1. Data are drawn from interviews conducted in 1972–73 with over 250 leaders of six major groups: (1) businessmen—bankers and industrialists, both domestic and multinational (n = 84); (2) labor leaders (53); (3) top civil servants (56); (4) bishops (11); (5) senators and deputies of the government party, ARENA (33); and (6) politicians of the opposition MDB party (14). Sampling procedures are outlined in the appendix. The period of the early seventies, during which General Emílio Garrastazu Medici was president of Brazil (1969-74), is covered in Alfred Stepan (ed.). Authoritarian Brazil (New Haven: Yale University Press, 1974).

2. This conceptualization implies neither that the two axes are equally important nor that they are orthogonal—that is, completely independent from one another. As the evidence to be presented shows, the second inference is false. Moreover, in the course of the analysis, no hard-and-fast distinction is made between “the state” and “the regime,” even though some observers consider the refinement to be of theoretical consequence. See Fernando Henrique Cardoso, “On the Characterization of Authoritarian Regimes in Latin America,” in David Collier (ed.). The New Authoritarianism in Latin America (Princeton: Princeton University Press, 1979); compare Peter Gourevitch, “The International System and Regime Formation,” Comparative Politics 10 (Apr. 1978): 419–38.

3. The comparison with agrarian bureaucracies is merely illustrative; their elite structures may be simpler than those of developmental dictatorships but they are prone to elite factionalism all the same. Small-scale communal societies may exemplify the ideal-type more closely. See Theda Skocpol, States and Social Revolutions (London: Cambridge University Press, 1979).

4. More elaborate treatments of this idea can be found in Andrew Martin, A General Theory of Secularization (New York: Harper & Row, 1978) and Robert N. Bellah, Beyond Belief: Essays on Religion in a Post-Traditional World (New York: Harper & Row, 1970), especially pp. 64–73.

5. For Brazil, see Thomas C. Bruneau, The Political Transformation of the Brazilian Catholic Church (London: Cambridge University Press, 1974) and Ralph Della Cava, “Catholicism and Society in Twentieth Century Brazil,” LARR 11 (1976): 7–50. For Spain, see Norman Cooper, “The Church: From Crusade to Christianity,” in Paul Preston (ed.), Spain in Crisis (New York: Harper & Row, 1976) and the literature cited therein.

6. Compare Carlos Guilherme Mota, Ideologia da Cultura Brasileira, 1933–1974 (São Paulo: Editora Atica, 1974), José Honório Rodrigues, Conciliação e Reforma no Brasil (Rio de Janeiro: Editora Civilização Brasileira, 1965) and Edward E. Malefakis, Agrarian Reform and Peasant Revolution in Spain: Origins of the Civil War (New Haven: Yale University Press, 1970). This interpretation need not entail a stereotypical contrast between Brazilian and Spanish “temperaments.” Instead, it refers to distinctive institutional histories that facilitate or impede the expression of political conflict in “fanatic” as compared to “pragmatic” terms. Nor does it imply that the styles of either Brazilian or Spanish politics are unchanging. Indeed, the Spanish case—specifically, the period of post-1950 Francoism—furnished evidence for the seminal argument that Spanish politics is less than fully ideological. The locus classicus is Juan J. Linz, “An Authoritarian Regime: Spain,” in Eric Allardt and Yrjo Littunen (eds.), Cleavages, Ideologies, and Party Systems (Helsinki: Academic Bookstore, 1964). For all this, it does appear that Spanish politics, at least at the elite level, has a greater propensity toward intransigence than is the case in Brazil. See Bartolomé Bennassar, The Spanish Character (Berkeley: University of California Press, 1978). One crucial institutional factor distinguishing the Spanish and Brazilian “styles” is that, in Spain, the Church has long been influential in the educational system, with Church schools continuing to receive government subsidies. In Brazil, the hold of the Church on the schools has been much less firm. The hypothesis that different educational experiences contribute to differential propensities toward ideological combat is eminently plausible.

7. At first glance, this hypothesis seems to stand in direct contrast to the explanation of the severity of bureaucratic-authoritarian rule as a function of the perceived threat of populist mobilization, joined with inflationary pressures and the bottlenecks associated with the changeover from light to heavy industrialization. See Guillermo A. O'Donnell, Modernization and Bureaucratic-Authoritarianism (Berkeley, Calif.: Institute of International Studies, 1973) and “Reflections on the Patterns of Change in the Bureaucratic-Authoritarian State,” LARR 13, no. 1 (1978):3–38. But this is not necessarily the case. “Severity” is a fuzzy dependent variable: it may refer to the durability as well as the onset of authoritarianism. See John Sheahan, “Market-Oriented Economic Policies and Political Repression in Latin America,” Economic Development and Cultural Change 28 (Jan. 1980): 267–91. In the long run, cultural-institutional tradition may constitute a swing factor, at least as significant as economic conditions, in influencing the harshness of authoritarian episodes. Some discussions of Latin American authoritarianism recognize this possibility even when they do not pursue it. See for example Robert R. Kaufman, “Industrial Change and Authoritarian Rule in Latin America,” in Collier (ed.), The New Authoritarianism.

