Five—
A Theory of Stable Democracy

Author's Note: This paper is the watershed that divided my academic life into two rather distinct parts. I have described both its prehistory and a good deal of its aftermath in an extensive monograph, The Natural History of Congruence Theory (University of Denver: Monograph Series in World Affairs, vol. 18, no. 2, 1980). Here I will be a good deal briefer.

Prehistory first. In 1959, when I began teaching at Princeton, I inherited two courses on modern democracies from Gabriel Almond. As a faithful revolutionary in comparative politics, I decided to teach the courses in a "problem-oriented" way, the problem chosen being how to explain stability and instability in democracies. At that time, the pertinent theoretical notions we had were largely concerned with formal-legal aspects of the structure of government, such as electoral systems and the way executive-legislative relationships were defined. In 1959 Lipset had published Political Man; his theories about relationships between levels and rates of economic development and political stability were also widely known. Kornhauser, a couple of years before, had published The Politics of Mass Society , in which the principal variable related not to stability, but to the tendency to join political mass movements. This could be related to stability, in the manner of Tocqueville, through the variable of intermediate associational structures in societies. Not associated with any particular individual was the idea that religion was a critical variable: Catholic countries could not be stable democracies; only Protestant countries could. There also existed an "evolutionary" explanation. It was held, in essence,

This essay first appeared in Division and Cohesion in Democracy: A Study of Norway (Princeton: Princeton University Press, 1966), Appendix B, 225–288. Reprinted with permission of the publisher. Copyright © 1966 by the Princeton University Press.

that if unavoidable, thus universal, problems like national identification, distribution, participation, and so on, are solved successively, then governments will be stable and politics moderate. If, however, they occur simultaneously, as in that prototype of unstable democracies, Germany, the opposite will occur. In the air at the same time, as always, were two vague theoretical notions involving consensus and homogeneity. Stable government was thought to follow from a high degree of consensus, particularly on what were called fundamentals, as against merely circumstantial matters. Homogeneity theory held, of course, that the more alike people are the less likely they are to be at each other's throats.

To become more fully acquainted with the literature I compiled an "inventory" of ideas pertinent to the problem. Most of these were buried, as offhand interpretations, in mostly descriptive case studies. The full inventory contained about 150 theoretical propositions: there was no lack of ideas but, on the contrary, an embarrassment of riches. Weeding seemed manifestly needed, as well as statements of general propositions that had wide applicability. So, as a next step, I reviewed the inventory to see if it contained patterns that might be the basis of more general propositions.

Something immediately became evident. There seemed to be two general schools of thought about the problem. In one case, the stability or instability of democracies was treated as dependent on the way they were endogenously constituted. In the other, the explanatory variables were external to government; governments were held to be, in one way or another, superstructural (on a vast variety of substructural variables, not only economic factors).

For several reasons, which would consume too much space to explain here, I thought both approaches deficient. Still, it seemed unlikely that governmental structure should have no significance at all and just as unlikely that the social contexts of governments should have no significant bearing on their performance. Yet, it seemed "obvious" that explanatory variables must be either endogenous or exogenous. The categories surely exhausted all the possibilities. So I worked, as best I could, with what existed—until a third possibility somehow came to mind. Governments are patterns of authority. Other social units also have authority patterns: modes of governance. Here, then, was a factor both endogenous and exogenous, and also neither: a possible "linkage-variable" that might connect governments and their contexts, and through which brute contextual factors could work effects upon governments (and vice versa). Polities and other social units, in fact, are most alike in that all have patterns of authority; that alone made it likely that "private" governments (as Charles Merriam called them) would affect public ones through the commonality of modes of governance.

And so, I wrote the article that follows—not as a final word, but as a

first attempt (an "explanation sketch") needing elaboration theoretically, conceptually, and operationally. I have been at elaboration, off and on, alone and with collaboration, ever since. The later period started with the attempt to examine the theory's plausibility for costly further work—I had hoped it would lead to a comparative test—in a country about which I knew almost nothing, except that observations in it had to support the theory. The country was Norway, and the result of inquiry there was a book that I had not intended to write, Division and Cohesion in Democracy (1966). Other publications that followed include the following:

1. "Authority Relations and Governmental Performance," Comparative Political Studies 2 (October 1969): 269–326. A more elaborate, more precise, statement of the theory.

2. Political Performance (Beverly Hills: Sage Monographs in Comparative Politics, 1971). An attempt to lay the groundwork for the dependent variable of the theory. The attempt is carried forward and empirically applied in a companion monograph by Ted Robert Gurr and Muriel McClelland.

3. With Ted R. Gurr, Patterns of Authority (New York: Wiley, 1975). A full presentation of my work, with Gurr and others, in trying to bring congruence theory to a decisive test, but a work more concerned with the study of authority relations in general than congruence theory only. Also contains a full list of others' work related to congruence theory.

4. Support for Regimes: Theories and Tests (Princeton: Center of International Studies, Research Monograph 44, 1979). This presents my view on how congruence theory may decisively be tested.

For as long as men have studied politics they have tried to discover connections between political life and its social setting. In this search one aspect of social life has been strangely neglected—strangely, because it is the element of social life most obviously and immediately relevant to political behavior on the governmental level; I refer to authority patterns in nongovernmental social relationships, in families, schools, economic organizations, and the like. We have propositions about the relations between politics and geography, politics and economic organization, politics and social stratification, politics and religion, politics and child training, politics and role structure, politics and education. But, beyond allusions to the subject, we have no general propositions about the relations between politics on the governmental level and politics in nongovernmental social structures. Under any circumstances this would be a serious omission, for it stands to reason that if any aspect of social life can directly affect gov-

ernment it is the experiences with authority that people have in other spheres of life, especially those that mold their personalities and those to which they normally devote most of their lives. The omission is all the more serious when one considers that the propositions now available for relating society and polity are all, in one way or another, highly imperfect, however much some of them may be supported by evidence and logic. Certainly none has as yet been established so rigorously as to have led to even a moderate consensus among political scientists, nor has any line of analysis yet proved so promising that political scientists can feel easy about leaving other possibilities unexplored.

There is another reason why one should be mystified by the lack of propositions relating governmental authority to other forms of social authority. For some time now numerous developments in political science seem to have been converging on just such propositions. One such development is what the political philosophers call pluralism: the belief that the state has neither a moral nor a practical monopoly upon political authority. Another is the belief that political science must concern itself with a subject matter broader than the state for methodological reasons: for example, on the ground, stated by Catlin already thirty years ago, that political science, if it is to be a true science, must have a subject matter which is simple, general, commonplace, and frequent of occurrence—the state, as we know it, satisfying none of these criteria, while the "act of control," in whatever context it may occur, allegedly satisfies them all.^[1] Still another development leading in the same direction is the growing concern of political scientists with power and influence relations as such. Finally, political scientists, and others, have been concerned for many decades now (since Mill, at least) with the possibilities of private despotism in a context of public liberty and with "informal" politics within the context of constitutional law.

These trends have in fact produced a small number of studies dealing with the politics of nongovernmental structures^[2] —a surprisingly small number, considering the enormous variety of cases available. They have also produced many exhortations to political scientists to produce more such studies, by, among others, Merriam, de Jouvenel, and Dahl.^[3] But while political scientists have not entirely neglected to study nongovernmental relations as political structures, and while they have certainly not neglected the possibilities of social determination of the operation and fate of governments, no one seems even to have thought of systematically linking the first interest to the second, of producing propositions which explain certain facts of public authority on the basis of certain facts of nonpublic authority. I can think of only one good reason for this omission—namely, that the modern idea of the state has so tenacious a hold even on those who disavow it that they tend not to think of private authority when they are concerned

with public authority, but think of it only when they are not concerned with states at all. However that may be, propositions linking the two do not yet exist, and the present paper seeks at least to begin remedying this state of affairs.

