Chapter Two—
A Divided Society

South Africa is appropriately viewed, I said earlier, as a society characterized by ascriptively determined superior-subordinate relations. That is the nature of racial domination. South African society is also characterized by ethnic cleavages within the racial categories. Underpinning both sets of cleavages are distinctive intergroup attitudes and orientations. Although I shall spend more time here on the neglected cleavages within the various racial groups, it is worth emphasizing as well that interracial differences are also likely to persist.

Black and White:
Some Preliminaries

In a survey of White elites conducted in the late 1960s, Heribert Adam found decidedly negative attitudes toward Africans. Strong majorities of Afrikaner officials and English entrepreneurs labeled "true" statements to the effect that there are important differences between "Bantu" and Whites, that the "average Bantu is a child, some hundreds of years behind the white man in development," and that some Black-White differences are biologically determined.^[1] Agree-disagree questions have some methodological problems attached to them; however, there is no mistaking the bigoted character of most of the replies. A decade later, a

[1] Heribert Adam, "The South African Power Elite: A Survey of Ideological Commitment," in Heribert Adam, ed., South Africa: Sociological Perspectives (London: Oxford University Press, 1971), pp. 73–102.

large-scale study by Hanf, Weiland, and Vierdag, which generally asked different questions, nevertheless suggested the emergence of somewhat more flexible attitudes and, on the one question overlapping Adam's—racial job reservations that excluded Blacks—produced dramatically different results.^[2]

Change, then, was occurring on some dimensions, but virtually every study shows unfavorable White attitudes toward Blacks.^[3] Afrikaner attitudes are more unfavorable than English attitudes are. Blacks generally reciprocate Afrikaner hostility—indeed, to an extreme degree—although, as we shall see, they are far more favorably disposed toward English-speaking Whites.

We shall review these studies shortly. The best place to do so is in the consideration of African attitudes, because the attitudes of groups toward each other are closely interrelated. For example, we shall observe that, even as Black hostility to Afrikaners was growing, Black views of English-speaking Whites were remarkably friendly. Some categories of Black respondents evaluated the attributes of English speakers more favorably than those of any other group, including other Black groups. By the same token, as Afrikaner views of Africans hardened, their views of White English speakers, Coloureds, and Indians softened. All of this shows the interplay of interracial and interethnic attitudes and relations in South Africa. The results also show that White and Black (especially if Black includes Coloureds and Indians) are not quite the categories in which respondents think. The strongest reciprocal hostility is between Afrikaners and Africans.

In any case, for the moment, it is obvious that, if attitudes are any guide, a nonracial society is not around the corner. Neither is a nonethnic society. Much of the future of South Africa lies in the interplay of race and ethnicity and in responses to these affiliations that do not wish them away.

[2] Theodor Hanf et al., South Africa: The Prospects of Peaceful Change (London: Rex Collings, 1981), pp. 127–241. On job reservations, compare ibid., pp. 133–34, 213, with Adam, "The South African Power Elite," pp. 87-88.

[3] Ans E. M. Appelgryn and Johan M. Nieuwoudt, "Relative Deprivation and the Ethnic Attitudes of Blacks and Afrikaans-speaking Whites in South Africa," Journal of Social Psychology 128, no. 3 (June 1988): 311–23; J. M. Nieuwoudt and C. Plug, "South African Ethnic Attitudes: 1973 to 1978," Journal of Social Psychology 121, 2d half (December 1983): 163–71; Graham C. Kinloch, "Racial Attitudes in South Africa: A Review," Genetic, Social, and General Psychology Monographs 111, no. 3 (August 1985): 261–81; Patrick C. L. Heaven, "A Historical Survey and Assessment of Research into Race Attitudes in South Africa: 1930–1975," South African Journal of Sociology 16 (September 1977): 68–75.

A New Context

In Asia and Africa, prior to independence, ethnic differences among the subject peoples generally were muted. Following independence, however, the context and the issues changed. With independence secured, the question was who in the new state would control it. At that point, ethnic differences became relevant, contradicting the expectations of those who saw in anti-colonial movements the makings of enduring transethnic nationalisms. As Anderson, von der Mehden, and Young have written, " . . . the transcendent obligation of resistance to the colonizer did largely obscure the vitality of ethnicity as a basis of social solidarity. Only in the cold dawn of independence did the potency of this factor begin to become clear." All over Africa, they point out,

scholars, highly sympathetic to the aspirations of African nationalism, looked naturally for the factors of cohesion rather than for elements of potential discord. "Tribalism" was a retrograde force of merely historical interest; the future belonged to the "detribalized." The inevitable urbanization and industrialization would create a social system dominated by economic classes. . . . "Class formation," wrote one distinguished anthropologist, "tolls the knell of tribalism."^[4]

For a brief but intense period in the 1960s, a literature on "nation building" emerged. The problems of nation building were generally conceived as entailing construction of an institutional infrastructure, creating symbols of national identity, fostering economic development, and closing the perceived gap between "modern" elites and "traditional" masses. The problems posed by ethnic, racial, religious, and linguistic pluralism that often came to dominate post-colonial politics were largely neglected.^[5]

The same flavor of neglect is now in the South African air. The struggle against apartheid has created illusions about the homogeneous character of a future South Africa. Part of the problem derives from linear,

[4] Charles W. Anderson et al., Issues of Political Development, 2d ed. (Englewood Cliffs, N.J.: Prentice-Hall, 1974), p. 29, quoting Daniel F. McCall, "Dynamics of Urbanization in Africa," Annals 298 (March 1955): 151–60, at 158.

[5] This depiction, of course, has a certain element of caricature about it, but it also resembles the writing of the period under discussion. See, e.g., Karl W. Deutsch and William J. Foltz, eds., Nation-Building (New York: Atherton Press, 1963); Wendell Bell and Ivar Oxaal, Decisions of Nationhood: Political and Social Development in the Caribbean , Social Science Foundation and Department of International Relations Monograph Series (Denver, Colo.: University of Denver, 1964); Leonard Binder, "National Integration and Political Development," American Political Science Review 58, no. 3 (September 1964): 622–31.

developmental thinking, rather than contextual thinking about group identity. Philip M. Mayer's view of urban African identity is rather widely shared. "The evidence suggests that it is too late for [homeland] political ethnicity" to take hold among Africans in town, he contends; "its place is likely to be taken more and more by Black ethnicity."^[6] When the context changes, however, alternative levels of identity can become more relevant. Similarly, it is argued that the emergence of common cultural norms in urban areas precludes the emergence of ethnic rivalry, on the implicit but certainly unproved assumption that a strong sense of cultural difference is a necessary condition for such rivalry.^[7] South Africa could well enter a period of fundamental change with the same mind-set that prevailed at the time of African decolonization, unprepared for the ethnic diversity that will undoubtedly be one of its conspicuous characteristics.

To be sure, there are bits and pieces of evidence that support the position that Black ethnicity is unlikely to have significant political salience. Progressive South Africans, steeped in the anti-apartheid movement, point especially to the nonracial, nonethnic ideology of the ANC, to the solidarity of various groups in the movement, to noncommunal tendencies in places like Soweto, and to the contrived nature of ethnic identities in South Africa, resulting from the regime's manipulation of ethnicity for its own ends.

The evidence, however, is not convincing, for four reasons. The first rests on the logic of changing contexts. The second relates to the selective character of the data on which the conclusions are based. The third derives from the thrust of the data that a more wide-ranging search elicits. The fourth in lies the propensity for even invented identities to develop their own resonance. I shall take these reasons up briefly in order.

First of all, in South Africa today, the overriding issue is, of course, race. The context of White domination is hardly the best test of Black division. While racial divisions are highly salient, every incentive points to Black solidarity, and even then Black solidarity is incomplete. When White domination is past, a different set of incentives will take over.

[6] Philip M. Mayer, "Class, Status, and Ethnicity as Perceived by Johannesburg Africans," in Leonard Thompson and Jeffrey Butler, eds., Change in Contemporary South Africa (Berkeley and Los Angeles: University of California Press, 1975), p. 153. For endorsement, see Martin E. West, "The Urban African Population of South Africa," in Robert M. Price and Carl G. Rosberg, eds., The Apartheid Regime: Political Power and Racial Domination (Berkeley: University of California Institute of International Studies, 1980), pp. 141–45.

[7] Mayer, "Class, Status, and Ethnicity," p. 155.

Ethnic divisions will become more important, coexisting with and, for some purposes, even superseding racial divisions. That has certainly been the experience of those Caribbean societies that were doubly divided—first on the basis of a color hierarchy like South Africa's and second on the basis of African or East Indian ancestry. When universal suffrage undercut the position of Whites, the focus of conflict shifted rapidly to tensions between Afro-Caribbeans and East Indians.^[8]

As I have indicated, the underlying logic is the logic of context. Who else is in the common environment? Which collectivities are similar, and which are different? Whose interests seem similar, and whose seem incompatible? Through half-perceptual, half-conscious judgments of this sort, group affinities and disparities are discerned, and group alignments and juxtapositions are established. South Africa is beginning a period of rapid change that will alter the context but will not render obsolete judgments like these, which lie at the core of ethnic group formation and re-formation and the ethnic politics that follows.

Second, the arguments that have been made are based on the behavior of selected sectors: certain urban areas but not others (Soweto but not the Durban or Pretoria townships, for example), trade unionists but not dwellers in the so-called homelands, and so on.^[9] Research on rural South Africans is difficult, but they are far more likely, as we shall see, to have retained their ethnic loyalties at this stage than are urban Blacks, who have been on the front lines against apartheid.

Third, the data available on Black ethnicity throughout South Africa, when pieced together, suggest that it is generally similar to ethnicity elsewhere in Africa and in much of Asia. On the whole, ethnicity is a fairly strongly held affiliation that is reflected in a variety of attitudinal and behavioral measures of group difference, an affiliation that can, under certain circumstances, give rise to serious conflict.

Fourth, the argument predicated on the artificial character of some Black South African ethnic divisions is based on a non sequitur—that contrived groups cannot become serious antagonists.^[10] In the discourse

[8] See, e.g., Ralph R. Premdas, "Politics of Preference in the Caribbean: The Case of Guyana," in N. Nevitte and P. Kennedy, eds., Ethnic Preference and Public Policy in Developing States (Boulder, Colo.: Lynne Rienner, 1986), pp. 155–87.

[9] For the suggestion of a fairly strong rural-urban cleavage based on a study of self-rankings, see A. A. Dubb et al., "African Attitudes in Town: The Search for Precision," African Studies 32, no. 2 (1973): 85–97. For further evidence of anti-urban ideology, see P. A. McAllister, "Political Aspects of Xhosa Beer Drink Oratory," English in Africa 15, no. 1 (May 1988): 83–95.

[10] The assumption that, but for the South African state, there might not be any ethnic groups in South Africa is widespread in South African academic discourse. For afairly typical taste, see Wilmot G. James, "Reinforcing Ethnic Boundaries: South Africa in the 1980s," in Susan Olzak and Joane Nagel, eds., Competitive Ethnic Relations (London: Academic Press, 1986), pp. 137–50.

of the South African regime, there has been a tendency to describe ethnic differences among Black groups as if they were given and immutable. At an earlier stage, ethnic affiliations might well have been depicted as deriving from God's plan, the same plan that ostensibly divided the world into races. Research on ethnicity makes it abundantly clear that ethnic groups are by no means given, that ethnic identities have an element of malleability, that groups form and re-form their boundaries.^[11] In South Africa, for example, the Afrikaners are an amalgam of several European components; the long-standing boundary between the English and Afrikaans groups shows increasing signs of porosity, as I shall note below; the "Coloured" category is assuredly a human product, frequently described as "an artificial community created by the official race classification policies of South Africa";^[12] and various Black ethnic groups are also "very much a human construct, a social product. . . ."^[13]

Race, too, is a construct, but in South Africa it is, nonetheless, highly significant. It is a conceptual mistake to leap from an understanding of ethnicity as a construct to a conclusion that ethnicity is politically insignificant. In many parts of Africa, there are ethnic groups whose identity can be traced to official policies, missionary categorizations, and encounters with neighboring groups. These "artificial" creations are often significant political actors.^[14]

The same can be true in South Africa. Tsonga (Shangaan) ethnicity, for example, was initially a product of early European classification, missionary work, later government policy, the desire of members of the category to benefit from largess distributed along ethnic lines, mobilization to counter that of more powerful and centralized groups (such as Zulu and Swazi), and the establishment of a Tsonga Bantustan named Gazankulu, created in 1973 and eventually involved in border disputes with a neighboring Pedi homeland, Lebowa. The Tsonga are neither a group existing from time immemorial nor a merely ephemeral product

[11] See, e.g., Frederik Barth, ed., Ethnic Groups and Boundaries (Boston: Little, Brown, 1969).

