strengthened segments of the workforce. For without trade unions to represent their interests, workers often sought political and social solidarity in a shared ethnicity. As early as 1907, a Benevolent Society was formed by largely Tsonga-speaking East Coasters on the Witwatersrand to

help sick members, to provide burials, and to aid Christian proselytizing at home. Concomitantly, groups such as the African Union of Natives of Mozambique emerged on the mines with distinct political interests.^[95] In the absence of broad working class organizations, workers often used ethnic ties to mobilize themselves in the interests of their class, and Shangaans as a group, often in concert with other ethnic groups, played a militant role in initiating boycotts, work stoppages and strikes. The contacts which the home group maintained with its area of origin also gave it the power to influence recruitment, which in turn pressured employers to adhere to contracts. Beyond the compounds the basis of an urban ethnicity was laid when home groups with strong rural links started to form burial societies and savings groups. The sense of ethnic community that emerged on the mines and in the urban areas was constantly funnelled into the rural areas by men returning home. These 'sons abroad' had developed an ethnic consciousness as a means of survival in an unfamiliar and competitive world and they spread their ideas and experiences into rural areas where people had little concept of the existence of a Tsonga or Shangaan 'people'. The emergence of this ethnic consciousness was the product of a complex interaction of forces; these were as much working class as they were petty bourgeois; as much urban as they were rural. It was also to become increasingly clear that ethnicity was as much an imposition of the central state, as it was a product of the people themselves.

The Role of the Apartheid System

In 1948 the Afrikaner National Party came to power with a policy aimed at transforming the reserves into African homelands. In numerous speeches, the new government's ministers stressed the central role of the chief and tribe in the implementation of apartheid in the rural areas. In a speech, typical of the period, the Minister of Native Affairs expressed his determination in 1950 to:

reclaim . . . and restore tribal life as far as possible by seeing to it that the chiefs and the whole tribal government adapt themselves to the exigencies of our times and thereby automatically regain the position of authority which they forfeited to a large extent through their backwardness . . . the natives of this country do not all belong to me same tribe or race. They have different languages and customs. We are of the opinion that the solidarity of the tribes should be preserved and that they should develop along the lines of their own national character and tradition. For that purpose we want to rehabilitate the deserving tribal chiefs as far as possible, and we would like to see their authority maintained over the members of their tribes. Suitable steps will be taken in this direction.^[96]

The first step was the passage of the Bantu Authorities Act of 1951 which bolstered the power of the chiefs by modernizing and expanding their tax basis to include all the members of the tribal authority. This gave the chiefs a new element of patronage through their control of the tribal account and through their participation in the decisions of the Regional Authority. It was estimated at the time that the northern Transvaal reserves were carrying sixteen times too many cattle and five times too many people.^[97] To provide the chiefs with the ability to distribute land and in an attempt to retain the productive base of the reserves, the government started to purchase released land. In concert with the extension of the reserves and the devolution of powers to the Bantu Authorities, a renewed assault was made on African tenants living on white farms, beyond the control of the chiefs in the 'homelands'.

An amendment to the Natives Trust and Land Act in the mid-1950s stipulated that only African families settled on white-owned land before the passage of the act in 1936 were entitled to stay on white farms. All Africans who had been registered after 1936 as 'squatters', working for less than four months annually, were deprived of their rights to stay outside the reserves. The graduated fees payable by farmers for registered 'squatters' were levied uniformly and increased annually. Tenant control boards, consisting of farmers and officials, which had been established under the 1936 Act in order to check 'squatting', started to determine and restrict the number of labour tenants required by individual farmers.^[98] This caused many African tenants, who had formerly 'sold' their labour to the farmers for at least four months a year, to become permanent labourers. This transformation of the farm labour force from part-time to full-time workers made as many as two-thirds of the labour tenants on some farms 'redundant', and these were then 'repatriated' to their 'homelands' as were tenant families whose younger sons had fled to the more open urban labour market.