Despite its title, it presents only an idea, not a full-fledged theory. Nevertheless, it seems to me worthwhile stating the idea now, even in an obviously rudimentary form. The main reason for this emerges from the conclusion of the paper: before the idea can claim to be a full-fledged theory, so much theoretical and empirical work still needs to be done that it might be years before one could say much more about the subject than can be said about it now. In the meantime, I hope that the present statement of the idea will provoke others to look into it, so that its theoretical and empirical elaboration will not be a job falling only to myself. There is more than enough work here for all who may have an interest in it.

Charecteristics of Stable Democracy

A theory should always begin by defining its subject. In the case of stable democracy this might seem to be a simple matter, but the concept actually raises one or two knotty problems.

Ordinarily we mean by a stable democracy one that has demonstrated considerable staying power, a capacity to endure, without great or frequent changes in pattern. On this basis, the French Third Republic, to take a simple and familiar example, was certainly a stable democracy; like M. Talleyrand in the French Revolution, "it survived," longer at any rate than any other French constitutional order, and longer than many other constitutional orders elsewhere. Yet this example also indicates why a definition of stability as mere longevity will not quite do, even if we could agree on just how long a democracy must endure to be considered stable. Taking the term in this sense, a system may be stable because of its own effectiveness or simply because of the ineffectiveness (or bad luck) of its opponents; it may persist, as did the Third Republic, for no better reason than that it never quite manages even to collapse, despite much opposition and many hairbreadth escapes. But this sort of tenuous survival surely requires an explanation different from that required by the persistence of a constitutional order because of its capacity for adapting to changing conditions, for realizing political aspirations and holding fast allegiances. The first can be explained only on the basis of particular historical circumstances (or "accidents")—the misjudgments of a MacMahon, the untimely relations of a Boulanger with his mistress, the timely realization of the Communists that helping the Fascists would not expedite the proletarian revolution. The second requires an explanation, not in terms of fortuitous circumstances, but in terms of settled social and political con-

ditions; it is, therefore, for political science the more interesting phenomenon, and indeed the only one political science can really handle. The theory of this paper, consequently, does not concern itself with stability merely in the sense of endurance, but stability in a more complicated sense. Endurance is a part of it, but only a part.

What else does the concept imply? First of all, it implies effective decision making—"effective" not in the sense of right action on the basis of some particular scheme of values, but in the basic sense of action itself, any sort of action, in pursuit of shared political goals or in adjustment to changing conditions. The political order must not be "immobilistic," for immobility can lead to desired results only by inadvertence, if at all. A government must govern, whatever else it may be supposed to do, otherwise it is not a government at all; and on this criterion the Third Republic, for all its staying power, certainly comes out badly. In addition, we can hardly call a government a stable democracy unless it is genuinely "democratic," and this implies that it must satisfy at least two conditions. One is that democratic structures must not be mere facades for actual government by nondemocratic structures; the decisions of government, which make it "effective" and perhaps capable of survival, must come from the democratic process, at least in large part. On the basis of this criterion the Third Republic comes out badly again, for it was a government carried on in normal times mainly by a nearly autonomous bureaucracy ("the Republic on top, the Empire at bottom") and in times of crisis mainly by temporary dictators armed with plein pouvoir; it was, in effect, largely a system of democratic politics and nondemocratic government. The second condition is that elections in such systems must decide, in some basic way, the outcome of the competition for power and policies; for what else can one mean by democracy? On this basis, too, the Third Republic fares badly, for its elections decided very little, compared with the traditions of the bureaucracy, the whims of parliamentary dictators like Laval, Chautemps, and Daladier, and the irrepressible trasfarmismo of the Radicals, the party always indispensable to governments and always in power, whatever the outcome of elections.

The term "stability," when applied to democracies, thus implies three conditions: persistence of pattern, decisional effectiveness, and authenticity. And barring the son of large good fortune that saved the Third Republic repeatedly upon the point of extinction, we may be sure that these three conditions hang closely together. The Third Republic, in fact, is the only case that does not completely illustrate this fact—it would indeed be strange if there were many others, for the sort of accidents that preserved it have only a slight statistical probability. In other cases—the Weimar Republic, for example, or the Fourth Republic, or pre-Fascist Italy—the interdependence of all three conditions is clearly revealed. Governmental

paralysis, particularly in the face of crisis, is generally a prelude to the demise of democracy. It always leads, even before the end of the formal representative system, to government by freewheeling bureaucrats, or parliamentary dictators, or inconsequential minority leaders like Facta and von Papen. And even if a democratic system survives in form by surrendering power to its adjuncts or to a minority, it cannot in that case be said to survive in fact; nor is it then likely, on the evidence, to survive for long in form. That is why all three conditions can be sufficiently stated in a single adjective, stable democracy, without any cumbersome additional ones (e.g., "effective").

The Problem of Stable Democracy

Certain conditions obviously favor stable democracy, so obviously in fact that no decent theory ever stops with them. When someone says, for example, that democracies tend to be stable if they enjoy wide support,^[4] that individual has obviously not said very much, and probably nothing that we really want to know. Neither is it very instructive to be told that a highly integrated party system (like the British) is more conducive to stable democracy than a fragmented party system, or that a climate of moderation and pragmatism in politics is more favorable to it than one of extremism and ideological dogma.^[5] There is in fact a well-known syndrome of conditions connected with stable democracy which is practically synonymous with the term, in that it would be difficult to imagine the one without the other. This syndrome of conditions includes, above all, consensus on the form of government, a high degree of political pragmatism, and a certain kind of party system: either a two-party system or a larger number of parties possessing to a high degree what the Germans call Koalitionsfähigkeit (coalition capacity, to be very literal). These conditions do "explain" the existence of stable democracy, but only in a very superficial way: the same way that the statement, "X pulled the trigger of a loaded revolver while pointing the muzzle at himself," explains how X came to commit suicide. In both cases, the explanation we really seek is one step further removed. We do not really want to know what son of political party systems and climate of opinion favor stable democracy. We really want to know what conditions underlie the requisite syndrome of favorable conditions, whatever it may be, for only on this level of the problem are the answers not obvious.

Getting beyond the obvious in regard to the problem of stable democracy requires also that we do not dwell very long on the general functional and structural requisites of viable social systems that sociologists and general systems theorists have so far managed to establish, even though these requisites are meant explicitly to explain the stability or instability of sys-

tems.^[6] We could, of course, discuss the collapse of democracies like the Weimar Republic, pre-Fascist Italy, or the French Fourth Republic in terms of these requisites; and keeping them in mind will certainly help us to look for and to find certain important weaknesses and malfunctions in these political systems. But the use of system requisites is also unlikely to lead us to the desired level of explanation. To some extent these categories will merely help establish the fact that the political systems mentioned do indeed fit our definition of unstable democracy: for example, that they failed to "maintain" the democratic "pattern" by the surrender of power to nondemocratic adjuncts of the system. To some extent, they will only lead us to the more familiar truistic theories stated earlier: to lack of "integration" in party systems, for example, or lack of "conformity" to the norms of the democratic polity on the part of many of its members. To some extent they will lead us to certain symptoms rather than preconditions (effects rather than causes) of instability: for example, the inadequate "adaptation" of fluid social resources through faulty economic policies, as in the constantly misshapen budgets of the Third and Fourth republics; or the failure of basic mechanisms of accountability and consent, as in the virtually unchecked use of decree powers by parliamentary autocrats; or the failure of a government's structure of coercion, as illustrated by the unconstrained violence in pre-Fascist Italy by D'Annunzio's orgiastic nationalists, pillaging peasants and workers, and Mussolini's own fascisti . In every democratic system that has failed, we do find a failure to satisfy some of the system requisites that men like Parsons, Levy, and Apter have proposed; if nothing else, this means that their claims to have established truly general categories are not invalidated. But the fact remains that these categories simply do not gel us to the level of the problem with which we are concerned in this case.