[12] Albert J. Venter, "The South African Plural Society: Reflections and Musings towards Its Understanding," Plural Societies 19, no. 1 (September 1989): 1–20, at 7.

[13] Patrick Harries, "Exclusion, Classification and Internal Colonialism: The Emergence of Ethnicity among the Tsonga-Speakers of South Africa," in Leroy Vail, ed., The Creation of Tribalism in Southern Africa (Berkeley and Los Angeles: University of California Press, 1989), p. 83.

[14] See, e.g., the classic treatment of the emergence of the Bangala by Crawford Young, Politics in the Congo (Princeton: Princeton University Press, 1965), chap. 11.

of Bantustan politics. In the judicious view of Patrick Harries, "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."^[15] Neither will the end of apartheid diminish the political significance of ethnicity among other Black groups.

In 1978, a sample of Black men in Soweto with at least mid-high-school education was asked: "What problems do you think might arise in the future if black people should rule South Africa?" Soweto is the most successful Black ethnic melting pot in South Africa, and better educated respondents, such as these, are generally less likely to provide ethnic responses to identity questions. Soweto after the 1976 violence was an angry place. Pan-Africanism and other versions of transethnic Black affiliations held considerable sway. It is all the more remarkable, therefore, that, for this open-ended question, the largest fraction of respondents, 37 percent, named intra-Black ethnic tensions as a future problem.^[16]

These results are discussed by Hanf, Weiland, and Vierdag,^[17] who postulate that apprehension of Black ethnic conflict underlies the frequently expressed Black rejection of unqualified Black majority rule options in sample surveys—a survey result we shall examine in Chapter 3. Here, however, the aim is not to explore this possible relationship. Instead, in most of the remainder of this chapter, I simply wish to inquire into evidence bearing on the perception of the Soweto respondents who anticipated Black ethnic tensions. How is Black ethnicity manifested, and how might it be manifested in a future South Africa?

African Ethnic Attitudes

In 1986, the estimated population of South Africa, including all of the homelands, was about 33 million, consisting of approximately 24.0 million Africans, 2.8 million Coloureds, 0.9 million Indians, 3.0 million Afrikaners, and 1.9 million other Whites (principally British, but also Portuguese, Jews, Germans, and Greeks). The African population was divided into nine main, linguistically based ethnic groups. Very rough, rounded figures for the nine groups are provided in Table 1.^[18]

[15] Harries, "Exclusion, Classification and Internal Colonialism," p. 110.

[16] Lawrence Schlemmer, "Change in South Africa: Opportunities and Constraints," in Price and Rosberg, eds., The Apartheid Regime, p. 278.

[17] South Africa , pp. 439–41.

[18] That these are estimates, rather than conclusive figures, can be ascertained byconsulting different estimates, such as those contained in Marianne Cornevin, "Populations noires d'Afrique du Sud," Afrique Contemporaine , no. 141 (January 1987): 32–49, at 36. For present purposes, however, these estimates are sufficient.

The enumerated categories are by no means perfectly demarcated, either externally or internally. Boundaries between them are not always sharp. Like ethnic groups elsewhere, most of the South African groups could easily be subdivided into segments. Subgroup identification with the overarching group is frequently imperfect,^[19] and conflicts among subgroups are often serious. The Xhosa speakers, for example, would include Xhosa, Pondo, Thembu, and Mfengu, among others. But this is not unusual in ethnicity. The category of English-speaking Whites, for example, also embraces considerable heterogeneity, but that does not render the category politically insignificant.

From Table 1, it is obvious that the two largest amalgams, Zulu and Xhosa, alone comprise about half of the African population, with, respectively, 27 percent and 23 percent of the total. In politics, these two groups are especially important. They also have relatively homogeneous

[19] For an example, see Jean Comaroff, Body of Power, Spirit of Resistance: The Culture and History of a South African People (Chicago: University of Chicago Press, 1985), pp. 38–39. For historical heterogeneity at the subgroup level and below, see Peter Delius, The Land Belongs to Us: The Pedi Polity, the Boers and the British in the Nineteenth-Century Transvaal (Berkeley and Los Angeles: University of California Press, 1984).

home areas, so that surveys in those areas yield responses that are more or less ethnically identifiable.

Consider first an unpublished but important survey administered in 1986 to a large female African sample in four areas: Soweto; Pretoria and the industrial Reef; Durban; and the Eastern Cape.^[20] Although their ethnic mix is different, Soweto and Pretoria, both in the Transvaal, are heterogeneous areas, with many migrants. Durban, in Natal, is about 90 percent Zulu, and the Eastern Cape is Xhosa.^[21]

The sample was asked an identity question, "Would you say you are first of all . . .," and there followed three choices: an ethnic group response such as Zulu or Xhosa, a "Black" response, and a nonracial "South African" response. The responses differed markedly by area. In Soweto, the most frequent response, provided by 46.0 percent, was "South African," followed by "Black" (33.6 percent) and then various ethnic responses (20.4 percent). On the Reef and in the Pretoria townships, the rank order of responses was completely different: 40.9 percent said "Black," 29.9 percent ethnic, 29.3 percent "South African." The rank order in Durban was different from these two, and it was different again in the Eastern Cape.

Since Durban and the Eastern Cape are largely homogeneous, it is instructive to contrast their responses, depicted in Table 2. The largely Zulu respondents of Durban gave an ethnic response about three times as frequently as the largely Xhosa respondents of the Eastern Cape. (The Zulu figures may actually have been higher, because the small non-Zulu minority in Durban, which is mainly Xhosa-speaking, may have provided responses more like those of the Eastern Cape.) For the Xhosa respondents of the Eastern Cape, the "South African" response was over three times as frequent as for the Durban Zulu respondents, and the "Black" response was significantly higher as well.

This pattern was repeated in responses to a question asking, "To which of these regional groups do you feel you belong?" The choices ranged from one's township to one's homeland, to one's city, to South Africa as a whole, and to Africa as a whole. The pan-African response was three to four times as frequent in the Eastern Cape as anywhere else (23.3 percent, compared to 3.7 percent to 6.7 percent elsewhere).

[20] This survey was administered by Markinor to a sample of 800 respondents in the various metropolitan areas. I am indebted to Nick Green of Markinor for making a copy of the results available to me.

[21] For figures on the ethnic composition of Soweto, Durban, and Pretoria, see Hanf et al., South Africa , p. 266.

The Durban sample was at the lowest end here, with only 3.7 percent naming Africa as a whole. More than a quarter of the Durban respondents gave their homeland, compared to about 10 percent of the remainder of the sample. Only 15.6 percent of the Durban people named South Africa as a whole, compared to 43.4 percent in Soweto, 47.5 percent in Pretoria and on the Reef, and 36.7 percent in the Eastern Cape.

Too much weight should not be put on any one survey. The exclusively female sample is one limitation. Another is the urban focus. Like many surveys of Black South Africans, this one does not tap rural responses. For present purposes, however, this is no real shortcoming. On the contrary, the findings can be interpreted to suggest that, even among those with a similar urban experience, ethnic differences in response patterns stand out. Such differences would likely stand out more decisively among rural respondents. A hint to this effect is that poorer and older respondents in this sample tended to give more parochial and more ethnically focused responses.

Especially prominent in the survey is a degree of polarity between Xhosa in the Eastern Cape and Zulu in Durban. Zulu responses are markedly more parochial, more place oriented, assuredly more ethnically focused, and less universalistic. Xhosa responses are more South African, more pan-African, less tied to locality. It is notable that Xhosa responses in the Eastern Cape mention homeland ties only about as

often as do all respondents in Soweto or the Pretoria-Reef area, even though Xhosa in the Eastern Cape are in or near what some might plausibly consider their home areas, whereas many of the Transvaal respondents live far from what they might consider their home areas, and many of their home-place ties are greatly attenuated by now.

This is not, of course, to equate Zulu ethnic-identity and locality responses to the two questions with Zulu ethnocentrism or necessarily to exempt Xhosa from ethnocentrism on the basis of their responses that refer to a wider identity and a wider field. The responses, however, do suggest different approaches to these questions, different conceptions of correct answers to them, perhaps different foci of identity.

As I shall point out, the Eastern Cape, with its long history of intense and often hostile contacts with White invaders, is the center of resistance movements and the center of ANC-UDF ideology. Natal has generally been peripheral to these movements. The Natal Indian Congress (NIC), rather than any African organization, has been perhaps the most important component of the UDF in that province.^[22] Natal is the center of what has become a separate, Zulu-based mass organization, Inkatha—albeit an organization that does not necessarily command an uncontested hold on Zulu loyalty, particularly in urban areas and certainly not in urban areas of the Transvaal. One consequence is that the survey tells us nothing about what Zulu responses are in the Transvaal, where there are many Zulu, or what Xhosa responses are in the Transvaal, although there are many fewer Xhosa there. Despite these reservations, the polarity of the responses is a revealing piece of evidence that needs to be matched against other pieces of evidence: it is no more, but it is no less either.

This survey put the choices more explicitly than most others, but its results are congruent with others, including one that asked an open-ended question of a sample of Transvaal township students: "When asked what group do you belong to, what do you answer?" Zulu gave the highest proportion of ethnic responses, 24 percent, and Xhosa the lowest, 7 percent, with the median of all groups in the mid-teens.^[23]

[22] A circumstance that occasioned the remark of Mangosuthu Buthelezi, leader of Inkatha, that prominent among the UDF affiliates in Natal was the NIC, composed of "a brand of Indian who is poison to the black struggle for liberation." Quoted in Carole Cooper et al., Race Relations Survey, 1987 / 88 (Johannesburg: South African Institute of Race Relations, 1988), p. 33.

[23] Mark Orkin, "'A Divided Struggle': Alienation, Ideology, and Social Control among Black Students in South Africa," Journal of Intercultural Studies , 4, no. 3 (1983): 69–98, at 76.

Again, the contrast is marked, although the total range of difference is relatively modest.

The matter goes considerably beyond questions of identity. There are also Xhosa-Zulu attitudinal contrasts revealed by surveys. On some central political issues, Zulu and Xhosa are out of step with each other.

Giliomee and Schlemmer present the results of a survey of a sample of Black coal miners in 1987, in which respondents were asked whether they preferred undifferentiated majority rule or either of two "arrangements for multiracial power sharing."^[24] Although most of the sample preferred majority rule over power sharing, by an average of 27 percent, Zulu were below average, at 18 percent, and Xhosa were considerably above average, at 39 percent. This does not imply that Zulu are more conservative, for when a sample of Black industrial workers in seventeen locations was asked to choose a description of their feelings about the South African political situation, 62 percent of Zulu respondents, compared to 44 percent of Xhosa, 45 percent of Sotho, and 30 percent of other groups, chose the "angry and impatient" label.^[25] Controls for education, urbanization, age, or other variables did not change the results by ethnicity. It would not be amiss to conclude that Zulu, Xhosa, and perhaps Sotho have the highest levels of political consciousness, but that consciousness is expressed differently by the two largest groups.

An important qualification relates to location. Given an ethnic or Black alternative, the results vary in Soweto. The question was, "If all black people voted in South Africa, [name of respondent's group] would be small in number and other groups might have more power in government. Would people like you feel weak and insecure or do you agree with the statement 'blacks are all one people—it would not matter at all'?" Only 24 percent of Soweto Xhosa and 21 percent of Soweto Zulu took the "weak and insecure" option. (Zulu are by far the largest group in Soweto, about a third of its population.) Rural Xhosa, Eastern Cape Xhosa, and rural Zulu, however, selected that option at between 44 percent and 50 percent—a significant fraction given both the counter-ideological wording of the choice and the homogeneous (presumably secure) area in which those respondents lived. But 64 percent of Zulu migrant workers chose the "weak and insecure" response, rather than the "blacks are all one people" response.