In 1957 chiefs living on white-owned farms were threatened with the forfeiture of their chieftaincies if they refused to accept resettlement in the reserves. Squatters were also rapidly expelled as their registration fees began to exceed the rents received by the farmer. Labour tenants suffered a similar, although more gradual fate for, as farmers mechanized their methods of production and transportation and increasingly used insecticides, the value of land occupied by tenants and their livestock rose while, in relative terms, the value of their labour declined. The 'natural' movement of tenants from white farms to the reserves, caused by the capitalization of agriculture, was turned into a flood in the early 1960s when labour tenancy was prohibited in the northern Transvaal. Small white farmers were badly affected by the dissolution of the labour tenancy system as it deprived them of their cheap, 'captive' labour force and in many cases obliged them to sell their farms to highly capitalized agro-businesses. The movement of Africans off the white farms that started in the 1930s and reached a peak in the late 1960s and early 1970s reversed a century-old tendency for Tsonga-speakers to disperse throughout the northern and eastern Transvaal. Of central importance was the influx control legislation that directed this efflux from the white farms away from the industrial centres and into impoverished 'homelands'. It was in these areas that ethnicity, particularly in the form of the population relocation brought about by ethnic consolidation, was to become a crucial issue in the 1970s.^[99]

The Bantu Authorities Act of 1951 did not affect the ethnically heterogeneous nature of the population living in the northern Transvaal reserves. A Tsonga homeland was not envisaged and Tsonga-speaking Tribal Authorities were grouped administratively in Regional Authorities, dominated by Venda and North Sotho-speakers in, respectively, the northern and eastern Transvaal. The first move towards an ethnic segregation of the area came from a number of northern Tsonga-speaking chiefs who felt their Regional Authority was dominated by Venda-speaking chiefs. Their calls for a separate Tsonga-dominated Regional Authority and a separate commissioner-general were made at roughly the same time that the report of the Tomlinson Commission indicated a move towards segregation on ethnic rather than merely spatial lines.^[100] But it was only in 1959 with the promulgation of the Bantu Self-government Act that Pretoria asserted that "The Bantu people of South Africa do not constitute a homogeneous people, but form separate national units on the basis of language and culture.' This act formally declared the 'Shangane/Tsonga' to constitute a 'national unit' and allowed the government to accede to the Tsonga chiefs' wishes

for a separate homeland. Four Regional Authorities dominated by Tsonga-speakers were then cut out of the old multi-ethnic regional authorities in the northern and eastern Transvaal and in November 1962 they combined to form the Matshangana Territorial Authority. Seven years later a Tsonga-Shangane commissioner-general was installed and a legislative assembly was opened at the newly-constructed capital at Giyani.

The segregation of the rural areas into ethnically-defined units was paralleled by similar movements in the urban areas. In Johannesburg, culturally mixed communities like Sophiatown were torn down and replaced by townships, like Soweto, that were built on an ethnic grid. The urban population living in rural towns like Louis Trichardt and Pietersburg was also shifted and rearranged according to the apartheid policy of 'ethnic consolidation'. At the same time, Bantu education and government-controlled Radio Bantu orientated people towards the 'traditional' forms of culture dominant in the rural areas. Thus apartheid blocked the process of social integration and cultural hybridization that had emerged as the economy required a geographically mobile African workforce. Under apartheid, the movement to the towns, both spiritual and material, was increasingly directed along ethnic conduits.

The government's new divide-and-rule policy, partly stemming from its own experience of the growth of Afrikaner ethnic consciousness, generated bitter ethnic conflicts over local resources, for in the early 1960s the northern Transvaal reserves remained crushingly overpopulated and overgrazed.^[101] Starting with the break-up and distribution of the funds and powers of the old multi-ethnic Regional Authorities, ethnic bitterness reached a peak as politically arbitrary borders such as roads, railways and farm boundaries were defined in order to separate the different 'homelands.' This immediately created disadvantaged ethnic minorities on both sides of the border. It is the competition for resources that has compelled many people to adopt an ethnic consciousness. This competition takes place between ethnically differentiated people living under the same chief, between ethnic minorities and a Bantustan government or between the members of different Bantustans. These pressures, together with the use by the government of physical force, have obliged large numbers of people to join the flow of labour tenants and 'squatters' into resettlement camps or closer settlement villages in the homelands.