Our real problem is not to find whether system failures have indeed occurred in unstable democracies or to link their occurrence to manifestly unfavorable conditions. We need to discover the deeper, or more remote, conditions that rule out or make possible stable democracy, either directly or by bringing into being the obviously deleterious or obviously favorable conditions discussed above.

The Congruece of Authority Patterns

The Universality of Authority

To solve this problem, a theory is required that, up to a point, is also a general theory of governmental stability. Since democracy is a special kind of government, it seems plain logic that a theory of stable democracy should consist of two parts, one stating the general conditions that make

governmental stability probable, the other stating the particular conditions required to make democracies stable. These particular conditions should of course be special instances of the more general conditions that produce governmental stability. It may be that no general theory of governmental stability can really be developed, that every kind of government is sui generis in this regard; but at this stage of inquiry we have no reason whatever to think that this is the case. Hence the first proposition of the present theory is in fact a proposition about the stability or instability of any governmental order, whatever its special character. This proposition concerns the nature of, and the relations among, the different authority patterns in a society.

In every society we can discover numerous authority patterns, both attitudes regarding authority and, to use Lasswell's terminology, authority "practices." Certainly this is so if we use the term "authority" in its broadest and most conventional sense, to denote relationships of superordination and subordination among individuals in social formations, relationships in which some members of the formation take decisions and others treat the decisions as binding. In this simple sense of a hierarchy of wills, the state certainly has no monopoly upon authority. It may be quite possible, and even useful, to distinguish between the authority of the state and other kinds of authority; and it is easy enough to define authority in a way that will confine it arbitrarily to the state. But as the term is used here, authority in some form is a characteristic of practically any persistent social aggregate, at least in that certain actual practices of subordination and superordination will be found in such aggregates, and probably also in that there will exist in the society as a whole and in its subunits certain dominant notions as to how such practices should be conducted. Authority exists not only in the state itself, but also in parties and pressure groups, in economic organizations, in various kinds of associations, in schools and families, even in friendships, bands, clubs, and gangs. We can discover it in any set of social relations which, in the not too happy jargon of social psychology, is "cooperatively interdependent"^[7] —that is, very simply, not competitive. The only persistent social relations in which we are pretty certain not to find it are "competitively interdependent" relations, like the bargaining relations that take place in a free economic market or in international politics—granted that a certain amount of bargaining can be an aspect of authority relations as well, and a certain amount of authority sets limits to the scope and content of bargaining.

I assert this universality of authority patterns in noncompetitive social relations, not because it is absolutely necessary to my theory to assert it, but because it is a palpable fact. All that is necessary to the present theory, however, is that authority should exist in some social relations other than those of formal government, particularly in those social relations, like the

family or economic organizations, which one finds in any society; and this assertion will surely not be disputed by anyone without some sort of redefinition of authority.

The Idea of Congruence

Stated very briefly, the first proposition of the theory I would suggest is that a government will tend to be stable if its authority pattern is congruent with the other authority patterns of the society of which it is a part . The crucial term in this proposition is of course "congruence," and it needs to be defined, particularly since, as used here, the term is not at all self-explanatory.

Authority patterns are congruent, in the first place (but only in the first place), if they are identical (that is to say, since we are dealing not with an abstract geometric universe but real life, if they very closely resemble each other). An example of congruence in this sense is furnished by the authority patterns in British government and British political parties, at any rate if we accept the standard analyses of the latter by Beer and McKenzie.^[8] Both patterns consist of a curious and very similar mixture of democratic, authoritarian, and, so to speak, constitutional elements; this despite the fact that British government can be traced back to the eras of medieval constitutionalism and royal absolutism, while political parties are, in almost every respect, creatures of a much later period, the era of the mandate; and this also despite the fact that the formal constitution of the Labour party makes it seem very different from both the Conservative party and the British governmental structure.^[9] In both government and parties, the idea of the mandate is formalized and paid considerably more than lip service, in one case in the House of Commons, in the other in the Annual Conferences. In both cases, however, the leaders actually enjoy long tenure in office and a great deal of autonomy, even though the autonomy of the Conservative Leader rests on formal rules and that of the Labour Leader "merely" on actual practice; in both cases, moreover, the idea of the mandate is contravened by the fact that the leaders are widely expected to govern, in the sense of taking personal policy initiatives and sometimes even acting contrary to opinion in the rank and file. In both cases, too, this autonomy of leadership is mitigated by the expectation—on the part, incidentally, of both elites and masses—that authority will be exercised "constitutionally"; that is to say, that it will be exercised, if not in conformity with written documents, then at least within a framework of widely accepted and well-understood limits and rules, including, for example, the rule that authority inheres always in a collective structure, whether this structure is provided for in a formal constitution, as in the Labour party, or not, as in the Conservative party and the governmental machinery. In addition to these absolutely fundamental resemblances, there are also many less basic, although no less striking, similarities between

British government and parties. For example, both Parliament and the parliamentary organizations of the parties have largely an advisory and exhortatory role in decision-making processes; in both Parliament and the party conferences, the leaders are given certain traditional privileges in debates (they speak longer, for example, and more frequently); in both government and parties, bureaucracy plays an indispensable but subordinate and unusually self-effacing role, even in this age of massive government and massive parties; and, of course, government and party leaders entirely coincide, a fact that the Weimar parties show to be by no means inevitable in a parliamentary system.

The essential patterns of cabinet government, and the essential attitudes on which it is based, thus all have their counterparts in the major British parties. In fact, it has been argued, cogently I think, that cabinet government on the British model compels a certain correspondence between party structure and governmental structure, certainly while a party is in power, and therefore also perhaps while it is in opposition and presumably aspiring to power. Cabinet government could not otherwise work at all on a party basis; hence the anxieties of many English citizens when the Labour party strays, as it infrequently does, from the model of the cabinet system and acts as if it really believed in its formal constitution. But this argument should not be taken to mean that British parties cannot help but have a structure similar to that of British government. "Compel" does not in this case mean "cause." The argument means merely that British government can work smoothly only if such a congruence of governmental and party structures exists, not that things could not actually be otherwise. There are plenty of parliamentary systems in which the same logic holds, but few in which the same congruence can be found.

The most extreme and plainest form of congruence, then, is identity. Mixing metaphors, we might speak in this case of isomorphic authority patterns. But identity cannot exhaust the meaning of congruence when applied to social phenomena, for it is difficult even to imagine a society in which all authority patterns closely resemble each other. Certainly such a state of affairs is impossible in a democracy. Some social relations simply cannot be conducted in a democratic manner, or can be so conducted only with the gravest dysfunctional consequences. Take, for example, those social units which link different generations—families and schools. An infant cannot be cared for democratically, or a child brought up and schooled democratically. Families and schools can be permissive, but this is merely to say that they can be authoritarian in a lax and lenient manner. Families and schools can also carry on a certain amount of democratic pretense, and indeed more than pretense, and when they do so on a large scale, that fact is not without significance; but by and large they cannot carry such simulation and imitation of democracy to very great lengths, if they

are not to produce warped and ineffectual human beings. One of the most basic and indispensable functions in any social system, the socialization function, must therefore always be to some extent out of tune with democratic patterns and potentially at odds with them. The same point applies, almost as obviously, to certain relations among adults. We have every reason to think that economic organizations cannot be organized in a truly democratic manner, at any rate not without consequences that no one wants; and we certainly know that capitalist economic organization and even certain kinds of public ownership (like the nationalization in Britain of industries absolutely vital to the health of the whole economy) militate against a democratization of economic relations. The case of military organizations is even plainer in this regard, and the case of public bureaucracy just as clear. Again, there can be some simulation and imitation of democracy in firms, or public offices, or military units, but only within rather narrow limits. Precisely those social relations in which most individuals are engaged most of the time—family life, schools, and jobs (most kinds of jobs)—are the least capable of being democratically organized. To expect all authority relations in a democracy to be identical would therefore be unreasonable, and we could probably demonstrate the same thing, in other ways, for other kinds of governmental structures. In any complex society, but above all in democracies, we must expect some heterogeneity in authority patterns, even if we deal only with fundamental patterns and not circumstantial details.