[24] Hermann Giliomee and Lawrence Schlemmer, From Apartheid to Nation-Building (Cape Town: Oxford University Press, 1989), p. 167.

[25] Ibid., p. 168.

Organizational Affiliations and African Ethnicity

Another matter on which ethnic differences emerge relates to partisan and leadership preferences among Blacks. These, however, cannot be understood fully without understanding the ethnic composition of parties and movements.

It is, of course, well known that Inkatha is largely a Zulu organization centered on the nonindependent KwaZulu homeland. Led by Chief Mangosuthu Buthelezi, it is the lineal descendant of the Zulu National Congress, founded in 1928, but it has had wider ambitions. Increasingly, however, Inkatha seems unable to secure support from other homeland leaders and members of Black ethnic groups that gave Buthelezi strong support previously: Ndebele, Swazi, Tswana, and Venda.^[26]

Much less well known is the ethnically skewed leadership composition of two of the main panethnic extraparliamentary organizations, the African National Congress and the Pan Africanist Congress. In both cases, Xhosa speakers have been disproportionately represented at the top.

Despite its nonracial ideology, the ANC, as noted in Chapter 1, had no non-African members on its National Executive Committee until 1985. After 1985, it had two Whites, two Coloureds, and two Indians. Of the remaining twenty members, as of mid-1989, ten were Xhosa, five Tswana, four Pedi, and one Zulu. In short, fully half the Black leadership of the ANC was Xhosa-speaking, which means that Xhosa were overrepresented by a factor of about 2.2. Zulu were dramatically underrepresented, by a factor of more than 5.^[27]

The most conspicuous ANC leaders are also disproportionately Xhosa. Nelson Mandela, Walter F. Sisulu, Oliver Tambo, Govan and Thabo Mbeki, Chris Hani, and Pallo Jordan are all Xhosa speakers. Of the major future contenders for power, only Joe Modise and one or two others are not Xhosa. This imbalance and the factionalism related to it

[26] See Hanf et al., South Africa , p. 355.

[27] These calculations, and those that follow on the PAC, are mine. Data were supplied by several South African informants. I am particularly grateful to Pierre du Toit and W. J. Breytenbach for coordinating the data gathering and to Steven Wilkinson for double-checking ethnic affiliations from written and oral sources. Since Black ethnicity in politics is rarely discussed, and data on identity are hard to come by, it is possible that I have mislabeled the identity of one or two leaders, who, for example, may be products of mixed marriages. But the overall tendencies are undeniable.

have been subjects of concern within the ANC.^[28] There is thus support for the rather casual—and, even then, rarely stated—judgment that the ANC "has been dominated by Xhosas. . . ."^[29] Following coups in the two Xhosa "homelands," Transkei and Ciskei—the latter shortly after the release of Nelson Mandela from prison in 1990—the new military rulers made clear their alignment with the ANC.^[30]

The leadership composition of the PAC is roughly similar to that of the ANC. Taking its Central Committee and principal foreign representatives together, there were, as of mid-1989, a total of nineteen members, two of whom were Coloured and one of whom was Indian. The remaining sixteen were as follows: ten Xhosa, three Zulu, two Tswana, and one Venda. A clear majority was Xhosa—nearly a threefold over-representation. Xhosa are also prominent in AZAPO.^[31]

As I shall explain later, there are important historical reasons for Xhosa to be in the forefront of radical anti-regime activity. For now, the situation is unlikely to change dramatically within either of these movements.

After a long period of factional disputation, culminating in an upheaval in its Central Committee in 1987, the PAC settled into a more stable period, during which it increased its standing with foreign governments.^[32] Further upheaval would be very costly.

The ethnic balance of leadership in the ANC was also related to the balance of new recruits who moved across the border. This in turn was largely a function of developments inside South Africa, which provided waves of recruits. These waves were, in some degree, ethnically differentiated, depending on the location of the key events propelling the move into exile. After the Soweto rising of 1976, there was an exodus of youth from Johannesburg townships to ANC training camps. In the mid-1980s, there was another flow of recruits, this time disproportion-

[28] See Africa Confidential (London), January 12, 1990, p. 3; ibid., February 23, 1990, p. 4; ibid., April 20, 1990, p. 3.

[29] Graham Leach, South Africa: No Easy Path to Peace (London: Routledge & Kegan Paul, 1986), p. 111.

[30] See Argus (Cape Town), March 5, 1990.

[31] Of the AZAPO officers elected in 1990, seven were Africans, three of them Xhosa. The PAC has reportedly made a point of emphasizing its diverse composition, contrasting it with alleged Xhosa domination of the ANC. Africa Confidential , April 6, 1990, p. 8.

[32] Gary van Staden, "Return of the Prodigal Son: Prospects for a Revival of the Pan Africanist Congress" (paper presented at the Research Colloquium of the Political Science Association of South Africa, October 6–7, 1988). For the background, see Tom Lodge, Black Politics in South Africa since 1945 (London: Longman, 1983), pp. 305–17. See also Africa Confidential , October 16, 1985, p. 7.

ately from the Xhosa-speaking townships of the Cape Province. The Cape was the location of the school and rent boycotts that triggered the 1984–86 resistance movement. There the UDF also emerged most strongly.^[33] It has been asserted that rivalries within the second echelon of the ANC, particularly between Joe Modise and Chris Hani, involve to some extent their mobilization of, respectively, "the Jo'burgers" and other Transvaalers, on the one hand, and the newer militants from the Cape, on the other.^[34] In considerable measure, these would be ethnically differentiated support bases. How reliable such accounts are is open to question, but there seems little doubt that the influx of the 1980s was disproportionately Xhosa.

Against this background, Xhosa and Zulu responses to survey questions about identity are more comprehensible. When Xhosa respond that they are Black or South African, they are providing an answer very much in line with the officially declared policy of the ANC and the PAC, with either of which it is easy for Xhosa to identify, given their leadership composition. More than that, of course, such a response re-

[33] See Steven Friedman, "Black Politics at the Crossroads," South African Institute of Race Relations, PD1 / 86 (Johannesburg, January 2, 1986; mimeo.), p. 9.

[34] W. J. Breytenbach, The ANC: Future Prognosis , University of Stellenbosch Institute for Futures Research, Occasional Paper no. 4 (July 1989), pp. 24–26; Belinda Barrett, "A Profile of the African National Congress (ANC)" (Inkatha Institute of South Africa, May 1989; mimeo.), pp. 24, 60–63; Africa Confidential , August 12, 1988, pp. 1–3. See also Africa Confidential , July 6, 1983, pp. 1–4; ibid., December 11, 1985, pp. 1–5.

flects the exigencies of the present struggle—including the preeminence of race over ethnicity—and its obvious battle lines, the state versus the mainly Black resistance. When Zulu provide a different response, it presumably signifies that they are, for understandable reasons, less wholly identified with ANC or PAC positions and that there may be more than one struggle and more than one set of cleavage lines.

The same contrasting responses are revealed in surveys that tap support for various political organizations. Consider support for Buthelezi and Inkatha versus support for Mandela and the ANC among Zulu in Natal and non-Zulu (essentially, Xhosa) in the Cape, as revealed in the 1986 survey depicted in Table 3. The ANC seems to have about as much support among Zulu in the Transvaal as it does among non-Zulu there. But drop out the two middle columns, and the contrasts between Zulu and Xhosa are striking. For Zulu in Natal, the ANC has some resonance, but much less than Inkatha has. For Xhosa in the Cape, Inkatha is anathema; and the ANC is supported by a strong majority.^[35] Polarity is not too strong a word to use in characterizing these responses, which range from zero to 34 percent and from 19 percent to 62 percent.

The percentages just quoted show Buthelezi's support in decline, but this does not mitigate the Xhosa-Zulu difference. In the late 1970s, Buthelezi had more support in Natal, and the ANC had less in Natal. But Xhosa support was almost equally divided between Buthelezi and the ANC, as the 1977 comparison in Table 4 shows.

Although the respondent categories are not exactly the same, still the two tables together show that the exclusivity of Xhosa identification

[35] See also Martin Meredith, "The Black Opposition," in Jesmond Blumenfeld, ed., South Africa in Crisis (London: Croom Helm for the Royal Institute of International Affairs, 1987), p. 87.

with the ANC has grown dramatically in a decade. The distance between the Xhosa position and the Zulu position is, despite growth in ANC support in Natal, certainly no less than it was, particularly because Inkatha support has fallen off so completely in the Cape. In percentage-difference terms, the two groups were further apart in preferences in 1986 than in 1977. The average difference on each item was 36.0 points in 1977 and 38.5 in 1986. (The 1977 difference might have been larger if, instead of Durban Zulu, the respondent category had been Zulu in all of Natal.) And, as usual, the political preferences of every other African ethnic group—Tswana, Pedi, Sotho, Tsonga, Swazi, Venda, and Ndebele—fall in between those of the Durban Zulu and the Xhosa.³⁶

A survey conducted in 1984–85 by the Institute of Black Research at the University of Natal confirms the Zulu-Xhosa divergence and adds some further regional qualifications as well. A question about organizational support divided Africans into three cities and one township, each of which represented a major region: Durban (Natal), Langa (Western Cape), Johannesburg (Transvaal), and Port Elizabeth (Eastern Cape). Inkatha received 55.5 percent support in Durban but 3.1 percent in Johannesburg and virtually none anywhere else. The UDF had 51.0 percent in Port Elizabeth, 30.8 percent in Langa, 21.3 percent in Durban, and 12.7 percent in Johannesburg. The Eastern Cape was clearly the most pro-UDF area. The three Black Consciousness or Africanist groupings, AZAPO, the PAC, and the National Forum, were weak everywhere (with a combined percentage share of 6.8 percent in Port Elizabeth and Langa, practically no share in Durban) except the Johannesburg townships, where the three organizations received support of 15.6 percent, 15.2 percent, and 3.3 percent, respectively. ANC figures were lower than in other recent surveys—their high was 27.1 percent in Langa—but it is uncertain from the question whether respondents would have felt free to name a then-illegal organization as an appropriate response.^[37] Again, the difference between Natal and the Eastern Cape stands out above all other differences.

Leadership ratings are an indication of consensus or dissensus. In the

[36] With a single exception, the Pedi, who are slightly below even the Xhosa in their enthusiasm for Buthelezi, but only by 3 percentage points. On every other response relating to the ANC or Buthelezi, Zulu and Xhosa form the extremes within which other groups fall. See Hanf et al., South Africa , p. 355.

[37] Though some, as indicated, named the PAC, which was then illegal. Institute of Black Research, University of Natal, Durban, tables, based on a 1984–85 survey, from an untitled forthcoming manuscript, table 2. Some figures are subject to final correction.

Eastern Cape, Mandela, the Reverend Allan Boesak, and Bishop Desmond Tutu are all "admired" by at least 94.6 percent of respondents, with "disliked" ratings of less than 5.0 percent each. In Durban, although Mandela still leads the pack, admiration for these three declines to an average of 67.0 percent. In Durban, Buthelezi is "admired" by 76.0 percent; elsewhere, he scores much lower, and, by now predictably, lowest by far in Port Elizabeth, with only 6.3 percent positive ratings and 65.4 percent negative. To put the point in terms of ratios of "admired" ratings to "disliked" ratings: in Durban, Buthelezi's ratio is +4.3; in Port Elizabeth, it is -10.4.^[38]

Finally, a question asked how unfair laws could be changed. African responses varied enormously by city. In Durban, nine respondents out of ten chose "dialogue," "government reform," or "peaceful opposition." Only one in ten chose "extraparliamentary peaceful" or "violent" strategies. In Port Elizabeth, these preferences were turned upside down: more than nine out of ten chose the latter two responses (83.5 percent chose "extraparliamentary peaceful" and 10.2 percent "violent"). In Langa and in Johannesburg, responses were, once again, in-between, although Langa respondents tended to prefer softer options and Johannesburg respondents produced a 43.6 percent plurality for violence.^[39]

All of these data illustrate what Mark Orkin has called "a divided struggle," and they show with special sharpness how the Xhosa and Zulu heartlands anchor two ends of a spectrum. The data on which I have drawn in this section are derived from surveys conducted by researchers of varying political inclinations. In addition to a market research organization, the data come from the work of Theodor Hanf and his colleagues, Lawrence Schlemmer, Mark Orkin, and Fatima Meer and her colleagues—researchers who are otherwise often at odds. Some of them have fought survey wars over whether Blacks prefer foreign investment or divestment and over whether Buthelezi still has a strong support base.^[40] That their results are so congruent on ethnicity and political preference is, therefore, all the more significant.