A year after the creation of the Mashangana Territorial Authority, government implementation of its policy of ethnic consolidation resulted in large-scale forced population removals in the Bushbuckridge area of the eastern Transvaal. At roughly the same time an ethnic border was drawn between the Tsonga and Venda 'homelands' that cut well south of the Levubu river and threatened with removal 40,000 Tsonga living north, and 10,000 Venda-speakers living south of the border.^[102] For these people ethnicity became of central importance overnight as did the Territorial Authority in its role as defender of the Tsonga people. Venda-speakers living south of the historical Levubu border were no longer considered integrated neighbours. They were seen in historical terms as foreigners who, in the nineteenth century, had been given asylum by Tsonga chiefs and who consequently had no right to be placed over the heads of the 'indigenous' Tsonga-speakers. The people faced with removal were threatened with great hardships as most were to be settled in the hot, dry and infertile areas close to the Kruger National Park, far from labour and produce markets. Schools, churches, clinics, homes, irrigation works, dams, debushed arable land and various other forms of immovable property constructed out of tribal levies were to be handed over to Venda chiefs and their followers, many of whom were outsiders resettled

from distant farms and 'black spots'. In these ethnically consolidated Venda areas, Tsonga teachers and principals were replaced and Venda became the medium of instruction. Resistance to this resettlement scheme was widespread and resulted in increased hardship when large numbers of people were imprisoned for cutting thatching grass and for continuing to plough their fields. The position of the disadvantaged ethnic minorities was concisely expressed in a letter from one of the affected communities to the Minister of Bantu Affairs:

We will be allowed to stay if we accept Venda laws and allow Venda to be taught to our Shangaan school children. And all the money we are paying is going to the Venda Tribal Authority . . . how can we sell our Shangaan birthright to accept the authority of a Venda man who has been imposed on us from outside?^[103]

The Mashangana Territorial Authority took up the defence of 'its' people by taking the boundary dispute to court, by challenging the dismissal of the Tsonga teachers, by calling for the installation of a separate Shangane/Tsonga Commissioner-General to protect their interests, and by asking for an impartial redelimitation commission on the grounds that the state ethnologist had close links with the Lutheran Church (ex Berlin Mission) in Vendaland. Although the removals did take place in the late 1960s, it was later felt that the stolid resistance of the chiefdoms, when combined with the backing of the Territorial Authority, was to a great extent responsible for the alleviated conditions of their removal.

In 1968 the government allocated land south of the Klein Letaba river to Vendaland, a move which 16 years later, according to a Gazankulu Bantustan government memo, continued to 'cause friction and poor relationship between the Shangaan and Venda . . . the situation is tense and violence may break out at any time'.^[104] Despite this hostility the government continued to enforce ethnic identities on the African population through the passage in 1970 of the Bantu Homeland Citizenship Act. According to the Minister of Bantu Administration and Development, the object of this bill was to:

mature the slumbering national pride and in this way to develop the spiritual and cultural assets of that national milieu and to help develop a healthy self-respect and national pride.^[105]

In 1973, the year that the Territorial Authority became the 'self-governing' Bantustan of Gazankulu, the delineation of one of its southern borders led to threats of war being made in the Giyani legislature against the Northern Sotho of the Lebowa Bantustan.^[106] Further ethnic hostility arose over the allocation of the eastern Transvaal Shiluvane mission hospital to Gazankulu in mid-1981. This caused the chief minister of Lebowa to remove all Northern Sotho patients, nurses and hospital staff to nearby 'Sotho hospitals' and to ban the Tsonga Presbyterian Church (ex Swiss Mission), which he held responsible for the government action, from at least one area of Lebowa. Because of the ethnic bitterness generated by this affair, southern Tsonga-speakers in 1984 would rather spend the money and time involved in travelling an extra 60 kilometres to a Tsonga hospital' than face possible ill-treatment at the nearby 'Sotho hospital'.^[107] In an area of the eastern Transvaal where the Tsonga-speaking Nkuna and Sotho-speaking Maake chiefdoms had lived in harmony for almost 150 years, ethnic antagonism was fanned by the forced removal of several thousand Sotho from land incorporated into Gazankulu. Those who refused to accept the rule of an imposed Tsonga chief or who used grazing or arable land or cut wood or grass in areas from which they had been removed were arrested by the Bantustan

authorities.^[108] This simmering ethnic antagonism led to violent conflicts in mid-1985 between 'Tsongas' and 'Pedis' (North Sotho) in the southern Mhala and Ritavi districts.^[109]