In that case, however, one can still speak meaningfully of a congruence of authority patterns if the patterns have a certain "fit" with one another—if they dovetail with, or support, the governmental pattern, however indirectly. One way in which they can do this is by the partial imitation of the governmental authority patterns in other social structures. Democratic (or other) pretenses, if taken seriously and carried far, may have important consequences for the operation of the governmental structure, even though they are pretenses. Furthermore, structures like economic or military organizations may, in some cases, willingly incur certain functional disadvantages for the sake of acting out norms associated with governments in their substantive decision-making processes. For instance, capitalistic economic organizations, which play a great deal at democracy and permit certain deviations from the logic of the double-entry ledger in order actually to carry on certain democratic practices, may be said to be more congruent with democratic government than those that stick closely, both in ritual and process, to the economically most rational practices.

In view of this, we might be tempted to say that authority patterns are congruent if they have, not everything, but something in common. But if the equation of congruence with identity makes demands that are too great, its equation with mere resemblance, however slight, does not demand

enough. On the first basis, we shall almost never find a society in which authority patterns are really congruent; on the second, we shall assuredly not find any in which authority patterns are incongruent. However, by congruence I do not mean any resemblance at all among authority patterns. Where authority relations are not all highly similar, the term refers rather to a particular pattern of resemblance among them, one that makes stringent requirements, but not requirements impossible to fulfill—a pattern of graduated resemblances , so to speak.

To grasp the concept of graduated resemblances, one must think of societies as being composed of segments that are more or less distant from government. Governments themselves are adult structures, and for this reason families, for example, are more "vertically" distant from them, in terms of age levels, than schools, and schools more distant from them than purely adult structures. In the same way, adult structures may be "horizontally" segmented, so that some appear close to, others distant from, government. Parties, for example, ordinarily are situated closer to government than pressure groups; among pressure groups certain types may be particularly closely involved in government or parties; and all pressure groups are located more closely to government than nonpolitical organizations. These are very rough breakdowns; in some concrete cases, moreover, it may be difficult to make unambiguous distinctions, and the same social structures will not always fall into the same positions in every society. But none of this affects the definition: that social authority patterns are congruent, either if they are very similar, or if similarity to the governmental pattern increases significantly as one approaches the governmental segment itself.

On the basis of these explications of the term "congruence," we can now restate the first proposition of the theory. Government will be stable , (1) if social authority patterns are identical with the governmental pattern , or (2) if they constitute a graduated pattern in a proper segmentation of society , or (3) if a high degree of resemblance exists in patterns adjacent to government and one finds throughout the more distant segments a marked departure from functionally appropriate patterns for the sake of imitating the governmental pattern or extensive imitation of the governmental pattern in ritual practices . Conditions (2) and (3) are both, of course, looser and less demanding versions of condition (1); all refer to a basic need for considerable resemblance in authority patterns if government is to be stable, particularly in those segments of society which impinge directly on government. Condition (3) may be regarded, in this way, as the minimum required for governmental stability (and the minimum meaning of congruence), but perhaps the most that can be realized in relation to some particular pattern of government. By the same token, governments will be unstable (and the authority patterns of a society incongruent) if the governmental authority pattern is isolated (that

is, substantially different) from those of other social segments, or if a very abrupt change in authority pattern occurs in any adjacent segments of society, or if several different authority patterns exist in social strata furnishing a large proportion of the political elite (in the sense of active political participants). In the last case, congruence with the authority patterns of a particular part of the elite—say, a particular social class—may be quite possible, but congruence with the overall authority patterns of a society is logically out of the question.

Two Examples:
Great Britain and Germany

To make these propositions less abstract, let us look at two concrete cases that illustrate them: contemporary Britain and Weimar Germany.

We have already seen how closely the authority patterns of government and political parties resemble one another in Great Britain and that this resemblance helps to make effective the processes of cabinet government. One can similarly find great resemblances in authority patterns between British government and other aspects of British social life. As one moves away from the governmental segment these resemblances do decline, but never markedly or in a very abrupt manner.

For example, there is a quite striking resemblance in the authority structures of government and pressure groups, a resemblance also required for effective cabinet government, at least in this age of the social service state (as I have pointed out in another work).^[10] This resemblance is perhaps greatest in the case of groups constituted almost exclusively for political purposes (that is, in "attitude groups"),^[11] but it also exists to a surprising

extent in functional organizations, like professional and economic organizations, which go in for politics only as a more or less important sideline. Among such functional organizations, moreover, the resemblance to government is particularly great in groups most directly involved in governmental and party affairs, for example, trade unions, large-scale employers' organizations, cooperative societies, and the like. It is true that, on the whole, involvement by the group in nonpolitical affairs, such as economic bargaining, tends to decrease resemblance to the governmental pattern. Also, certain nonpolitical activities (for example, economic activities as against those of professional associations) act as inhibitions on too great an imitation of the governmental pattern. But throughout the whole universe of British pressure groups, resemblance to the governmental pattern is quite surprisingly great, even in organizations where, for functional reasons, one would least expect this.

British pressure groups tend to follow more or less closely, but ordinarily very closely, a certain ideal-typical authority pattern, which will immediately ring familiar to anyone acquainted with British government and parties. At the apex of this authority pattern, one usually finds a ceremonial figure who symbolically represents the group, perhaps presides over important meetings, and makes solemn speeches, but occupies a largely ritualistic and honorific position—although in a few cases ceremony and "efficiency" may be combined, as it is also in the leadership of the Conservative party. Under this figure, there generally is a council or executive committee, a collective body that usually exercises the real decision-making power, at least so far as higher decisions are concerned; more often than not this body is dominated, despite pretenses of collective decision making, by its chairman and a small handful of especially powerful members. Under the council, one generally finds a large number of other collective bodies—functional committees—with overlapping memberships; the dominant roles on these committees are played by the more powerful figures on the council. The functional committees and council are formally considered "responsible" to large annual conventions (a conference, or representative meeting, or whatever it may be called), which are always supposed to exercise the ultimate decision-making power, but generally act only as bodies acclaiming the leaders, occasionally criticizing them, and very infrequently making substantive contributions to policy; nevertheless, the leaders take care not to deviate too far from the public opinion of the convention and at the very least put up a show of responsiveness to the rank and file. Finally, there are a number of paid, full-time officials, a secretary and the assistants, who administer the organization's activities, play an extremely important role in regard to all affairs and an all-important role in regard to routine business, work closely and unobtrusively with the chairman and other important council members, generally get their way,

but know when to stay in the background and when not to press their views. Throughout this structure operate norms typical of all public authority in Great Britain: that decision making must be carried out by some sort of collective leadership, which is both responsive to the mass of the organization and, to a large extent, autonomous of it, which is expected to behave according to some sort of code well-understood in the group but normally not explicitly defined, and which, although resting to some extent on an elective basis, enjoys great tenure of office, often for as long as the leaders want to keep their positions.

In short, in the typical British pressure group we find the same mixture of basic authority forms that characterizes British government and parties, as well as imitations of less basic British authority practices, from the ubiquity of committees to ceremonial headship. This resemblance of the pressure-group pattern to the governmental pattern even extends to very minor matters, such as styles of debate at the annual conventions (directing remarks at the presiding officer, referring to other speakers in a florid and impersonal fashion, rarely delivering set speeches), or ways in which motions are introduced and processed, or ways in which orders of speakers are determined. We find this pattern, in almost every aspect, in a professional association like the British Medical Association,^[12] in a trade union like the National Union of Teachers, and in an employers' organization like the Federation of British Industries.^[13] Names may differ—the B.M.A. has an Annual Representative Meeting and a Council, the N.U.T. a Conference and an Executive, the F.B.I. an Annual General Meeting and a Grand Council—but the substance is the same, in structural forms and in actual practices.