8. The implications of this hypothesis stretch back to the debate over the relative effects of cultural versus class factors in accounting for the authoritarian renaissance in Latin America. See inter alia Alfred Stepan, The State and Society: Peru in Comparative Perspective (Princeton: Princeton University Press, 1978) and Howard J. Wiarda, “Corporatism in Iberian and Latin American Political Analysis,” Comparative Politics 10 (Jan. 1978): 307–12. A major point of the present article is that the either-or nature of much of this debate is misguided especially insofar as the explicandum—the onset, the durability as well as the decline of authoritarianism—is ill-defined. Compare David J. Elkins and Richard E. B. Simons, “A Cause in Search of Its Effect, or What Does Political Culture Explain,” Comparative Politics 11 (Jan. 1979): 127–45.

9. Many scholars of Latin America have noted that elite as well as nonelite networks may be composed of several kinds of linkages. But it is rare to encounter a formalization of the idea that different ties may produce different networks. A seminal work for the understanding of social networks in Brazil is Anthony Leeds, “Brazilian Careers and Social Structures,” American Anthropologist 66 (1964): 1321–47; for comparative analyses, see Steffen W. Schmidt et al. (eds.), Friends, Followers, and Factions: A Reader in Political Clientelism (Berkeley: University of California Press, 1977).

10. Compare Ellen Kay Trimberger, “A Theory of Elite Revolutions,” Studies in Comparative International Development 7 (Fall 1972): 191–207.

11. Because of space restrictions, analysis of the socioeconomic backgrounds of the elites—their class origins and their education—is omitted. Briefly: most Brazilian industrialists and financiers are not self-made men; many of them are the sons of prominent figures in industry and banking. At the other extreme are the labor leaders, who are almost exclusively of working-class provenance. It is primarily among the civil servants and, to a lesser extent, the politicians that formal education assumes a crucial role in mobility. Most of the state managers are from the middle sectors, and they depend on higher education to overcome their class origins. For a fuller discussion, see Peter McDonough, Accommodation and Confrontation among Brazilian Elites (Princeton: Princeton University Press, 1981).

12. Compare the studies conducted by Maurice Zeitlin and his colleagues in the United States and Chile of family ties among elites: e.g., Zeitlin and Richard Earl Ratcliff, “Research Methods for the Analysis of the Internal Class Structure of Dominant Classes: The Case of Landlords and Capitalists in Chile,” LARR 10 (Fall 1975): 5–61 and Zeitlin, “Corporate Ownership and Control: The Large Corporation and the Capitalist Class,” American Journal of Sociology 79 (Mar. 1974): 1073–1119.

13. See Lois M. Verbrugge, “Multiplexity in Adult Friendships,” Social Forces 57 (June 1979): 1286–1308.

14. An index of endogamy has been devised which can be used with the type of data presented in table 1; see David J. Strauss, “Measuring Endogamy,” Social Science Research 6 (1977): 225–45. However, percentage differences are adequate to the task of assessing the extent of endogamy/exogamy, and they focus attention on the shifts between specific occupational categories. The seven-category occupational classification is a simplified version of an original 20-category code. “Manual” includes urban and rural labor, both skilled and unskilled; “routine nonmanual” includes low-level public functionaries, elementary school teachers, enlisted men, salesmen, etc.; “middle class” includes shopkeepers, highschool teachers, middle-level government functionaries; “professionals” are doctors, lawyers, engineers, accountants, university professors, etc.; and the top categories are largely self-explanatory. “Top military/political,” for example, takes in nationally known politicians, ministers, generals and admirals. Later on, the three highest categories are collapsed into one to form a five-point scale, since it is impossible to rank-order these categories. The name given this grouping is “upper class.”

15. Endogamy within the upper class may appear to be lower than it actually is because table 1 retains the tripartite disaggregation of this stratum. When the categories are joined, it can be seen that endogamy at this level reaches 60 percent.

16. See Warren Dean, The Industrialization of São Paulo, 1880–1945 (Austin: University of Texas Press, 1969).

17. Technically, it is possible for correlations between the occupational rankings of fathers and fathers-in-law to be quite high even though elites do not marry within the same class. There could be a constant difference between the rankings, such that the manual category regularly into the routine nonmanual; the routine nonmanual into the middle sectors, and so on. But inspection of the empirical distributions shows that this is not the case. Hence, the correlations shown here (and in figure 2) are proper measures of endogamy, and they permit a more economical presentation than a series of crosstabulations. The role of education in breaking down barriers to mobility is not, of course, uniquely Brazilian. Compare, for example, Sheila Fitzpatrick, Education and Social Mobility in the Soviet Union, 1921–1934 (London: Cambridge University Press, 1980).