The results are, moreover, congruent with long-standing patterns of

[38] Ibid., table 68b.

[39] Ibid., table 8.

[40] See, e.g., Mark Orkin, Disinvestment, the Struggle, and the Future: What Black South Africans Really Think (Johannesburg: Ravan Press, 1986; pamphlet), pp. 5–28, 37–44, 62–67; Fatima Meer and Alan Reynolds, "Sample Survey of Perceptions of the Durban Unrest—August 1985," in Fatima Meer, ed., Resistance in the Townships (Durban: Madiba, 1989), p. 276.

organization. The Eastern Cape is the historic heartland of ANC organization. Although it had nationwide objectives, the ANC's best-organized protest, the Defiance Campaign of 1952, drew nearly three-quarters of its participants from the townships of the Eastern Cape, especially around the two main cities of Port Elizabeth and East London.^[41] Port Elizabeth was, according to Tom Lodge, "consistently the strongest centre of ANC mobilization," and East London was "the second most important centre of the civil disobedience campaign in the country."^[42] By contrast, the Defiance Campaign "had only limited success among Africans in Natal. There, the ANC still had few branches. . . ."^[43] Although the ANC did manage to penetrate Natal in the next several years, no doubt aided by the election of a Natal Zulu, Albert Luthuli, as ANC national president in 1952, its following in the Eastern Cape has been consistently much stronger and better organized.

Underlying such patterns, and comparable patterns in Zimbabwe, Angola, Namibia, and Nigeria, where anti-colonial movements were divided along ethnic lines, is the intertwining of rational and affective elements in ethnicity. These elements affect followers as well as leaders, as Masipula Sithole has pointed out for Zimbabwe:

It is the masses who respond negatively or positively to tribal ideology, otherwise politicians would not use this resource. The masses calculate that they stand to benefit one way or the other from "our leader" or "our son," or "our homeboy," or indeed "our tribesman" in power, or representing "us" in the corridors of power. Indeed the tribesman politician might appeal to the masses in this way, but the response is not based on simple emotional false-consciousness; the masses do give thought to the promises made and the base of their credibility. It all makes rational sense to them in the same way they would rationalize an economic argument. A homeboy will deliver the goods.^[44]

That, despite such considerations, the extraparliamentary organizations have thus far achieved the degree of interethnic solidarity that they have

[41] Tom Lodge, "Political Mobilization during the 1950s: An East London Case Study," in Shula Marks and Stanley Trapido, eds., The Politics of Race, Class and Nationalism in Twentieth-Century South Africa (London: Longman, 1987), p. 318.

[42] Ibid., pp. 315, 323.

[43] William Beinart, "Worker Consciousness, Ethnic Particularism and Nationalism: The Experience of a South African Migrant, 1930–1960," in Marks and Trapido, eds., The Politics of Race, Class and Nationalism in Twentieth-Century South Africa , p. 298.

[44] Masipula Sithole, "The Salience of Ethnicity in African Politics: The Case of Zimbabwe," in Anand C. Paranjpe, ed., Ethnic Identities and Prejudices: Perspectives from the Third World (Leiden: E. J. Brill, 1986), p. 57.

is a measure of the overwhelming urgency of the struggle against apartheid.

Exogamy, Social Distance, and Stereotypes

So far, all we know is that the two largest groups, which are also disproportionately important in politics, occupy polar positions on some key questions of ethnic identity, ideology, organizational affiliation, leadership preferences, and strategic inclinations, and that one of these groups is significantly overrepresented and the other underrepresented in the leading extraparliamentary opposition organizations. Polarity may reflect or ripen into antipathy, but it is not necessarily the same thing.

The strongest case that ethnic differences will not ripen into recurrent ethnic conflict is made by examining data on Africans living in certain urban areas, above all Soweto. We have already seen that Soweto Zulu do not share a number of attitudes held by Durban Zulu. The passions against Inkatha and its leader can run rather deep among Zulu in Soweto. But a broader point can be made about Soweto. Africans in Soweto tend to hold quite liberal views toward ethnic intermarriage and toward ethnic differences in general.^[45]

Historically, around Johannesburg, interethnic marriage has been a variable rather than a constant. As early as the 1930s, Ellen Hellmann found that nearly half of all marriages in Rooiyard, a small slum area in a Johannesburg suburb, were interethnic.^[46] In the early 1950s, however, a much more wide-ranging and rather careful local government survey showed exogamy to be decidedly a minority trend in several townships on the west side of Johannesburg. Most people then married not only within their own language group but within their own subgroup as well. The exceptions concerned some specific patterns of intermarriage: Pedi men marrying Tswana women, Swazi men marrying Zulu women, and Shangaan marrying members of several other groups. For the rest, endogamy was practiced with considerable consistency.^[47] Even today, exact rates of marrying out are hard to come by. As William

[45] Hanf et al., South Africa , pp. 339–41.

[46] Ellen Hellmann, Rooiyard: A Sociological Survey of an Urban Native Slum Yard , Rhodes-Livingstone Papers, no. 13 (Cape Town: Oxford University Press, 1948), p. 13.

[47] See "Report of a Sample Survey of the Native Population Residing in the Western Areas of Johannesburg" (Non-European Affairs Department, City of Johannesburg, 1951; mimeo.), pp. 74, 78. For some later West African comparisons, see Margaret Peil, Cities and Suburbs: Urban Life in West Africa (New York: Africana, 1981), pp. 149–51.

Beinart says, they have been "little studied."^[48] Nevertheless, within a generation, attitudes toward intermarriage have changed. Estimates put exogamy in Soweto in the range of one-third to one-half of all marriages.^[49] Melville Leonard Edelstein's Soweto sample in 1971 consisted of elite pupils, 42 percent of whose parents had married exogamously.^[50] These are very high rates indeed, but not incompatible with the persistence of ethnic identities.^[51] Similarly, the extent to which actual political behavior reflects norms of interethnic accommodation is an open question. In precept, at least, "exclusive tribal patriotism seems to have almost died in Soweto," replaced by an ideology of the melting pot: "We are all Africans here, suffering from the same malady—the injustice of Government."^[52] Although the results were not uniform, Edelstein's study showed that children of exogamous marriages generally displayed slightly less social distance toward racial and ethnic out-groups than did children of endogamous marriages.^[53]

Even in Soweto, there has been lethal behavior at variance with these norms, as we shall see later; and, as of the early 1970s, the various groups in Soweto held significantly more favorable in-group than out-group stereotypes and associated more frequently with in-group than with out-group members.^[54] Outside of Soweto, social distance studies show significant amounts of ethnocentrism and some patterns of intergroup hostility. As exogamy is understudied, so are ethnic attitudes among Africans,^[55] but they are far from being wholly unstudied.

In the Johannesburg township of Dube (part of Soweto), Zulu, Sotho, and Xhosa were asked to rank several groups on an adjectival scale representing a wide variety of desirable personal attributes.^[56] Groups

[48] "Worker Consciousness, Ethnic Particularism and Nationalism," p. 295.

[49] A. A. Dubb, "The Impact of the City," in W. D. Hammond-Tooke, ed., The Bantu-speaking Peoples of South Africa (London: Routledge & Kegan Paul, 1974), p. 465.

[50] Melville Leonard Edelstein, "An Attitude Survey of Urban Bantu Matric Pupils in Soweto, with Special Reference to Stereotyping and Social Distance: A Sociological Study" (M.A. thesis, University of Pretoria, 1971), p. 108.

[51] See Donald L. Horowitz, "Conflict and Accommodation: Mexican-Americans in the Cosmopolis," in Walker Connor, ed., Mexican-Americans in Comparative Perspective (Washington, D.C.: Urban Institute, 1985), pp. 81–84.

[52] Mayer, "Class, Status, and Ethnicity," pp. 152, 153.

[53] Edelstein, "An Attitude Survey of Urban Bantu Matric Pupils in Soweto," pp. 110, 130–31.

[54] Melville Leonard Edelstein, What Do Young Africans Think? (Johannesburg: South African Institute of Race Relations, 1972), p. 116.

[55] See J. W. Mann, "Attitudes towards Ethnic Groups," in Adam, ed., South Africa: Sociological Perspectives , p. 66.

[56] Dubb et al., "African Attitudes in Town."

to be ranked included the respondent's own, the two other African groups, and Whites. Majorities of all the respondent groups had either been born in town or, if not, had scant contact with rural birthplaces. As expected, respondents in each group ranked their own group highest. Both Sotho and Xhosa ranked Whites higher than they ranked Zulu; Xhosa ranked Whites above both Zulu and Sotho. Zulu ranked Whites last, but, for Zulu respondents, there was not much to choose. The three groups, apart from the subjects' own, formed a "compact set."^[57] For Zulu and Sotho, the distances between subject groups and target groups were not very great, but Xhosa respondents emerged in this study as more ethnocentric. They evaluated other groups "a good deal less favourably than themselves. . . ."^[58]

Subjects in a nationwide study of positive and negative qualities, conducted under the auspices of the Human Sciences Research Council (HSRC), were Afrikaans-speaking Whites, English-speaking Whites, Coloureds, Sotho, and Zulu.^[59] The instrument employed semantic differential scales.^[60] Respondents were asked to characterize groups in terms of 32 sets of adjective antonyms, such as courteous or rude, kind or cruel, aggressive or submissive . Differences in group evaluations are revealed by degree of agreement among subjects on positive and negative adjectives that characterize a given target group. These evaluations were based on research from the early 1970s to the early 1980s. Over this time span, the intergroup rankings are reported to have been quite consistent. A graphic depiction of the values over time, for all subject and target groups, is displayed in Figure 1.

From the figure, it is obvious that all respondent groups had the most positive views of themselves. The greatest spread in evaluations of other groups was displayed by Afrikaners. While they evaluated themselves only modestly favorably, Afrikaners held less positive views of Blacks and Coloureds than any other subjects in the sample displayed toward any group. The range of difference between in-group evaluations and out-group evaluations displayed by English-speaking White and by Coloured subjects was markedly narrower. The range of evaluations by Zulu respondents was in-between, but the evaluative range exhibited by

[57] Ibid., p. 89.

[58] Ibid., p. 90.

[59] HSRC Investigation into Intergroup Relations, The South African Society: Realities and Future Prospects (Pretoria: Human Sciences Research Council, 1985), pp. 81–82.

[60] On the logic of semantic differentiation, see Charles E. Osgood et al., The Measurement of Meaning (Urbana: University of Illinois Press, 1957).

Figure 1.
Average Semantic Differential of Own and Other Groups in South Africa, 1973–82
SOURCE:
HSRC Investigation into Intergroup Relations,  The South African Society: Realities and Future Prospects
(Pretoria: Human Sciences Research Council, 1985), p. 82. © HSRC 1985. Reproduced by permission. 
NOTE:
Sample sizes varied with respondent group and year, but were very large. For example, the Sotho sample in 1979
consisted of 1,199 respondents; the Zulu sample in the same year consisted of 1,168 respondents.

Sotho was close to that shown by Afrikaners. Both Sotho and Zulu gave English-speaking Whites the most positive evaluations, second only to their own, and both gave Afrikaners and Coloureds the least positive. English-speaking Whites, however, did not reciprocate the positive views held of them by Sotho and Zulu. For Sotho, Zulu and Indians ranked in the middle; both of these groups were placed by Sotho respondents at a substantial distance from the positive evaluations by Sotho subjects of their own group. For Zulu, Sotho and Indians occupied the middle ground, not far behind Zulu evaluations of English-speaking Whites. As in the study of Dube, out-group evaluations, on the whole, were not powerfully negative. But it is very clear that Black South Africans are as keenly aware of ethnic differences as of racial differences.

Perhaps the most important findings are the sharp evaluative distinctions made by Zulu and Sotho respondents between Afrikaans and English speakers—a result found in other studies of Black respondents^[61] —the least positive evaluations accorded by those respondents to Afrikaners and Coloureds, and the fact that both Zulu and Sotho evaluated each other less positively than either evaluated the English speakers. This last, despite the prevalence of color discrimination in the society.