In nonpolitical adult organizations, from friendly societies and clubs to business organizations, smallness of scale and functional considerations generally lead to significant departures from these forms and practices, but in the vast majority of cases one finds in these organizations at least a great deal of imitation, if only as ritual, of the governmental pattern. Even small-scale neighborhood clubs, like the many lawn tennis and social clubs that dot the landscape of middle-class England, generally have their committees and secretaries, their relatively inconsequential elections and stable oligarchies, their formalities, petty constitutionalisms, and ritualistic annual meetings; and still more is this true of larger clubs and friendly societies, like the famous snob clubs of Pall Mall and that holiest of holies, the Marylebone Cricket Club. It may be easier in such small organizations for elites to maintain themselves in authority, and also for members to participate in decision making, if they wish, but the essential forms and actual patterns of authority do not differ very much from the great political associations.

In business organizations, greater departures from the typical pattern

occur, but even in such organizations there is usually some acting out of the governmental forms (most conspicuously, much reliance on committees, both on the level of management and among workers). In business organizations, however, congruence with the authority patterns of the political organizations may be found mainly in seemingly inconsequential patterns of behavior, some of which have driven countless American efficiency teams, taking as their frame of reference behavior strictly "rational" in economic terms, to uncomprehending distraction—such as the tendency of British executives to keep civil service hours, and above all their tendency to carry on endless consultations instead of reaching quick and independent decisions. The general tone of relations among bosses and workers is also relevant here, although this is a matter difficult to deal with very explicitly. The typical British boss, like any other boss, is an authoritarian figure, but he is rarely an overbearing tyrant or a stern taskmaster. More often than not, he is a paternalistic authoritarian, in the general style of the British upper classes in relation to the British lower classes. It would not occur to him to be on intimate terms with his workers. Neither would it occur to him that his position might entitle him to treat his workers in an insulting manner, or to claim special privileges over them outside of the business organization, or that his functionally specific authority might extend to other aspects of his workers' lives,

Business organizations depart to a rather great extent from the governmental pattern, not only for obvious functional reasons, but also because of a fact just touched upon: they bring into close relation members of the upper and lower social classes. Wherever this occurs in British life, the authoritarian elements of the authority pattern tend to become enlarged and the democratic elements diminished. This is so not only in relations between economic bosses and their workers, but also between domestic servants and masters, enlisted men and officers, and members of the Administrative Class and other classes in the civil service. While all of these relations are governed by a high sense of propriety and functional limitation (by constitutionalist norms), those among members of the higher levels tend to be quite surprisingly democratic, or at least consultative and comradely; here again we might note the ubiquity of committees at every conceivable level in the higher civil service, the unusual use of staff committees in the military services, and the easy relations among officers of all ranks in military regiments, especially in elitist regiments like the Guards. But between members of the Administrative Class and their underlings, officers and their men, managers and their hired help, relations are highly nonconsultative and certainly not comradely; the observance of propriety and functional limitations in these cases is complemented by a considerable separation of individuals from each other, a general lack of contact among them for purposes other than functional ones.

All this is quite in keeping with a governmental pattern that is as markedly elitist as the British. In general, the governmental authority pattern is conformed to most in Britain in relations among members of the upper classes; it is conformed to less in relations among members of the lower classes and least in interclass relations. In other words, the more important the role in government that members of a social stratum are likely to play, the more their relations tend to be molded in the governmental pattern; and this is just what one would expect in a system like the British.

British schools and family life are at least partly responsible for this state of affairs. Family life in the lower strata is much more authoritarian than in the upper strata; there is less reasonableness, less consultation, less courtesy, less formality, more punishment, and more arbitrariness. Although even in the lower classes we can find no counterpart of the menacing paterfamilias of certain continental European families, the matriarchs whom Young and Willmott found in Bethnal Green or the arbitrarily despotic fathers depicted in working-class novels,^[14] like those of Lawrence and Sillitoe, have no upper-class counterparts. The same thing applies to schools. A marked change of atmosphere occurs as one moves from the secondary modern and technical schools to the grammar schools and public schools, especially the latter. Of course, even on the upper levels, British schools, like British government itself, tend to be rather authoritarian—more so, certainly, than schools in the United States—but relations among masters and pupils involve also a great deal of rather formalized good fellowship (in school games, for example) and a strict adherence to well-defined codes of conduct, while behavior among pupils is modeled to a remarkable extent on the political system. A case in point is described in Duff Cooper's book of memoirs, Old Men Forget . In Cooper's days at Eton, and possibly still today, debates were conducted in a manner obviously a carbon copy of House of Commons procedure, even though political questions were never discussed. A motion would be introduced, and any one of the students attending ("the House"!) might be called upon to captain the discussion; prior to "business," questions would be asked as in the Commons, and boys referred to one another as "honorable members" rather than by name. Not always do we find quite such faithful copies of governmental patterns in the higher educational institutions, but the governmental style is noticeable in almost every case; indeed, British politics has a readily recognizable style precisely because politicians are brought up to it practically from the time they wear rompers. Again, this is not to imply that relations among public school boys or grammar school boys are remarkably democratic or egalitarian; it would be absurd to call the prefect systems of British schools egalitarian. But then neither is British government remarkably egalitarian. And British schoolboys do enjoy an unusual degree of freedom from the direct supervision of their elders, a

great amount of self-government, even though it is self-government modified by relatively well-defined authoritarian relations among themselves.

British life thus illustrates the congruence of social authority patterns in all its aspects and degrees. There is a great resemblance between the authority pattern of government and those most closely adjacent to it, parties and pressure groups. Between government and those authority patterns that least resemble it intervene authority structures that are to a fairly large extent like the governmental structure. At no point in the segmentation of British society is there any abrupt and large change in authority patterns, and throughout one finds at least some imitation of governmental forms. And resemblance to the governmental pattern is greater in elite structures than in those of the nonelite. This does not validate the present theory, but it does support it in the case of perhaps the most stable of all modern political systems.

For support from the opposite end of the spectrum, we might look at one of the least stable of all modern governments, the Weimar Republic. How did the authority patterns of interwar Germany differ from modern Britain? Basically, in two ways. On the one hand, the German governmental pattern was much more one-sidedly democratic, at any rate if we confine analysis to the level of parliamentary representation and decision making and do not take into account the instrumental adjuncts of government, bureaucracy, the military, and the judiciary. On the other hand, social life, including life in parties and political interest groups, was highly authoritarian and relatively little "constitutionalized" compared with Britain. Not only were society and polity to some degree incongruent; they existed in unprecedented contradiction with one another. And on the basis of the theory of congruence, in consequence, the Weimar Republic could only have been what indeed it was: nasty, brutish, and short-lived—unless, like the Third Republic, it had been more lucky than any political order can expect to be.

Democracy, in interwar Germany, was, for all practical purposes, isolated at the level of parliamentary government, but at that level it was organized in an almost absurdly pure and exaggerated manner. Weimar Germany was governed by a Reichstag chosen on the basis of universal suffrage and by means of one of the purest systems of proportional representation ever devised. The chief of state was a plebiscitary president, and an effort was made, through run-off provisions, to assure that he would have the support of an absolute, not merely a relative, majority. Ministers were easily removable both by the popularly elected Reichstag and popularly elected president, and government was conducted on the basis of a very lengthy and detailed bill of rights. The Weimar constitution was proclaimed in its day as the most perfect of all democratic constitutions, and for good reasons.