The unfinished demarcation and arbitrary nature of 'ethnic borders' continues to give rise to much inter-ethnic bitterness as ethnic solidarity has emerged as a vital element in the mobilization of people in defence of their property. This has inevitably lent a degree of legitimacy to the Bantustan government as the political embodiment and protector of the Tsonga people. Hudson Ntsanwisi, the chief minister of Gazankulu, excused Tsonga actions against the Pedi by qualifying Pretoria's border delimitation as 'throwing a bone to two starving dogs'. At the same time he threatened the possibility of bloodshed if the Venda, 'the central government's beloved children', occupied a long-established Swiss mission farm that, because of border delineations, had become an island of Tsonga-speakers in Vendaland. In a related speech he declared that, 'the ethnic fires kindled by the Republican government will spread until this continent [sic ] is destroyed irrevocably'.^[110] Yet the Bantustan system with its 'homeland' governments is structurally the major instigator of ethnic feeling.

Gazankulu constitutes a geographically defined and poverty-stricken colony within the borders of South Africa. Influx control and resettlement schemes have caused a gross overcrowding; the average population density per square kilometre is 76 in Gazankulu and 17 in 'white' South Africa. As a result only about 6 per cent of the population can make a bare living from agriculture.^[111] In 1982 there were at least twice as many Gazankulu citizens working as migrant labourers outside the Bantustan (over 40,000) as there were wage earners within its borders (21,000). Although Gazankulu's importance within the regional division of labour has increasingly become that of a dumping ground for surplus labour and a political alternative to African nationalism, it remains a directly exploited periphery within South Africa. The 1983 estimated per capita GNP of R121 (R3157 for South Africa as a whole) does not reflect the sharp regional differences within the Bantustan. In the heavily resettled northern district only 1.3 per cent of the population can live off agriculture while farming in central Gazankulu can theoretically support 10 per cent of the population. Thus in some areas the cost of feeding and reproducing the population is still partially met by agricultural production which consequently continues to subsidize low industrial wages. Women have also come to play an increasingly important role as wage earners who commute to neighbouring white farms. Perhaps more important in the continuing system of centre-periphery exploitation is the fact that the costs of social reproduction in the Bantustans are far lower than in the urban areas whose political importance, magnified by their closeness to the centres of economic power, demands adequate supplies of clean running water as well as transport, housing, medical, educational and other facilities. In the Bantustans much of the cost of providing these services falls on the members of the tribal authorities, a factor that tends to reinforce the underdevelopment of peripheral areas such as Gazankulu.^[112]

One of the few legal ways in which Africans may oppose their regional exploitation is through political mobilization along ethnic lines. This was certainly a major reason behind Ntsanwisi's formation in 1981 of a political party named Ximoko which, modelled on Chief Gatsha Buthelezi's Zulu Inkatha organization in Natal, promotes 'worthy indigenous customs and traditions' and attempts to unite all Tsonga/Shangane against the land claims and imposition of 'independence' by Pretoria. The allied Gazankulu Women's Association wishes 'to revive what our ancestors were doing' and promotes games, cooking, dress,

music and dances that it defines as 'Tsonga/Shangane'.^[113]

While much of Gazankulu has been reduced to the status of a proletarian dormitory for the industrial areas of South Africa, sections of the petty bourgeoisie have benefited from the Bantustan system for as Gazankulu citizens they are protected from 'foreign' entrepreneurial competition. The Bantustan government provides business loans and a wide spectrum of licences to its citizens. In 1983 it controlled over half of the jobs and, in the form of pensions, a further 45 per cent of all regular incomes generated within Gazankulu. It is common for members of the petty bourgeoisie to adopt and stress their Tsonga identity when competing for control of, or access to, those limited resources.