The negative attributes imputed to Afrikaners by majorities of Sotho and Zulu are politically relevant ones. The adjectives used by Sotho were snobbish, unjust, fickle, and stubborn; those used by Zulu were stubborn, aggressive, prejudiced, and quarrelsome.^[62] In an earlier Soweto social distance study of four African respondent groups, Afrikaners were uniformly seen as the most distant group, as we shall see.

In KwaMashu, a satellite town outside Durban, Zulu respondents were asked a variety of social distance and stereotype questions.^[63] The vast majority of the respondents had been born in town or had lived there for more than twenty years. Social distance scales gauge attitudes

[61] See, e.g., Nieuwoudt and Plug, "South African Ethnic Attitudes: 1973 to 1978," pp. 168–69.

[62] S. J. Kruger, Sienswyses wat Suid-Sotho's oor hulself en ander Suid-Afrikaanse bevolkingsgroepe gehuldig het, 1979–1982 (Pretoria: Human Sciences Research Council, 1984), p. 21; S. J. Kruger, Sienswyses wat Zoeloes oor hulself en ander Suid-Afrikaanse bevolkingsgroepe gehuldig het, 1979–1982 (Pretoria: Human Sciences Research Council, 1984), p. 19. The three target groups eliciting the largest number of majority responses (Sotho, Zulu, and English) also received higher percentages of agreement on group characteristics than did the remaining target groups. The same is true for Zulu evaluations of English speakers and Sotho versus Zulu evaluations of Afrikaners and Coloureds. I am indebted to G. A. Thiele for making these reports available to me.

[63] Brian M. du Toit, "Ethnicity, Neighborliness, and Friendship among Urban Africans in South Africa," in Brian M. du Toit, ed., Ethnicity in Modern Africa (Boulder, Colo.: Westview Press, 1978), pp. 143–74.

toward others by measuring the willingness of respondents to join with those others in activities involving varying degrees of intimacy, from occasional visits to residence in the same neighborhood, to friendship, to work relationships, and to marriage. As intimacy increased in the activity postulated by the question, so did reluctance to engage in that activity with ethnic strangers. For these respondents, as for other Zulu in Durban townships,^[64] marriage with a member of any out-group was regarded as nearly unthinkable.^[65] The groups that seem to respondents to be most distant from them, by far, were Indians, Whites, and Coloureds (in that order).^[66] Fairly elaborate stereotypes were held of all the out-groups. Strongly negative characterizations were provided for the most distant African groups. Ndebele were "backward," Tsonga were "loose-living," and Xhosa were "clever" people who would make the Zulu feel "lost" and "inferior" or would cheat them.^[67] In fact, "Zulu really get their ire up when they refer to the Xhosa," whom they described as "too clever," "crooks," "scheming," "very cunning," "crafty," and "stiffheaded."^[68] Among males, the most highly educated respondents displayed the greatest social distance.^[69]

Although KwaMashu, like other Durban townships, is overwhelmingly Zulu in composition, it contains non-Zulu minorities, among whom Xhosa speakers are predominant. The articulated stereotypes, then, are not based on an absence of intergroup contact. Some of the contact is not particularly friendly, as we shall see.

The reliability of such findings can be gauged by matching them against what other studies find. For all the differences between Durban and Soweto, young, educated Zulu in Soweto display roughly the same rank order of distance from various groups as those in Durban do.^[70] In Soweto, as in Durban, Xhosa and Shangaan are seen by Zulu as somewhat distant, while Swazi are seen as close. In Soweto, both Zulu and Xhosa agree, as do all categories of respondent, that Zulu are "hardworking,"

[64] See Hanf et al., South Africa , p. 341.

[65] Du Toit, "Ethnicity, Neighborliness, and Friendship among Urban Africans in South Africa," pp. 151–52.

[66] Ibid., p. 153.

[67] Ibid., pp. 158, 153.

[68] Ibid., p. 158.

[69] Ibid., p. 161. So did the oldest respondents, but below age 54 little difference was observable. Among females, the most highly educated respondents—those with tertiary education—displayed the least social distance, but the numbers in that cell were very small. Ibid., pp. 162, 149.

[70] Edelstein, What Do Young Africans Think? p. 106.

while Xhosa, the most educated group in Soweto, are "intelligent."^[71] And, finally, Soweto Xhosa do not reciprocate Zulu feelings of distance; on the contrary, they see themselves as close to Zulu, Tswana, and Sotho and—like Soweto Zulu, only more so—most distant from Coloureds, Indians, and Whites.^[72]

It is worth a closer look at the Soweto social distance study. The data, gathered in 1971, are rather old, but they are quite consistent (especially in the Zulu and Sotho responses) with the HSRC data, which were gathered into the 1980s. Moreover, the Soweto respondents were then matriculation pupils—those qualifying for some form of higher education and, therefore, future elites—in their teens.

Where respondents are asked to scale the distance of a variety of both ethnic and racial groups, one can ask whether other racial groups are seen to be more distant from the respondent than other ethnic groups are. In South Africa, the answer to this is generally "yes." In this respect, as we have seen, the results are different from stereotype studies, which often show more favorable evaluations of at least some other racial groups. But a further question can be asked: Are the groups ordered continuously or discontinuously? Does the distance separating the cluster of ethnic groups from the cluster of racial groups indicate a great gap between the two? An answer to this question for South African Black elites lurks in Edelstein's Soweto data. As one might have hypothesized from the Dube township and nationwide HSRC studies, the results vary by the ethnicity of the respondent, even in Soweto.

Edelstein breaks the results out for four groups of respondents: Zulu, Xhosa, Tswana, and Sotho. Figure 2, which diagrams his results, shows that Zulu and Sotho see other ethnic and racial groups as being placed at various points on a single continuum. For Zulu, it is a narrower continuum overall than for Sotho, who, in the HSRC study, also spread their evaluations out more than Zulu did. Tswana create a bit more distance between the furthest African group and the nearest non-African group than either Zulu or Sotho do. For Tswana respondents, there are at least three clusters: the three African groups they see as being very close to them, then the four more distant African groups, followed by the non-African groups. Xhosa also group Africans into two clusters—the extremely close and the fairly close—but, unlike Tswana, they

[71] Ibid., pp. 107, 110.

[72] Ibid., p. 106.

Figure 2.
Social Distance Evaluations of South African Ethnic Groups by Soweto Matriculation Pupils, 1971 ( N  = 200)
SOURCE:
Based on data in Melville Leonard Edelstein, "An Attitude Survey of Urban Bantu Matric Pupils in Soweto, with Special Reference to
Stereotyping and Social Distance: A Sociological Study" (M.A. thesis, University of Pretoria, 1971), p. 132.
NOTE:
On the scale from 1 to 5, 1 denotes least distance and 5 denotes greatest distance from the respondent.

place the non-Africans in a completely different category. For Xhosa, there is a great gap between Africans and non-Africans. For all four respondent groups, the ranking of the non-Africans is close to the same: Afrikaners are furthest by far, English and Coloureds are closest. But the average distance varies enormously by respondent group, as Table 5 shows.

Edelstein concludes that while "Zulu respondents associate more readily with Venda than [with] South African English there is not a great difference between their attitudes to these two outgroups," and he makes the same point for the proximity of the least favored African and most favored non-African group for Sotho. (In all cases, the most favored non-Africans are either Coloured or English.) But Xhosa, he notes, place fully 1.4 points on a scale of only 1.0 to 5.0 between the furthest African group and the nearest non-African groups.^[73] And, as Table 5 shows, Xhosa mean distance ratings are markedly different from those of other respondents.

All of this is really another way of saying that, for Xhosa respondents, the definition of the situation seems different. For them, racial differences are so much greater that they are discontinuous from intra-African ethnic differences. The location of Afrikaners on the Xhosa response scale is truly extraordinary, and this is before the Soweto rising of 1976. The 4.6 distance rating on the scale, together with the overall bimodal pattern of Xhosa responses, indicates clearly that, for Xhosa, racial antipathy supersedes ethnic antipathy. When Xhosa provide a nonethnic response to an identity question on a sample survey, they are

[73] Edelstein, "An Attitude Survey of Urban Bantu Matric Pupils in Soweto," pp. 114, 116, 115, respectively.

reflecting their view of South African society. Given this view, it is no wonder that Xhosa are so strongly identified with militant movements.

In several of these studies, taken together, are hints of some sources of the different political alignments of Xhosa and Zulu. The antipathy toward Xhosa, coupled with fear of Xhosa craftiness, expressed in the KwaMashu study, suggests reasons why Natal Zulu might steer clear of organizations that seem to be controlled by Xhosa. The significantly greater distance between Xhosa subjects and Whites than between Zulu subjects and Whites may mean that, for Xhosa, the struggle for racial equality outweighs African ethnic differences to a greater degree than it does for Zulu. The frequent finding that Africans of various groups believe at least some Whites to possess more desirable attributes than those possessed by Black groups other than their own—and this finding holds even for Xhosa in the Dube township study—suggests rather strongly that there is potential for intra-African conflicts to supersede Black-White conflicts at some stage, for some purposes.

Against these findings, it is easier to understand the variety of settings in which Black interethnic attitudes crystallize. In Soweto and possibly a few other heterogeneous townships, the initial residential segregation by ethnic group has broken down; and, although "ethnic categories do still have considerable resonance in daily interaction," nevertheless, intra-African ethnocentrism, while not absent, meets with social disapproval.^[74] In more homogeneous urban or rural areas, a good deal of ethnocentrism is expressed,^[75] but it might or might not find a behavioral outlet. In heterogeneous areas composed of migrants from several home areas, the setting is conducive to ethnic exclusiveness. Most well documented are mines and factories where migrants comprise a multiethnic work force.^[76] Housing is often ethnically segregated, social life is largely conducted within ethnic groups, and intergroup hostility is in evidence. In such settings, the potential for interethnic violence is great,

[74] Mayer, "Class, Status, and Ethnicity," p. 152.

[75] See, e.g., Philip M. Mayer, Townsmen or Tribesmen? 2d ed. (Cape Town: Oxford University Press, 1971), pp. 40-41.

[76] See, e.g., J. K. McNamara, "Brothers and Work Mates: Home Friend Networks in the Social Life of Black Migrant Workers in a Gold Mine Hostel," in Philip M. Mayer, ed., Black Villagers in an Industrial Society (Cape Town: Oxford University Press, 1980), pp. 305–40; Dunbar Moodie, "Mine Culture and Miners' Identity on the South African Gold Mines," in Belinda Bozzoli, ed., Town and Country in the Transvaal: Capitalist Penetration and Popular Response (Johannesburg: Ravan Press, 1983), p. 185; Philip Bonner and Rob Lambert, "Batons and Bare Heads: The Strike at Amato Textiles, February 1958," in Marks and Trapido, eds., The Politics of Race, Class and Nationalism in Twentieth-Century South Africa , pp. 341–48; Beinart, "Worker Consciousness, Ethnic Particularism and Nationalism," pp. 292–93.

and indeed it frequently occurs, sometimes with considerable loss of life. In 1985, for example, 119 people were killed in fighting between Xhosa and Sotho miners at a gold mine.^[77]

Attitudinally, the Soweto rising of 1976 was a watershed in Black-White relations. There is good evidence that White attitudes toward Blacks hardened and Black hostility to Afrikaners in particular increased substantially after Soweto.^[78] Even in the early 1970s, however, as we have seen, interracial social distance studies among African respondents showed Afrikaners as the most distant group and English speakers (or, occasionally, Coloureds) as the closest, with Indians and Coloureds generally at various points in between.^[79] White attitudes toward Indians, and Afrikaner (but not English) attitudes toward Coloureds as well, grew more favorable after Soweto.^[80] This may reflect an implicit economy of antipathy (it is difficult to be equally hostile to all groups simultaneously) or contrast effects resulting from the nonparticipation of Indians and Coloureds in the main events of 1976.^[81] What we do not have are complete measures over time for rising and declining intra-African ethnic attitudes. Soweto may have had a slight softening effect on them as well—it certainly helped close the attitudinal gap between English- and Afrikaans-speaking Whites.^[82] As long as White-Black conflict is preeminent, there are limits to intra-White ethnic differences. For the time being, presumably, there are similar limits to intra-Black ethnic antipathy.