This unalleviated democracy was superimposed upon a society pervaded by authoritarian relationships and obsessed with authoritarianism. In his study of interwar German films, From Caligari to Hitler , Siegfried Kracauer has pointed out that a morbid concern with despotism, with raw power and arbitrary will, was a characteristic alike of reactionary and revolutionary German films, most obviously in films like Caligari, Waxworks, Dr. Mabuse , and Mädchen in Uniform . That the Germans should have been deeply preoccupied with naked power, large and petty, is hardly surprising in a society democratized on its parliamentary surface, but shot through with large and petty tyrants in every other segment of life. Compared with their British counterparts, German family life, German schools, and German business firms were all exceedingly authoritarian. German families were dominated, more often than not, by tyrannical husbands and fathers, German schools by tyrannical teachers, German firms by tyrannical bosses. Insolence, gruffness, pettiness, arbitrariness, even violence were so widespread that one could certainly not consider them mere deviations from normal patterns. In a sense, throughout the whole of German interwar society one finds authority patterns that in Britain are confined to the nonelite strata, but even among the lower classes in Britain authority relations are not quite so arbitrary or so unrelenting as they were in Weimar Germany, nor afflicted by such omnipresent intimations of violence.

Families, schools, and occupational contexts are the most basic (that is, the most absorbing and demanding) segments of life, and the patterns existing in them are bound to affect all other social relations. But perhaps a high degree of authoritarianism in these patterns would not matter from the standpoint of democratic government if there were interposed between them and government certain institutions having mixed authority relations—institutions that might mediate between the pervasive despotism of the primary segments and the pure democracy of government, so that individuals would not be tossed abruptly from stark domination in one segment of life to stark liberty in another. But nothing remotely like this was the case in Weimar Germany. Political parties in imperial Germany had served as the principal model for Michels's iron law of oligarchy, and their internal political characteristics persisted in the Weimar period; at a later date, German political parties served as the chief illustration for Hermens's argument that proportional representation with the straight list system inevitably makes for a highly centralized and oligarchical party structure. Associational life in Weimar Germany presents, if anything, an even sharper contrast with Great Britain. Quite apart from the fact that the great interest groups were intimately involved in the party system—every major interest, economic, religious, or sectional, had a party of its own—the main associations offering men opportunities for escape from loneliness or from the primary social relations were extraordinarily au-

thoritarian in structure. Germany was not a country of hushed snob clubs and demure whist drives, of jolly good fellowship and darts in the public bar, but a country of paramilitary organizations, trade union militants, beer hall conspirators, grimly serious Turnvereine and systematically joyful Gesangvereine . This, of course, is something of an exaggeration. Not all German associations were highly authoritarian in structure, and we have reason to think that there were important differences among the various regions of Germany in this regard—southern Germany, for example, being on the whole less monotonously authoritarian than northern Germany. But while any simple picture of any complex society will exaggerate and overemphasize to some extent, the essential picture of pervasive authoritarianism in German secondary associations given here probably does so relatively little. At the very least, we can say that life in German associations was much more authoritarian than it is in the great majority of democracies. Associations thus formed no bridge between government and the primary and occupational groups; if anything, they formed a barrier to their reconciliation. Certainly it is in organizations like the Stahlhelm , or the Freikorps , or the SA and SS, that we find the greatest contrast to the pattern of plebiscitary government.^[15]

The same argument applies to the instrumental appendages of the parliamentary system, the civil and military service. Just as the imperial German parties had inspired Michels's theory of universal oligarchy, so the Prussian civil service, which persisted in form and to a large extent even in personnel under Weimar, had served as the chief model for Weber's ideal type of bureaucracy, with its emphasis upon "hierarchical subordination" and "the distinct social esteem" of the official. To these characteristics, which are by no means found in anything like pure form in other modern countries (e.g., Britain), we might add another that innumerable works on German bureaucracy have remarked upon: its recruitment from the more authoritarian elements in German society and its open sympathy, throughout the Weimar period, for reactionary and authoritarian movements.^[16] As for the military, so frequently has the antagonism of its patterns and attitudes to the Weimar system been pointed out that nothing more need be said on this subject.

We have in the case of the Weimar Republic a government violently contradictory to all nongovernmental aspects of life. However, it would not be strictly accurate to say, as so many have said, that the Germans were simply thoroughgoing authoritarians who just had no use for political democracy, that in Germany governmental democracy was imposed upon a country that provided no basis at all for it, so that the first talented and lucky authoritarian to come along could easily demolish the whole structure. There is every reason to think that the great majority of Germans were convinced democrats during the Weimar period and even before—

in their attitudes toward government. Imperial Germany was one of the first countries to have universal suffrage, and the fact that prodemocratic, liberal, center, and socialist parties consistently won somewhere between 80 and 90 percent of the vote in elections before World War I is therefore a matter of some importance. This voting pattern, furthermore, continued under the Weimar Republic right up to the ill-starred elections of the early 1930s; the right-wing Nationalists (the DNVP), the party that best fits the stereotype of the unmitigatedly authoritarian German, rarely polled more than 10 percent of the vote. Consequently, when one says that there was no basis for democracy in interwar Germany, one says something much more complicated than that Germans did not really want a governmental democracy. One says that Weimar Germany could provide no proper basis even for a governmental system that the great majority indeed wanted, while imperial Germany, ironically, did furnish a proper basis for a type of government the great majority did not seem to want. Had the German taste for authoritarianism been absolute, the Germans would probably have constructed after World War I a much more stable, though not a purely democratic, government. The trouble was not that the Germans were so one-sidedly authoritarian; the trouble was rather that they were—and perhaps had always been^[17] —so remarkably two-sided (i.e., incongruent) in their political beliefs and social practices. Profound ideological commitment to governmental democracy is not a sufficient basis for stable democracy; in fact, it can be worse, in the long run, than a more qualified commitment to democracy.

One more point regarding the incongruities in Weimar authority patterns may be worth mentioning. Although the constitution of the Weimar Republic was democratic to an unprecedented degree, it did contain some authoritarian elements. The executive, as in Britain, was given an unlimited power of dissolution; and under Article 48, the article that played such a villainous role in the legal destruction of German democracy, the president, with the cooperation of a chancellor appointed by himself, could wield enormous emergency powers, including the virtually absolute suspension of civil rights and a nearly unlimited power to govern by decrees. But these authoritarian elements of the constitution were reserved for particular periods when regular parliamentary processes no longer operated; contrary to the British case, they were not built into the everyday management of government. They were no part of the normal pattern of governmental authority. The constitution really left nothing to choose between absolute democracy in "normal" times—rare enough in the Weimar period—and an absolute lack of it in times of crisis. As a result, government was carried on not only in contradiction with society, but even in a son of contradiction with itself; periods of pure parliamentary democracy alternated at a rapid rate with periods of practically unchecked

executive dictatorship, much as in the last years of pre-Fascist Italian democracy and during the 1930s in France.

Motivational Basis of the Theory

In addition to the fact that it seems to fit the most unambiguous cases of stability and instability, the theory of congruence has, prior to any concerted testing, one other important point to recommend it: it leads one immediately to the motivational (or psychological) links between the variables it relates. This is important. Any generalization about human behavior obviously lacks an important element of plausibility if it makes no sense in terms of what we know about human motivations. Conceivably societies may involve certain purely mechanical relations that require no motivational explanation at all, but follow simply from the fact of interaction, whatever may underlie it; but even theories about competitive interdependence seem always to proceed from particular views of probable human conduct. In any event, to ask why certain relations exist in social life is always to ask what there is about one state of society that induces behavior leading to another.