The chiefs are the other major local beneficiary of the Bantustan system and because of this they tend to stress factors, such as ethnicity, that link rather than divide the different chiefdoms. When Gazankulu became self-governing in 1973 the electoral system was structured so that of the 68 members of the legislative assembly, 42 were delegated by the Bantu authorities and hence, ultimately, by the chiefs. Of a five man cabinet, it was prescribed that not more than three and no fewer than two had to be chiefs.^[114] The chiefs monthly income, paid directly by the Gazankulu government, is partially dependent on the number of his followers, from whom he draws a plethora of taxes. If he has sufficient subjects, a chief earning R2,400 a year will be appointed to a seat in the Giyani legislature at the annual salary (in 1985) of R12,400 which makes him eligible for ministerial positions remunerated at over R50,000 annually.^[115] These salaries, in a poverty-stricken rural backwater, provide a strong encouragement for chiefs to accept resettled Tsonga-speaking people with whom the dominant clan and the tribal authority have no relationship other than that of a shared ethnicity. As the local organizers of Ximoko, the chiefs use ethnicity to bind their heterogeneous tribal authorities. Although they have lost to the petty bourgeoisie the final say in matters concerning labour bureaux, court cases and land allocation, the chiefs dispose of a good deal of patronage through their access to the tribal fund, which is the pool for the majority of local taxes. This gives them a firm control over virtually all local development projects from the siting of boreholes and water taps to the establishment of clinics and schools. Through their appeals to a common ethnic consciousness the chiefs and petty bourgeoisie are able to mask the sharp class and regional inequalities that have grown up within Gazankulu. At the same time ethnicity provides the petty bourgeoisie with indigenous 'traditions' and a degree of local authenticity that obfuscate its origins in, and total dependence upon, the system of exploitation controlled by the white minority government in Pretoria. As a form of social exclusivism dividing the African population, ethnicity creates an external enemy that, unlike Pretoria, which in 1985/6 provided well over 65 per cent of the Bantustan's budget, is not responsible for the monthly salaries of the Gazankulu bureaucracy.^[116]

The debates of the Gazankulu legislative assembly are full of rhetoric aimed at the 'Tsonga nation' and the 'Tsonga people'; they rejoice when Tsonga-speakers 'come back home' and decry 'the loss' of Tsonga-speakers who refuse to move to Gazankulu. Conflicts that are in reality between Bantustans over access to resources are often portrayed as attacks on 'the Tsonga-Shangaan people'.^[117] The creation of an historical basis for this ethnic consciousness has required some myth making. According to Ntsanwisi:

We are told that we must build a nation and we want to build ourselves. We want to unite the Shangaan-Tsonga tribes into one single cohesive unit from Komatipoort through to the Sabi river, Bushbuckridge, Letaba, right up to

Malamulele where we have many of our people. We have a common history, customs, language, religion, similar aspirations and a common way of life which will make it easier for us to develop a national spirit and character.^[118]

Ntsanwisi has tried to foster ethnic unity in various ways. The myth of origin of an important local clan has been adopted as a 'traditional' symbol, linking all Tsonga-speakers in the Transvaal in the belief that they are descended from a pair of founding ancestors.^[119] An ethnically-based nationalism is fostered by the Gazankulu radio service and newspaper, by symbols such as a national flag and administrative buildings, the continual codification of Tsonga-Shangaan law and particularly the education system. In the towns, an ethnicity has been encouraged by associations such as the Mashangana Urban Movement and the Tsonga Cultural Academy which have deep roots in the Bantustan.^[120]