Ethnicity, Violence, and "Homeland" Politics

Despite the obvious binding force of White rule on Black cohesion, modern South Africa has experienced a great deal of intra-Black vio-

[77] Carole Cooper et al., Race Relations Survey, 1985 (Johannesburg: South African Institute of Race Relations, 1986), p. 169. For other accounts of violence, see the works cited in note 76, above.

[78] Nieuwoudt and Plug, "South African Ethnic Attitudes: 1973 to 1978."

[79] H. G. Viljoen, "Stereotipes en sosiale afstand as dimensies van ethniese vooroordeel by Bantoe- en Indiërstudente" (Ph.D. diss., University of Pretoria, 1972), pp. 131, 137, 143; Edelstein, "An Attitude Survey of Urban Bantu Matric Pupils in Soweto," pp. 113, 132.

[80] Nieuwoudt and Plug, "South African Ethnic Attitudes: 1973 to 1978," p. 167.

[81] For the perceptual basis of contrast effects, see Muzafer Sherif and Carl I. Hovland, Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Change (New Haven: Yale University Press, 1961); H. Tajfel, "Quantitative Judgement in Social Perception," British Journal of Psychology 50, pt. 1 (February 1959): 16–29.

[82] Nieuwoudt and Plug, "South African Ethnic Attitudes: 1973 to 1978," pp. 168–69.

lence, some of it between competing Black extraparliamentary organizations and some between Black ethnic groups.^[83]

When the multiracial United Democratic Front was formed in 1983 to oppose the new Indian and Coloured legislative chambers, it immediately came into conflict with the Azanian People's Organization.^[84] The UDF and AZAPO saw themselves as rivals for support in the Black townships, especially in the Eastern Cape. Up through 1987, there was violence between the two, with the UDF generally emerging triumphant. Violence between Africanists and Charterists continued sporadically thereafter, accelerating in 1990, when the ANC clashed with the PAC as well as with AZAPO.^[85]

The balance of power was different in Natal, where UDF efforts to expand met with armed resistance from Inkatha. With its strength especially concentrated in the Cape, the UDF was surely seen by some in Natal as a non-Zulu movement.^[86] From 1986 through 1989, more than 2,000 people were killed in the townships around Pietermaritzburg, which had become the center of uncontrollable violence. From 1986 to 1988, 3,000 houses were burned, and 30,000 people were left homeless. More of the victims were UDF supporters than Inkatha supporters. In some cases, the police appeared more sympathetic to Inkatha, to judge by police inactivity, the appointment of some Inkatha activists as auxiliary constables, and the pattern of arrests.^[87] Interspersed in the Natal inci-

[83] For present purposes, I leave aside violence in 1985–86 at the Crossroads settlement outside Cape Town between so-called Witdoeke (white cloths, after strips of cloth by which they identified themselves) and generally younger comrades. With the apparent connivance of the security forces, a very large number of homes were destroyed. Some of those left homeless moved to Kaiyelitsha, a Black settlement near Stellenbosch. For an account of the violence, see Lawyers Committee for Human Rights, Crisis in Crossroads (New York: Lawyers Committee for Human Rights, 1988). For comparable episodes in the Eastern Cape, see Cooper et al., Race Relations Survey, 1987/88 , pp. 47–49.

[84] See Barney Mthombothi, "Introduction," in Meer, ed., Resistance in the Townships , pp. 5–6; Colin Legum, ed., Africa Contemporary Record, 1986–87 (New York: Africana, 1987), p. B744.

[85] See Carole Cooper et al., Race Relations Survey, 1986 (Johannesburg: South African Institute of Race Relations, 1987), pt. 1, pp. 142–43; Cape Times (Cape Town), February 14, 1990.

[86] See, e.g., the letter to the editor of Argus , April 26, 1990.

[87] The Pietermaritzburg violence has been well documented in Wendy Leeb, "Death, Devastation and Destruction—Refugees in Natal" (Pietermaritzburg: Centre for Adult Education, University of Natal, 1989; unpublished paper); J. W. W. Aitchison, "The Pietermaritzburg Conflict—Experience and Analysis" (Pietermaritzburg: Centre for Adult Education, University of Natal, July 7, 1989; unpublished paper); J. W. W. Aitchison, "Numbering the Dead: Patterns in the Midlands Violence" (Pietermaritzburg: Centre for Adult Education, University of Natal, 1988; unpublished paper). On the so-called Kitsonkonstabels , see Colin Legum and Marion E. Doro, eds., Africa Contemporary Record, 1987–88 (New York: Africana, 1989), p. B732.

dents was also some violence between UDF and Black Consciousness youth.

As the violence spread in Natal in the late 1980s and in 1990, it became more difficult to decompose its elements. In some areas, political violence was intertwined with conflict between old and young, migrants and residents, extended families, Zulu subgroups (Iziwe ), Zulu tribal authorities and their opponents, and local gangs.^[88]

It is difficult to say whether organizational rivalries have any clear ascriptive underpinnings within Zulu society or elsewhere. There is no doubt that subgroup divisions are occasional sources of violent conflict. In 1978, 260 people were killed in clashes between Madondas and Majolas in KwaZulu; and, in 1984, 70 people were killed in comparable fighting in Natal.^[89]

Much clearer have been episodes of outright interethnic violence. In 1976, during the Soweto revolt, Zulu migrants emerged from their hostel and killed a good many youth on the streets.^[90] At the end of 1985, Zulu outside Durban attacked Xhosa-speaking Pondo migrants who worked in the sugar industry, killing many and driving the rest out.^[91] Around the same time, homeland boundaries provoked violence—first in 1985 between Shangaan and Pedi over the boundary between Gazankulu and Lebowa^[92] and then in early 1986 between Ndebele and Pedi over a South African government plan to incorporate the largely-Pedi Moutse area outside Pretoria into KwaNdebele.^[93] Finally, in 1990, there was widespread violence between Zulu hostel dwellers, and Xhosa in Transvaal townships, resulting in many, many deaths.

To be sure, some of this violence appears to have been manipulated or encouraged by state authorities, and their role was even more direct when they used Zulu or Tswana police to remove Xhosa squatters, as

[88] Interviews, Botha's Hill, Natal, August 3, 1989.

[89] For the 1978 events, see Ernest Harsch, South Africa: White Rule, Black Revolt (New York: Monad Press, 1980), p. 67. For the 1984 violence, see Leach, South Africa , p. 111. For similar events, see Cooper et al., Race Relations Survey, 1985 , pp. 309, 318.

[90] Times (London), August 26, 1976; Colin Legum, ed., Africa Contemporary Record, 1976–77 (New York: Africana, 1977), pp. B792–93.

[91] For a helpful account, see "Umbumbulu," in Meer, ed., Resistance in the Townships , pp. 165–76; see also Los Angeles Times , December 27, 1985. For the flavor of Pondo-Zulu antipathy in the Natal cane fields, see Beinart, "Worker Consciousness, Ethnic Particularism and Nationalism," pp. 289, 293.

[92] "Umbumbulu," p. 166. For the background, see Harries, "Exclusion, Classification and Internal Colonialism," pp. 106–07. For an account of the dispute, see Cooper et al., Race Relations Survey, 1985 , p. 284.

[93] Financial Times (London), January 3, 1986; Los Angeles Times , January 3, 1986. For background, see Cooper et al., Race Relations Survey, 1985 , pp. 284–86.

they did, with lethal results, in Crossroads in 1985.^[94] But ethnic strangers were sent in for police duty precisely because it was known they would be reliable. The fact that ethnic affiliations were available for manipulation or encouragement suggests that ethnic violence is not just the product of the state's action in setting one group against another but reflects the continuing importance of ethnicity.

A similar point can be made about the so-called African homelands. They are creations of the apartheid regime, but they nonetheless reveal some important patterns of ethnic politics.

As I indicated earlier, a contextual view of ethnicity suggests that intra-African divisions would be muted while White-Black conflicts are preeminent but that they would become salient as White-Black conflicts recede in importance. Universal suffrage is likely to produce just such a change in context. To appreciate the increased role of ethnicity in these circumstances, a glimpse at politics in the various homelands is instructive.

Although the homelands were designed for each of the major African ethnic groups, several homelands contain significant minorities. In the Pedi homeland, Lebowa, Ndebele form a minority. In turn, Pedi are a large minority in the Tswana homeland, Bophuthatswana. There are Swazi in KwaZulu and Sotho in the two Xhosa Bantustans, the Transkei and Ciskei. In none of these have minorities been politically well represented. Early on, leaders of the dominant group generally did the equivalent of what the chief minister of Bophuthatswana did—namely, insist on a strongly Tswana identity for the territory.^[95] That, after all, was the ostensible raison d'être for the homeland. There was also a general suppression of party competition at an early stage in homeland development. In a single-party context, it was easy to neglect ethnic strangers, prevent them from securing legislative representation, and keep them out of the cabinet.

In most homelands, subethnic rivalries have also played a major role in politics. In the Ciskei, the long-standing rivalry between two major Xhosa-speaking subgroups formed the leitmotiv of political conflict. The Rharhabe, who fought the Europeans for more than a century, were

[94] See Lars Waldorf, "Life in Crossroads," New Republic , August 25, 1986, pp. 17–19. See also "Affidavits—Orange Free State," in Meer, ed., Resistance in the Townships , p. 127, reporting that township councillors in Thaborg in the Orange Free State are protected by vigilantes "drawn from migrants, thugs in the township and some Zulu people from Natal."

[95] See D. A. Kotzé, African Politics in South Africa, 1964–1975 (New York: St. Martin's Press, 1975), pp. 47–48.

aligned against the Mfengu, who had migrated to the Ciskei 150 years earlier. The political dispute in the Ciskei reflects a variety of underlying group-based differences.^[96] Comparable subgroup or lineage divisions are to be found in other homelands.

In Bophuthatswana, both ethnic and subethnic divisions are simultaneously important. In 1988, the precariously placed chief minister, Lucius Mangope, announced that he would sooner merge the homeland into neighboring Botswana than reincorporate it into South Africa in accordance with the views of the extraparliamentary opposition. After the legalization of the ANC and the PAC in 1990, demonstrations demanding reincorporation into South Africa were put down by the homeland police and army. Non-Tswana were particularly fearful of a connection with Botswana, with its overwhelming Tswana majority. In addition, Mangope had excluded the Bafokeng subgroup, located in the east of the homeland, from any real power, and that triggered coup attempts, clashes with villagers resisting integration into the homeland, and an alignment between the regionally based dissenters and the ANC.^[97]

South Africans have tended to think in terms of fixed rather than fluid categories and alignments. What homeland politics shows is that there is a substructure of allegiances and divisions available for activation when a new context brings African politics into the foreground.

The History of Racial and Ethnic Encounters

In many African colonies, the encounter between colonizers and colonized involved warfare during the initial conquest, the later "pacification," and the periodic suppression of rebellion. The dislocations caused by colonial occupation also produced population movements and conflicts among the colonized peoples themselves and did not necessarily prevent the continuation of warfare begun before the colonial occupation.

The patterns of colonial warfare, however, were varied. In Nigeria, a southward march by Fulani was interrupted by the British arrival.

[96] See "Ethnicity and Pseudo-Ethnicity in the Ciskei," in Vail, ed., The Creation of Tribalism in Southern Africa , pp. 395–413; C. W. Manona, "Ethnic Relations in the Ciskei," in Nancy Charton, ed., Ciskei: Economics and Politics of Dependence in a South African Homeland (London: Croom Helm, 1980), pp. 97–121; M. G. Whisson and C. W. Manona, "Maqoma and Ciskeian Politics Today," in Charton, ed., Ciskei , pp. 214–27. Cf. David J. A. Edwards, "Political Identity among South African Blacks in and near a Contemporary Homeland," International Journal of Psychology 22, no. 1 (February 1987): 39–55.

[97] See Africa Confidential , February 23, 1990, p. 6.

Later protected by the British, Hausa-Fulani were slow to join the independence movement. Conflict between them and southern ethnic groups occurred, but it had little to do with pre-colonial warfare.^[98] The same was not true in Gabon, where the Fang were on the move, conquering and absorbing other groups on their way to the coast, when the French stopped their advance. Rivalry between Fang and coastal peoples continued during the colonial period. After independence, the Fang, seen by others as domineering and ethnocentric, assumed leadership roles in every area of Gabonese life.^[99] In Zaire, a series of revolts by Lulua against Belgian rule began in the 1890s and lasted up to the 1940s. With much of the countryside decimated, non-Lulua fled to the towns. By 1930, Baluba constituted a majority of the urban population. As the Belgians departed, Lulua hostility was redirected toward Baluba, who in 1959 were victims of massive violence.^[100] Sometimes interethnic warfare was not resumed, sometimes it was, and sometimes anti-colonial violence was channeled into ethnic conflict.