Such motivational connections between different aspects of social life are by no means easy to see whenever one finds a positive correlation between social variables. Take the correlation between democracy and economic development. A positive relation of some sort seems to exist between the two, but why exactly should it? The only link which readily suggests itself is that a high degree of economic development leads to a high degree of economic satisfaction and thus reconciles people to their condition of life, including their government. But this does not tell us why it should reconcile them particularly to democracy, and, what is more, militates against the obvious fact that economic satisfaction is always relative to economic expectations—expectations that might conceivably outstrip any rate of economic growth and any level of economic development. This is not to say that there is no link between democracy and economic development at all, but only that the motivational link between them is not readily apparent from the correlation itself and probably very indirect, if indeed it does exist.

The motivational basis of the theory of congruence, however, is quite readily apparent. This is due to the fact that the conditions described by the term "incongruity" are very similar to the conditions denoted by two other concepts of social science that, appropriately enough, denote both certain social conditions and certain psychological states or propensities to act: the closely related concepts of anomie and strain.

Anomie exists, in its purest form, whenever there is a complete breakdown of a normative order governing action, when individuals lack clear and commanding guidelines to behavior, do not know what is expected

of them, and are thus compelled to rely solely upon their egos, their "rational" calculations, to inform their conduct. Anomie, in this sense, may be more or less acute and more or less widespread in a society; it may extend to few or many, important or unimportant, phases of life, and it may be found in society generally or only in certain of its members. It is always disturbing, but becomes, in its more acute form, unbearable; the actual responses to it depend, however, not only upon its acuteness but also upon the extent of its diffusion throughout a society. At its less acute levels, it manifests itself in merely annoying, possibly even constructive, anxieties, and the resort to perfectly innocuous means of relieving them. But in its more intense forms it has been linked, on the individual level, with serious functional disorders (even suicide), and, on the social level, with mass movements in general, particularly movements of religious fanaticism and political movements of a chiliastic and highly ideological character—in general, with movements that provide people with a sense of orientation, a sense of belonging to a bearable social order, or merely with the opportunity for escaping from the dilemmas of everyday life or submerging themselves in some comforting collectivity.

Anomie may result from many conditions. In individuals, it may be the result of inadequate socialization, or rapid mobility from one stratum of society to another, or transplantation from one culture to another, or indeed any important change in one's condition in life. In societies it may result from any social change (especially rapid change) requiring important adjustments in conduct, or from a widely successful attack upon traditional norms, or from large-scale mobility.

Among these many conditions that can give rise to anomie, the condition of "strain" is perhaps the most common in any complex society. "Strain" is used here in a technical sense; it refers, not to the utter lack of settled guides to behavior, or to ambiguous norms, but to ambivalent expectations—that is, the coexistence of different, perhaps even contradictory, norms of conduct in regard to a particular set of actions or an individual's actions in general.^[18] We may speak of strain whenever people are expected to conform to different, but equally legitimate, norms of conduct—as, for example, when an individual simultaneously performs some roles involving universalistic norms and others involving particularistic norms, or some roles permitting affective responses and others demanding affective neutrality; and strains are of course particularly acute if a single role makes contradictory normative demands. Strain thus exists for a doctor who is sexually attracted to a patient.

Incongruity between the authority patterns of a society, like any other incongruity among social patterns, is an obvious source of strain, and through strain of anomie, and through anomie of behavior potentially destructive to the stability of any pattern of government. This seems ob-

vious; yet one cannot let the argument go at that. Conflicting expectations inevitably exist in any highly differentiated society, and are perhaps given in the very nature of the human condition, for human beings are inherently multifaceted. One can no more imagine a bearable existence that is utterly devoid of affect than one that consists only of emotional responses; pure and complete universalism is unattainable, people being what they are, and pure and complete particularism leads to chaos and gross inefficiency in any kind of social life. It is not, therefore, the simple absence of strains that distinguishes an integrated and stable society from one that is unintegrated and unstable, but rather the successful reduction and management of strains that can never be eliminated. How then can strains be "managed," in order to prevent them from leading to acute anomie?

Perhaps the only reliable way is through the institutionalized segregation of roles—through preventing one role as much as possible from impinging upon, or even being mentally associated with, another, if the other makes conflicting normative demands. Such segregation of roles is in fact a feature of any society that is functionally differentiated to a high degree; it may be physical (note, for example, the fact that the doctor's office and domicile are generally separate) or psychological—that is, achieved only through the widespread mental disjunction of particular roles. Why then should not incongruities in authority patterns be similarly manageable through segregation? Why should it matter, for example, that authority is strong in one context and weak in another?

A number of things need to be said about this issue. First, the theory here sketched does not assert that any disparity among authority patterns is disastrous; it only asserts that disparities of a particular kind and degree have fatal consequences; I have already argued that the very notion of congruence encompasses certain kinds of disparities. Some disparate authority patterns can be tolerated well enough—anyway, without serious anomic consequences; the argument here is that incongruent patterns cannot be tolerated, partly because they are, by definition, patterns in which disparities are particularly stark and great, and partly because strains arising from incongruent authority patterns are not alleviated by "intermediate" patterns that help to reconcile the starkly disparate patterns in relatively distant segments of society. In any case, no one would argue that all strains can be managed equally well; and, obviously, the greater the strain, the more unlikely it is to be successfully managed.

Another point to bear in mind is that managing a strain is not the same thing as abolishing it; when we manage strains, we merely reduce them to a tolerable level, or, without reducing them, in some way accommodate ourselves to their existence. The strains, however, persist, and may at any time lead to behavior modifying the social relations that give rise to them. What is more, certain kinds of strains are hard to reduce or tolerate, and

strains among authority patterns, in my view, are of this type. The reason is precisely that authority relations are nearly universal in social relations, aspects of almost every social role, a fact that makes it inherently difficult to segregate authority patterns from one another. Whether a person is acting the role of parent, teacher, boss, or politician, that individual is almost always in some context involving authority; the operation of authority is one of the more inescapable facets of life. Not so for an individual playing both the affective spouse and affectively neutral professional, or both the particularistic parent and universalistic boss. In these cases, not only the structures but the functions also are different. There is, in other words, a crucial difference between performing different functions in different ways and performing the same function in different ways; conflicts of the latter son obviously impose incomparably the greater psychological burden. Imagine, on the one hand, doctors who are expected to take a coldly scientific attitude toward their patients and a warmly unscientific attitude toward their spouse; imagine, on the other, doctors who are expected to administer only to their wealthy patients but wash their hands of others, or who are supposed to help friends and kin but let strangers suffer, or to alleviate the pains of adults but not those of children. In the latter cases strain is bound to be the more severe, whatever grotesque value system might be used to legitimate such behavior—and functioning in incongruent authority relations is like the latter cases rather than the former.

This is not to say that an individual cannot rule in one context and be ruled in another. Although such a duality of positions does create strains which, apparently, are unmanageable by some people, most of us do just that most of the time. The question here is one of operating in conflicting authority patterns, not of occupying different positions in similar authority patterns. When one is subordinated in one pattern of authority, one may in fact learn very well how properly to be dominant in a similar pattern of authority, but being tossed back and forth among radically different authority patterns is another matter.

To some extent, however, it may be possible to segregate even very disparate patterns in the performance of the same or similar roles through highly rigid institutionalization. The likelihood of this is small, but not absolutely zero. Human ingenuity in ridding life of strains, or in creating the delusion that strains do not really exist, is very great. Political science itself furnishes a striking example of how far such delusions may go, in the long-held, and still carefully nurtured, belief that authority is characteristic only of the state, not of anything outside the state. This belief is quite modern in origin and has been dominant mainly in the liberal West. Why has it been dominant there? Has it become established only because of the development of political science departments in search of

some subject matter entirely their own? Or because of the effects of the theory of sovereignty? Or because of the gradual development of functionally differentiated, specialized structures of government? Or is it really a delusion that keeps liberal democrats from having to face up to the inevitable lack of real democracy in nongovernmental phases of social life? On the face of it, the notion that authority is a property solely of the state is so absurdly untenable that the compulsions making for it among eminently sensible people must be very great indeed. And that lends credence to the view that the notion performs neither an academic, nor a legal, but a psychological function: that it helps preserve the myth of democracy by keeping people from having to face up to, and incorporate in their political theories, uncomfortable disparities between governmental and nongovernmental life. In this connection, note that it has been mainly those out of sympathy with classical democratic ideas who have maintained a different position: pluralists, like Laski and Cole; Marxists, most obviously; and power theorists, like Lasswell and Michels.