18. To say that businessmen are inclined to marry within their class does not mean that they all marry into business families. It does mean that they are less mobile by way of marriage than the other elites. In principle, this suggests that the occasional businessman who rises from working class or routine nonmanual origins is more likely to have married into a family of similar background than into a “business family.” In practice, there are very few businessmen from the lower classes. It is their “studied” lack of education which seems to be the key factor in the relative immobilism of the industrialists and bankers.

19. The relevant correlations drop from.62 within the oldest cohort of businessmen (those 61 years of age and older) to.44 within the youngest (the 31 to 40 year-olds). I do not press this trend because of the possibility that it may reflect life-cycle rather than intergenerational changes, even though ancillary evidence indicates that a historical interpretation is most probably correct.

20. Compare Peter H. Smith, Labyrinths of Power: Political Recruitment in Twentieth-Century Mexico (Princeton: Princeton University Press, 1979), pp. 205–16. See also Merilee S. Grindle, “Patrons and Clients in the Bureaucracy: Career Networks in Mexico, LARR 12, no. 1 (1977):37–66.

21. See Mark S. Granovetter, “The Strength of Weak Ties,” American Journal of Sociology 78 (May 1973): 1360–80.

22. The questions are as follows: (1) “When you were growing up, before you reached the age of 18, do you recall if any of your relatives or members of your family belonged to the following 13 groups?” (2) “Could you tell me if any of your colleagues from the university now occupy positions in the following 13 groups?” (3) “Could you tell me if any of your relatives or friends currently occupy any position in the following 13 groups?” When responses were positive to (2) or (3), the elites were asked whether they had contact with their friends and/or relatives in the groups. In addition, for (3), kin were divided into familiares (immediate family) and sogro ou cunhado (father-in-law or brother-in-law). These distinctions have been abandoned in the analysis because they are practically nonexistent empirically: that is, contact is almost always maintained, and degrees of parentesco (kinship) are almost never distinguished.

23. The method used is smallest-space analysis. It generates results similar to those of factor analysis without some of the latter's restrictive assumptions (e.g., intervallevel measurement). The coefficients at the bottom of the diagram indicate the success of the technique in reducing the 13 ties to one, two, and three dimensions: the lower the coefficients, the better the fit. The one-dimensional solution is unsatisfactory. It would reduce the power structure to a split between the labor leadership and all other groups. The two-dimensional solution, shown here, is acceptable. The three-dimensional rendition, which would place the church on a separate axis, is even better. The fact that the ecclesiastical hierarchy forms the basis of the third, but still significant, axis of the power structure is, of course, quite in line with theoretical expectations. For a lucid presentation of the method, see Kenneth B. Bailey, “Interpreting Smallest-Space Analysis,” Sociological Methods and Research 3 (Aug. 1974): 3–29.

24. The results correspond to informed observations about the configuration of power in post-1964 Brazil; see for example Fernando Henrique Cardoso, O Modelo Político Brasileiro (São Paulo: Difusão Européia do Livro, 1973) and Helio Jaguaribe, Brasil: Crise e Alternativas (Rio de Janeiro: Zahar, 1974).

25. See Celso Lafer, O Sistema Político Brasileiro (São Paulo: Perspectiva, 1975).

26. See Charles Antoine, “L'Église Catholique: De la Résistance à l'Opposition,” Le Monde Diplomatique (Feb. 1980): 4–5. The important point is that, while the Church in Brazil has from time to time sided with both conservative and progressive currents, as an institution it has not been bound to one camp or the other for extended periods. This is the long-term context of the Church's present endorsement of liberation theology as well as of the internal factionalism that this change has precipitated. For diverse interpretations, see Paulo Sérgio Pinheiro, “Anos de Crise,” Isto É, 2 June 1980, pp. 58–65; Ralph Della Cava, “Church Leads the Cause of Human Rights in Brazil,” International Development Review 22 (1980): 40–43, and Penny Lernoux, “Latin America: The Revolutionary Bishops,” The Atlantic 246 (July 1980): 6–14.

27. The underlying orderliness of day-to-day interactions among Brazilian elites is a significant pattern, for it is not at all obvious to the casual eye, nor is it derivable from the more schematic macro-models of Latin American politics. For a discussion of this problem, see Linn A. Hammergren, “Corporatism in Latin American Politics: A Reexamination of the ‘Unique’ Tradition,” Comparative Politics 9 (July 1977): 443–61.