Warfare in South Africa was unusually bitter and protracted. The first colonists at the Cape owed allegiance to the Dutch East India Company, which had a skeletal organization and a limited capacity to reach into the territory. As the frontier grew away from the Cape peninsula, the company had little impact. It was nearly two centuries before the British consolidated their control over South Africa. During this interval and even after, there was warfare between Afrikaners on the move northward and the Africans they encountered along the way, as well as warfare among the Africans themselves. Until the British consolidated their hold, there was no single overwhelming force to crush opposition and impose a colonial peace.^[101]

In the Cape, as the Boers moved eastward toward the Fish River, they encountered the Xhosa. A series of raids back and forth matured into hostilities that lasted on and off for more than a century. Officially, nine wars were fought between Whites and Xhosa in the nineteenth century. As late as 1920, a strong Xhosa uprising was put down by the White government.

[98] See C. S. Whitaker, Jr., The Politics of Tradition: Continuity and Change in Northern Nigeria, 1946–1966 (Princeton: Princeton University Press, 1970).

[99] See Brian Weinstein, Gabon: Nation-Building on the Ogooué (Cambridge: MIT Press, 1966), pp. 32–69.

[100] See Thomas Turner, "Congo-Kinshasa," in Victor A. Olorunsola, ed., The Politics of Cultural Sub-Nationalism in Africa (Garden City, N.Y.: Anchor Books, 1972), pp. 218–24.

[101] See, e.g., T. R. H. Davenport, South Africa: A Modern History , 3d ed. (Toronto: University of Toronto Press, 1987).

Zulu, meanwhile, were in an expansionist phase in the early nineteenth century. Having undergone considerable subgroup amalgamation and centralization, the Zulu moved out in every direction, in a massive dispersion often called the mfecane , subjugating a variety of groups—Swazi, Sotho, and Pondo—and causing many to migrate from their home areas, including those Ndebele who fled to what is now Zimbabwe. Only the British occupation of Natal put an end to Zulu expansion.

It would be inaccurate to attribute all or most intra-African differences to this history, but the unusual turbulence of the eighteenth and nineteenth centuries certainly has contemporary reverberations. There is still, for example, a martial spirit revived from time to time among the Zulu. In 1949, when Zulu attacked Indians in Durban, the attackers were organized into impis , or regiments.^[102] Relations between Mfengu and Rharhabe Xhosa are colored by the fact that, although the Mfengu fled Natal from the mfecane in the early nineteenth century and found shelter among Xhosa, they then proceeded to ally with Whites against Xhosa in the various frontier wars from 1834 to 1879, for which they were well rewarded. The persistence of Zulu vocabulary in the Mfengu dialect of Xhosa also remains an irritant.^[103]

Among the Xhosa, too, there is a strong strand of exclusive, anti-White nationalism, found particularly among so-called Red Xhosa, those who adhere to a consciously traditional subculture reflecting a rejection of the White conquest:

The happy past to which the Red person looks back is the period . . . when there were no White men; the dream he would entertain for his future is getting away from the White men again—away into Xhosa independence and dignity. Meanwhile no inconsistent cultural aspirations complicate the straightforward political opposition. "There is nothing I like about them or their way of life. There can never be any peace between a Xhosa and Umlungu [the White man]."^[104]

Although the Red tradition is in decline,^[105] both it and the more acculturating "School" tradition among the Xhosa reflect "the different reactions of an earlier generation to a particular conquest situation over

[102] Maurice Webb and Kenneth Kirkwood, The Durban Riots and After (Johannesburg: South African Institute of Race Relations, 1949); Union of South Africa, Report of the Commission of Enquiry into Riots in Durban (Cape Town: Cape Times, 1949).

[103] Manona, "Ethnic Relations in the Ciskei," p. 103.

[104] Mayer, Townsmen or Tribesmen? p. 33.

[105] R. Hunt Davis, "School vs. Blanket and Settler: Elijah Makiwane and the Leadership of the Cape School Community," African Affairs 78, no. 310 (January 1979): 12–31, at 13.

a hundred years ago."^[106] One part of that conquest was the dispossession of Blacks by Whites. Another part was the participation of some groups in helping to subjugate others. Both components are remembered.^[107]

It is not surprising in such an environment that both nationalism and ethnicity should have claims on the loyalties of Black South Africans. The intensive contact and protracted warfare between Xhosa and Whites make Xhosa overrepresentation in the ranks of militant movements understandable.^[108] Some recrudescence of Zulu historical pride helps explain an organization like Inkatha, with its paramilitary trappings. The Zulu-Pondo Wars of the 1820s are perhaps in the distant background of recent Zulu-Pondo violence. Throughout the twentieth century, "Zulu ethnic nationalism" has been widespread in Natal, and among its sources were "the ways in which a pre-colonial past provided military metaphors for mobilization."^[109] And if some groups, like the Pedi, have a degree of enthusiasm even for their government-created homeland, that, too, can be understood, in part, by reference to the turbulent history of their relations to their neighbors.^[110] Ethnicity may be no more keenly felt than in other African countries, but historical recollections as a component of ethnic sentiment may be more powerful in South Africa.

Further Complexity:
Whites, Indians, Coloureds

The initial apartheid vision of a four-segment society—of Whites, Coloureds, Indians, and Africans—is confounded by the social complexity, not merely of the African category, but of the Indian, Coloured, and White categories as well. The picture of ideological and ascriptive divisions among Africans is mirrored by similar divisions within each of the other categories.

Political differences among Whites are greater now than they have

[106] Mayer, Townsmen or Tribesmen? p. 293.

[107] See Hanf et al., South Africa, p. 246; Cornevin, "Populations noires d'Afrique du Sud," p. 43. For the roles of historical memory and warfare in ethnic relations, see Anthony D. Smith, The Ethnic Origins of Nations (Oxford: Blackwell, 1986), pp. 30–31, 37–41.

[108] See Winnie Mandela, Part of My Soul Went with Him (New York: Norton, 1984), p. 48: " . . . you tell yourself: 'If they failed in those nine Xhosa wars, I am one of them, and I will start from where those Xhosas left off and get my land back.'"

[109] Shula Marks, "Patriotism, Patriarchy and Purity: Natal and the Politics of Zulu Ethnic Consciousness," in Vail, ed., The Creation of Tribalism in Southern Africa, p. 233.

[110] See Delius, The Land Belongs to Us .

been since before 1948.^[111] The most obvious difference is in the realm of party competition and the divergent opinions it reflects. Between the 1987 and 1989 elections to the White chamber of parliament, the political spectrum spread out considerably. The Conservative Party moved from 27 percent of the vote in 1987 to 31 percent in 1989. In 1987, the liberal opposition consisted of the Progressive Federal Party (PFP). Together with the small New Republic Party, the PFP had 16 percent of the vote. By 1989, the liberal opposition became consolidated in a new Democratic Party (DP), consisting of the former PFP and a number of defectors from the Nationalists. In the 1989 elections, the DP gained 20 percent of the vote. Quite correctly, the National Party had pointed out in its campaign that the Democrats were not committed to a White minority veto in a future South Africa. For the first time, moreover, an anti-apartheid party in parliament had more than a token share of Afrikaner support.^[112] Reciprocally, a great many English-speaking Whites gave their support to the National Party.^[113] Ethnicity is no longer the firm predictor of political attitudes within the White category that it once was. This, a mere dozen years after Black South Africans had begun to make sharper distinctions between Afrikaner and English Whites.

Changes among Afrikaner elites are palpable. In 1985, a "Stellenbosch '85" discussion group was formed at the Afrikaans-language university that has nurtured a large number of South African cabinet ministers. Out of its discussions came an acrimonious meeting of some 28 leading Afrikaner intellectuals with then-President P. W. Botha. The disaffection of these intellectuals coincided with the defection of several Afrikaner politicians from the National Party in 1987. These streams fed directly into the new Democratic Party.

The roots of Afrikaner liberalism, particularly in the Cape, go back much further, but its political significance is traceable more immediately to the declining legitimacy of the old regime of race relations among many educated Afrikaners.^[114] Among such people, there was a good

[111] For a denial of any change, based on an eccentric reading of the evidence, see Bruce W. Nelan, "Changes in South Africa," Foreign Affairs 69, no. 1 (January 1990): 35–51.

[112] For some of the sources of this change, see John D. Brewer, "Black Protest in South Africa's Crisis: A Comment on Legassick," African Affairs 85, no. 2 (April 1986): 283–94.

[113] See Hennie Kotzé, "A Whole New Poll Game for '89 Nats," Sunday Times (Johannesburg), August 6, 1989. In 1977, by contrast, only 28 percent of English speakers were Nationalist supporters. H. Lever, "Public Opinion and Voting," in Anthony de Crespigny and Robert Schrire, eds., The Government and Politics of South Africa (Cape Town: Juta, 1978), p. 153.

[114] For an example, see the HSRC Investigation into Intergroup Relations, The South African Society, p. 162, which called for "the establishment of public institutions that areworthy of trust and acceptance." At the meeting called in 1985 to discuss this report, there were repeated denunciations of the government's race relations policies by Afrikaner academics. Five years earlier, P. W. Botha had been heckled by students at the University of Stellenbosch.

deal of impatience with the government's reluctance to commit itself to fundamental change.

There can be no more graphic demonstration of this shift than the changing position of the secret society called the Afrikaner Broederbond. The Broederbond has historically numbered among its members virtually the entire Afrikaner political leadership, including the cabinet and three-quarters of the National Party members of parliament, and it has been described as "a secret communication channel between the government and the Afrikaner elite."^[115] Its 1986 confidential document, "Basic Political Values for the Survival of the Afrikaner," had a conventional title, but its content was anything but conventional. Reflecting a certain blurring of ethnic lines, the document equated the political interests of the Afrikaner with those of "the white man" in general.^[116] More remarkably, the paper went on to assert that the "abolition of statutory discrimination measures must not be seen as concessions but as a prerequisite for survival," that "there can no longer be a white government," that Christian principles require that a government "must govern fairly and justly in respect of all its subjects," and that "the exclusion of effective black sharing in political processes at the highest level is a threat to the survival of the white man, which cannot be countered by maintaining the status quo or by a further consolidation of power in white hands."^[117] And, to be sure that the message had been received, the Broederbond paper noted, almost en passant, that "the head of government does not necessarily have to be white. . . ."^[118]

Not surprisingly, these statements produced a storm of right-wing invective. By South African standards, they are revolutionary formulations, illustrating both the growing social convergence between English and Afrikaans speakers^[119] and the now-legitimate occupation by Afrikaners of positions all along the spectrum of White political opinion.

To both of these tendencies, there are limits. Although the boundary

[115] Hermann Giliomee, "The National Party and the Afrikaner Broederbond," in Price and Rosberg, eds., The Apartheid Regime, p. 41.

[116] Afrikaner Broederbond, "Basic Political Values for the Survival of the Afrikaner" (N.p., 1986; unpublished paper), pp. 1, 2, 7.

[117] Ibid., pp. 6, 8, 6, 7, respectively.

[118] Ibid., p. 7.

[119] "A common white South African nationalism is well on the way to replacing the historical group identities." Adam, "The South African Power Elite," p. 99. Cf. Hanf et al., South Africa, pp. 169–73.

between English and Afrikaners is less firm than it once was, it has not been obliterated. Recollections of the Anglo-Boer Wars and of divergent attitudes toward European fascism in the 1930s and 1940s have been muted but not wholly transcended. A 1989 survey of members of the Democratic Party, which had not yet chosen a leader, showed that English-speaking party members supported Denis Worrall, an English speaker, for the leadership position, over Wynand Malan, an Afrikaner, by 52 percent to 15 percent. Among Afrikaners, 44 percent chose Malan, while only 28 percent chose Worrall.^[120] Even in the Democratic Party, ethnic differences among Whites have political significance. Moreover, the two groups are not evenly distributed along the political spectrum. Afrikaners are a decided minority among Democrats, and English speakers are a rarity in the Conservative Party, which, while the Democrats were opposing apartheid, was attempting to reinstate segregation in Boksburg, a municipality whose town council the Conservatives controlled.^[121]

The limits of Afrikaner liberalism are emphasized in a series of papers based on sample surveys of university students, conducted by Jannie Gagiano.^[122] Gagiano identifies a number of important differences between Afrikaner Democratic Party and Afrikaner National Party supporters—including the firm rejection by the former of a White-controlled government—and he documents the extent to which White students are groping for an alternative to the status quo. Nevertheless, he is at pains to underscore the extent to which Afrikaner students remain symbolically attached to the institutions of the polity, despite the ferment around them.