Tsonga-Shangaan ethnic nationalism within Gazankulu is spread at a more popular level by the clientalism of the petty bourgeoisie. The patronage dispensed by the old individual chiefs has been replaced by that of the tribal authority and Bantustan government which provide villagers with pensions, jobs and anything from a bus service to drought relief aid. This popular ethnicity is often expressed in the form of songs that praise 'Gazankulu' and 'the Gazas'. Songs and other forms of social expression can take on an extremely aggressive tone when sung by people who have been forcibly removed from their birthplaces by the citizens of neighbouring Bantustans. They will often use the derogatory term 'Vesha' when referring to outsiders like the Venda or Northern Sotho. Various songs proclaim that Tsonga women should not marry Venda or Sotho men, that the latter 'stink' or, accompanied by threatening dancing and gestures, call on the members of the chiefdom to 'kill them all, the Vendas'.^[121] Most of these songs were composed during times of friction caused by the imposition of ethnic borders. But the animosity that they propagate lives as long as their popularity.

The sentiment of ethnic separation is reinforced when neighbouring people refer to Tsonga-speakers, not in a polite manner as 'Tsongas', but in a derogatory way as 'Tongas', 'Koapas' or Thokas'. In areas that have experienced a high degree of ethnic tension, people will claim that Tsonga-speakers have an offensive smell, that their cooking is unpleasant and their manner of dress obnoxious. It is of little relevance that these grounds of exclusion can, at best, be traced to localized brands of soap or fats or to regional forms of dress and cooking. What is important is that some people believe that Tsonga-speakers, as a group, dress, cook and even smell the same. Here the cultural factors of exclusion have become so extreme as to politicize the basic senses of smell, taste and sight. This change in the perception of reality both fuels, and is fuelled by, ethnic hostility.

Conclusion

This chapter has stressed that ethnicity has not grown out of the old values and symbols associated with the chiefdom or out of an inadequate modernization. The façade of traditionalism that is presented by the chiefs is interpreted by modernization theorists as an index of the degree to which primordial ethnicity has withstood the pressures of capitalist development. Ethnicity is viewed in residual or at best transitional terms: as a phenomenon that will disappear as the development of a modern economic infrastructure allows the dominant culture associated with capitalism to penetrate and transform provincial cultures. The Afrikaner National Party holds a more static pluralist view of South Africa as a land inhabited by different peoples, each with its own ethnos . These views of

ethnic identity as a natural order, as a remnant or atavistic manifestation of a historically concrete social unit, result at the ideological level in federal and Bantustan political policies.

Ethnicity is a central concept in South African politics. It is used by the South African government to divide the African population and by some Bantustan politicians, such as Ntsanwisi and Buthelezi, to mobilize opposition to homeland independence. Within the national liberation movement one wing calls for the denial and suppression of ethnic differences in the cause of national unity.^[122] The other wing recognizes the power of ethnicity as an instrument for political mobilization at the regional level. The Freedom Charter which is adhered to by the major part of the national liberation movement, holds that:

While we do not encourage 'tribal pride'—in fact we denounce it—we are far from being indifferent to traditions, languages and cultures of individual ethnic groups; we do not propagate ethnic nihilism . . . our reality is a multi-ethnic society. We respect and strive to develop all local languages and cultures and this helps us to combat all forms of reactionary nationalism, chauvinism and ethno-centricity; it also helps us to improve inter-ethnic relations thus facilitating the drive towards national and social emancipation.^[123]

I have argued that ethnicity should be seen in processual terms as the historical product of internal colonialism. But it has been stressed that ethnicity should not be seen in simple terms as the response, within one region, of a uniform class with identical interests to a situation of core-periphery exploitation and underdevelopment. Ethnicity has emerged out of the acceptance and propagation by various classes of cultural symbols that cut across class barriers and distinguish and unite people as 'Tsonga'. Ethnicity is thus a fluctuating, situational expression of group identity aimed at the achievement of specific political ends. The expression of an ethnic consciousness does not eradicate narrower loyalties to chief and clan; these can coexist with other feelings of class, national or religious consciousness. The individual will adopt, in response to a specific situation, any one of these identities for the purpose of group mobilization. Nor is ethnicity merely a product of Bantustan politics, and it is unlikely that the abandonment of apartheid and the Bantustan system will end the regional underdevelopment which, through the politicization of cultural differences, is one of the major causes of ethnic exclusivism.