28. President Medici was nominated as “most influential” by 41 percent of the elites, and Delfim Neto by 17 percent. No other individual approaches these men in the hierarchy. Correcting the power rankings in light of reported contact restores some balance to the hierarchy. The president is more powerful than the finance minister, but the elites have more contact with the latter. This does not work for any other individuals, because they are cited so rarely.

29. Actually, the elites nominated many more than 30 individuals as influential; the total figure is close to 200. However, if a person was not cited by at least two percent of the elites, he was grouped with other persons in his own field (e.g., business, government) rather than treated individually. In short, only 30 individuals were nominated as influential by two percent or more of the respondents.

30. The variables entered into the analysis are the measures of contact, scored from “1” (no contact) to “4”frequent contact). Virtually identical results are obtained when the influence measure is used, because the two variables are highly correlated. In this case “greatest influence” (that is, cited at the top of the list of the “five or six” most powerful figures) received a score of “6,” the second-to-the-top a score of “5,” and so on, with “0” representing “not mentioned as influential.”

31. The fact that the labor leaders and the bishops are both enmeshed in the noneconomic bureaucracy does not mean that they have joint interests, even though a tentative alignment has emerged between them (and the peasantry). See Warren Hoge, “Brazilians Battle over Land, with Church Backing Poor,” New York Times, 4 March 1980, A2, and “A Face Cruel do Brasil,” Veja, 16 July 1980, pp. 84–92.

32. To take only two indicators of the weakness of the Church's presence: in Brazil there are about 12,000 priests to minister to a population of over 100 million, and half of these priests are foreigners. See David E. Mutchler, “Roman Catholicism in Brazil,” Studies in Comparative International Development 1 (1968): 104–17.

33. For documentation, see Peter McDonough, “The Developmental Priorities of Brazilian Elites,” Economic Development and Cultural Change (forthcoming).

34. It seems probable that local politics in Brazil is even more strongly influenced by parentesco ties than we have plotted at the national level. See Linda Levin, “Some Historical Implications of Kinship Organization for Family-based Politics in the Brazilian Northeast,” Comparative Studies in Society and History 21 (Apr. 1979): 262–92.

35. See Glaucio Ary Dillon Soares, Sociedade e Política no Brasil: Desenvolvimento, Classe e Política durante a Segunda República (São Paulo: Difusão Editora do Livro, 1973), pp. 36ff, and Boris Fausto, A Revolução de 1930 (São Paulo: Editora Brasiliense, 1970). Compare D. A. Brading, ed., Caudillo and Peasant in the Mexican Revolution (London: Cambridge University Press, 1980).

36. Studies of cyclical blockages in mobility have grown in recent years, and some of them are plainly relevant to the succession problem among elites. See Joan M. Waring, “Social Replenishment and Social Change: The Problem of Disordered Cohort Flow,” American Behavioral Scientist 19 (Nov.–Dec. 1975): 237–56; I. William Zartman, “The Study of Elite Circulation,” Comparative Politics 6 (Apr. 1974): 465–88; Valerie Bunce, “Leadership Succession and Policy Innovation in the Soviet Republics,” Comparative Politics 11 (July 1979): 379–401, and John D. Nagle, System and Succession: The Social Bases of Political Elite Recruitment (Austin: University of Texas Press, 1977).

37. McDonough, Accommodation and Confrontation.

38. See Peter McDonough and Amaury DeSouza, The Politics of Population in Brazil (Austin: University of Texas Press, 1980).

39. See Juan Linz, “Europe's Southern Frontier: Evolving Trends Toward What?” Daedalus, no. 108 (Winter 1979): 369–91, and Peter McDonough, Antonio López Pina, and Samuel H. Barnes, “The Spanish Public in Political Transition,” British Journal of Political Science (in press).

40. See for example Stein Rokkan, “Nation-Building, Cleavage Formation and the Structuring of Mass Politics,” in Rokkan et al., Citizens, Elections, Parties (New York: David McKay, 1970). In Brazil, such a line of investigation would lead to an analysis not only of the ambiguous history of church-state relations but also of the almost equally complex ambivalence of the military with respect to economic and political goals. Rather like that of the Brazilian Church, the role of the Brazilian military has not been uniformly reactionary through most of its history. See Edmundo Campos Coelho, Em Busca da Identidade: O Exército e a Política na Sociedade Brasileira (Rio de Janeiro: Forense Universitária, 1976) and Paulo Mercadante, A Consciência Conservadora no Brasil (Rio de Janeiro, Editôra Saga, 1965).

41. See Allen H. Barton and R. Wayne Parsons, “Consensus and Conflict among American Leaders,” Public Opinion Quarterly 38 (Winter 1974):509–28. 1974):509–28.