If Whites present a picture of some interethnic convergence coupled with ideological divergence, somewhat analogous trends are present among both Indians and Coloureds. Both of these groups are ideologically conflicted, and the Indians are ethnically divided as well.

Indians are nearly everywhere a socially divided community, and South Africa is no exception. Between the descendants of the initial inden-

[120] Sunday Times (Johannesburg), August 6, 1989.

[121] Daily News (Durban), August 2, 1989.

[122] Jannie Gagiano, "The Scope of Regime Support: A Case Study," in Hermann Giliomee and Lawrence Schlemmer, eds., Negotiating South Africa's Future (New York: St. Martin's Press, 1989), pp. 52–62; Jannie Gagiano, "Ruling Group Cohesion in South Africa: A Study of Political Attitudes among White University Students" (paper presented at the Biennial Conference of the Political Science Association of South Africa, October 9–11, 1989); Jannie Gagiano, "Meanwhile, Back on the 'Boereplaas,'" Politikon 13, no. 2 (December 1986): 3–23.

tured laborers and the later merchants, there are predictable class differences. To some extent, these are compounded by overlapping ascriptive divisions of language, caste, and religion. A working-class person is more likely to be Tamil, Hindu, and low caste. A businessman is more likely to be Gujarati, perhaps Muslim, and not low caste. To some extent, such differences translate into political positions, for the Indians are a polarized community. Following Gandhi, many of them have joined the Natal and Transvaal Indian Congresses and aligned with the UDF. On the other hand, the serious Durban anti-Indian riots of 1949, and some much milder but threatening episodes in 1974, 1985, and 1990, have served to remind many Indians of the dangers that have befallen Asian communities elsewhere in Africa. We have seen that Coloureds and Indians usually are ranked by Africans as rather distant on social distance scales. But there are differences. In the nationwide HSRC study, Coloureds and Afrikaners received the least positive adjectival ratings, and Indians placed closer to the middle of the scale.^[123] However, in the study of Durban, where most Indians live, social distance was greatest from Indians.^[124]

Indians, therefore, have reasons for a certain conservatism, and there is no doubt that the unrest of the mid-1980s accelerated it. In a 1985 Durban survey that asked the respondents' choice of South African leader, most Indians (53.4 percent) selected none other than P. W. Botha; none chose Tutu, only 3.7 percent chose Mandela, and 3.1 percent chose Boesak. The African figures are in stark contrast, with a majority for Mandela, 11.9 percent for Tutu, 13.6 percent for Boesak, and none for Botha.^[125]

These results do not mean that Indians rushed to embrace their new legislative chamber. On the contrary, only about 20 percent of registered voters cast a ballot in 1984, dividing their support between two parties. Perhaps a better measure of Indian allegiance comes from a question on organizational support in the same 1985 survey. Half of all Africans gave their support to the UDF, and only 18.5 percent professed support for no organization. By contrast, only 8.7 percent of Indians

[123] HSRC Investigation into Intergroup Relations, The South African Society , pp. 81–82.

[124] Du Toit, "Ethnicity, Neighborliness, and Friendship among Urban Africans in South Africa," p. 153. See also Mann, Attitudes towards Ethnic Groups," p. 64.

[125] Meer and Reynolds, "Sample Survey of Perceptions of the Durban Unrest," p. 262.

supported the UDF, and another 8.7 percent supported the Indian parliamentary chamber. Half of all Indians professed support for no organization.^[126] Bifurcated between pro-UDF activists and pro-system moderates, the Indians are a community divided, in-between, and heavily apolitical.

To some extent, the same applies to the so-called Coloured community, which derives from a variety of sources: ex-slaves in the Cape, people of mixed Khoi or San and European ancestry, and "Malays" brought by the Dutch from Indonesia. For some purposes, the differences between the largely-Muslim Malays and other Coloureds are significant, but they do not coincide with firm political lines. The watershed events for Coloureds appear to have been their disfranchisement in the early apartheid years and their massive forced removal in the 1960s, under the Group Areas Act, from parts of Cape Town in which they had long resided.^[127] Two-thirds of those removed under the Group Areas Act were Coloured, and nearly one-third was Indian.^[128] These were traumatic, bitter experiences that formed the prelude to increasing militancy, some identification with the plight of Africans, and growing rejection of the Afrikaans language, the mother tongue of most Coloureds.

In the 1984 elections to the Coloured chamber of parliament, only 30 percent of eligible voters turned out. They overwhelmingly elected members of Allan Hendrickse's Labour Party, which has repeatedly emphasized the illegitimacy of the tricameral constitutional arrangements and threatened not to participate in further elections if the system is not transformed.^[129] The University of the Western Cape, officially a Coloured campus, has opened its doors to Africans and has been a center

[126] Ibid., p. 264.

[127] For the bitterness of this episode, see the novel by Richard Rive, "Buckingham Palace," District Six (Cape Town: David Philip, 1986).

[128] See Hendrik W. van der Merwe, Pursuing Justice and Peace in South Africa (London: Routledge, 1989), p. 29. In a study that followed the removals by only a few years, H. G. Viljoen found that, despite religious differences between the two groups, Indians displayed less social distance from Coloureds than from any other group. "Stereotipes en sociale afstand," p. 150. Since Viljoen had no group of Coloured respondents, we do not know whether the sentiment was reciprocated, but the finding is testimony to the power of common situation and official definitions in shaping affinities.

[129] For Labour Party noncooperation with government initiatives in the (Coloured) House of Representatives, see Cooper et al., Race Relations Survey, 1987/88 , pp. xxxv–xxxvi. See also Hennie Kotzé, "Adapting or Dyeing: Parliamentary Political Parties, Reform and Reaction in South Africa," in D. J. van Vuuren et al., eds., South Africa: The Challenge of Reform (Pinetown, South Africa: Burgess, 1989), pp. 150–52.

of radical anti-government activity. Many Coloureds have joined the UDF, and the community as a whole is somewhat more anti-government than is the Indian community.

At the same time, there are crosscurrents, as might well be expected, given the strongly negative social distance evaluations of Coloureds by African respondents.^[130] Asked in the 1985 Durban survey for their leadership choice, 31 percent, a plurality, still chose Botha, and 25 percent chose Frederik van Zyl Slabbert, then PFP leader of the parliamentary opposition. Mandela received only 11 percent support, Tutu 5 percent, and Boesak 3 percent.^[131] Likewise, Coloureds in Natal did not opt for a radical line. While half of Natal's Africans professed to support the UDF, only 7 percent of the Coloureds did. A clear majority of Coloured respondents (56 percent) said they supported no organization.^[132] In sample surveys, Coloureds show significantly higher levels of life satisfaction than Africans do—by levels of nearly 2 to 1.^[133] The Coloured levels (81 percent) are close to those reported by both Indians and Whites (89 percent). The judgment seems inescapable that the unrest of the mid-1980s produced, in Fatima Meer's words, "a radicalization of African political sentiment, and a marked shift towards conservatism on the part of Coloured and particularly Indian people."^[134] In recent years, more than a few Coloureds have emigrated to Canada and Australia.

Among students, moreover, Coloured responses to survey options

[130] Even in Brian du Toit's study "Ethnicity, Neighborliness, and Friendship among Urban Africans in South Africa," p. 153, conducted in Durban, which has few Coloureds, Coloureds were evaluated as the third most distant group from Zulu respondents, after Indians and Whites, in that order. In the HSRC study, Coloureds were seen as more distant than Afrikaners by Sotho respondents and just about as distant as Afrikaners by Zulu respondents. HSRC Investigation into Intergroup Relations, The South African Society , p. 82.

[131] Meer and Reynolds, "Sample Survey of Perceptions of the Durban Unrest," p. 262.

[132] Ibid., p. 264.

[133] V. Moller et al., "Quality of Life and Race in South Africa: A Preliminary Analysis" (Centre for Applied Social Sciences, University of Natal, Durban, September 1985; unpublished paper), pp. 8, 22.

[134] Meer and Reynolds, "Sample Survey of Perceptions of the Durban Unrest," p. 261. For general surveys of the two groups, see Kogila A. Moodley, "Structured Inequality and Minority Anxiety: Responses of Middle Groups in South Africa," in Price and Rosberg, eds., The Apartheid Regime , pp. 217–35; Anil Sookdeo, "The Transformation of Ethnic Identities: The Case of 'Coloured' and Indian [South] Africans," Journal of Ethnic Studies 15, no. 4 (Winter 1988): 69–83; Karl P. Magyar, "The Permanent Minority: Politicization of South Africa's Indian Community" (paper presented at the Congress of the Political Science Association of South Africa, September 20, 1985). For an argument that Indians were united and mobilized into the struggle against apartheid, see Pushpa Hargovan, "Apartheid and the Indian Community in South Africa: Isolation or Cooperation," Journal of Asian and African Affairs 1, no. 2 (December 1989): 155–73.

are notably different from those of African students. As we shall see in Chapter 3, African university students in the Western Cape overwhelmingly chose a "hard nationalist" option among possible future arrangements, and only 5.8 percent chose "a society in which group identity has ceased to be crucial in determining who governs."^[135] Among Coloured students in the Western Cape, this liberal option was chosen by literally ten times as many respondents: 59 percent.^[136]

The Political Significance of Ethnicity and Race

What we have seen is the continuing importance of racial, ethnic, and subethnic identities in South Africa. In all the racial categories, there are divisions along lines of ethnicity and ideology. Some intergroup cleavages have considerable conflict-producing potential. Eliminate White domination, and intra-African differences will be particularly important. They are very much on the order of the cleavages that in some countries translated into the serious post-independence conflict and violence. To ignore them in planning for a future South Africa would be to repeat the same fallacy of assuming in the 1950s and 1960s that an inclusive "nationalism" would be the universal solvent of differences in post-colonial Africa, a fallacy for which many people paid dearly. It is a fallacy the Nigerians, among others, want no part of, as their ventures into preventive measures attest.

It is not at all my purpose to paint a worst-possible-case scenario. All that needs to be said is that ethnic conflict is a recurrent phenomenon, that the ascriptive character of affiliations puts a strain on democratic institutions unless precautions are taken, and that having been through a long bout of White-Black conflict constitutes no inoculation against other varieties. Put simply, that is why South Africa's politics will not be nonracial and nonethnic but multiracial and multiethnic, and that is also why preventive arrangements should appeal to everybody in South Africa, regardless of color.

It might be thought that no African ethnic group is large enough to advance a claim to domination by itself; none even approaches 50 percent of the African population. Such arithmetic speculations would be quite misconceived in comparative perspective. To begin with, South

[135] Peter Collins, The Ethnic Factor in South Africa's Politics (University of Cape Town, n.d.; mimeo.), p. 95.

[136] Ibid.

African Whites, a small minority, maintained their domination for a very long time. In Nigeria, the Hausa comprised not more than 30 percent of the population, the Ibo and Yoruba 16–17 percent each; yet all three were major actors in an ethnic politics that, in the First Republic, produced essentially a Hausa-Ibo showdown. Likewise, the Kikuyu, with about 20 percent of the Kenyan population, managed to dominate Kenyan politics and exclude others for decades. Once the democratic process cracks under the strain of ethnic conflict and of parties aligned with particular ethnic groups—whether the crack comes from a military coup or the inauguration of a single-party state—the way is open for much smaller minorities to rule behind the facade of an allegedly inclusive military or an allegedly inclusive single party.^[137] That possibility cannot be excluded for South Africa unless powerful precautions are taken. Would it not be ironic, after all the struggle that has passed, for South Africa to end up, as other countries have, with simply another form of minority rule?

[137] For a description of how this can happen, see Donald L. Horowitz, Ethnic Groups in Conflict (Berkeley and Los Angeles: University of California Press, 1985), pp. 429–37, 486–508.