Undoubtedly, individuals are able—and must be able—to make bearable, by all sorts of devices, even the most crushing strains. Under one condition, however, we can be sure that the chances of successful role segregation are not merely small, but practically zero: if nongovernmental social relations are themselves highly politicalized—that is to say, if they are greatly concerned with governmental politics. One of the characteristics of life under the Weimar Republic was that social formations, in addition to being extraordinarily authoritarian, were concerned, to an extraordinary extent and with extraordinary fervor, with matters of government; many voluntary societies—even some of the least of gymnastic or choral groups—seemed to have some sort of governmental ideology, or even some affiliation with political parties or movements. Where this is the case, it is obviously much more difficult, indeed impossible, to segregate discrepancies in expected behavior, for the simple reason that government itself cannot then appear as a segregated social context. And in a democracy some politicalization of nongovernmental life is always unavoidable, even necessary; how otherwise could there be any political competition? Democracy presupposes, at the very least, some organized party life, and party life tends to draw into politics all sorts of other social units; it diffuses governmental politics throughout society, now more, now less, but always to some extent, and thus makes the management of strains by role segregation particularly unlikely.

My purpose here is not to show that strains arising from incongruent authority patterns cannot be managed or tolerated under any circumstances. Rather it is to show that, among the manifold strains of life in a complex society, such strains are unusually, perhaps incomparably, difficult to manage, so that it seems logical for people to try and cope with them

by reducing them at the source, that is, ordinarily, by changing the governmental patterns under which they live. Even if this is not granted, however, one fact is surely beyond doubt: societies possessing congruent authority patterns possess an enormous "economic" advantage over those that do not. In the more extreme cases, like Great Britain, individuals are in effect socialized into almost all authority patterns simultaneously (even if they belong to the nonelite strata), while in highly incongruent societies people must repeatedly be resocialized for participation in various parts of social life. For society, then, congruence in regard to authority patterns, at the very least, saves much effort; for individuals, it saves much psychological wear and tear resulting from uncertainty and ambivalence. In that basic and indisputable sense, the congruent society starts with a great advantage over the incongruent society; and we have many reasons to think that it enjoys greater advantages still.

If it is true that incongruities among authority patterns can be reduced only at the source (in democracies, and perhaps other systems of government as well), and if it is true that such incongruities are particularly hard to bear, then all sorts of things that are otherwise mysterious become more comprehensible. To cite only one example: perhaps these points explain why in totalitarian systems such massive efforts are made to accomplish seemingly trivial, even self-destructive, ends, particularly efforts to reshape all social relations, from the family up, in the totalitarian image—or else to destroy them. "What rational balance sheet," asks Inkeles, speaking of early Soviet attacks on family and church, "would have led a group of leaders who were concerned first and foremost with preserving their power to attempt that particular diversion of energy with its obvious consequences of social resentment and popular hostility"?^[19] His own solution is that the Soviet leaders' actions can be explained only by their messianic visions of an utterly transformed society; the theory of congruence, however, provides a still more logical explanation, even if it was only felt in some intuitive way, not grasped in the sense in which it is stated here, by Soviet leaders.

But the subject here is democracy, not totalitarianism; and for the analysis of democratic government another important deduction can be made from the proposition that strains due to ambivalences in authority patterns can be relieved only by reducing the ambivalences themselves: there must always be some strains among authority patterns in a democracy. Such strains will be nonexistent only if a society's authority patterns are identical, a condition quite unrealizable in a democracy, as we have seen. Obviously, however, they can be kept at a relatively low level if the society provides sufficient opportunities for learning patterns of action appropriate to democracy (if, that is, a good many of the authority patterns in a democracy are significantly democratized); and this will be all the more the case if the

more congruent patterns are directly associated with government or involve mainly interactions among the political elite. And there is also a second possibility: strains might be kept within tolerable limits in democratic governments which have certain characteristics rather than others—there are, after all, many varieties of democracy.

This is a subject better dealt with under the second theory we require, a theory that specifies the particular conditions which must exist if democracy, rather than government in general, is to be stable.

Balanced Disparitis in the Governmental Pattern

What are the special conditions that democracies must fulfill if they are to be stable? Undoubtedly one could list here a massive catalog of social characteristics that favor stable democracy, but if we confine ourselves to conditions not merely favorable but indispensable, to necessary conditions, we can deal with this question rather briefly. The most essential special requirement of stable democracy can in fact be deduced from the theory of the congruence of authority patterns—as indeed ought to be the case, since special theories should always be derivable from general theories. If governments tend to be stable when social authority patterns are congruent; if a great many social relations cannot be organized on a purely democratic basis without seriously dysfunctional consequences; and if some of these relations resistant to democratic structure exist in social segments adjacent or close to government; then it follows that governmental democracy will tend to be stable only if it is to a significant extent impure—if, in short, the governmental authority pattern contains a balance of disparate elements, of which democracy is an important part (but only a part) .

To forestall a tempting, but unjustified, retort to this proposition, it should perhaps be pointed out immediately that I am not now asserting, after arguing that governmental stability requires congruent authority patterns, that stable democracy requires an incongruent authority pattern after all. The notion of congruence applies to the relations between governmental and nongovernmental authority patterns, while the present point relates only to governmental patterns. It asserts merely that intolerable strains between governmental and nongovernmental patterns are likely to be avoided if the governmental pattern is not extremely, that is, purely, democratic.

One could even argue that the very minimum definition of congruence cannot be satisfied in a pure governmental democracy. Not only are the primary and occupational relationships of social life inhospitable to democratic organization, but we have every reason to think that associations, parties, and pressure groups also resist democratization after a point—not to mention the still more obvious cases of the civil and military service.

Parties, for example, are competitive organizations—Michels called them fighting organizations^[20] —which can hardly afford the luxuries of plebiscitary democracy; so are pressure groups. Associations in general are rarely very democratic, if only because of the relatively low rate of participation by members in their affairs, in the great majority of cases^[21] —not to mention functional requirements that act as barriers against democratization. The iron law of oligarchy seems to hold pretty well, which is not to say that democracy must always be a chimera in any form. It follows that between a governmental pattern as purely democratic as that of Weimar Germany and any kind of social life, even in the least authoritarian of political cultures, there are bound to be glaring, perhaps insurmountable, disjunctions. Governmental democracy, of course, is never really pure in practice; but, as the case of Weimar shows, it can come close, certainly in the forms and myths of authority, if not quite so readily in actual practices.

It is certainly curious how often one finds mixtures of heterogeneous characteristics, sometimes even contradictory characteristics, in stable democracies. Take Great Britain once more as the obvious and most essential case in point. British government and British authority beliefs, as already pointed out, are a mixture of all the elements out of which governmental authority can be concocted: popular government, the government of an autonomous elite, and government under an impersonal law; and none of these elements is clearly dominant over the others. British government combines (in a surprisingly easy fit) authority, responsibility, and responsiveness—the dominion of rulers, rules, and the ruled; for this reason there is something in British government to which every aspect of British social life, whether permissive or compulsive, traditional or modern, can have an affinity. Nor is this mixture of basic authority patterns the only balance of disparities that British government contains. We find in it a similar mixture of ceremonial and "efficient" institutions, of sober business and gaudy show. We find in it also a mixture of integration and pluralistic competition. Authority is concentrated in the cabinet, but the cabinet, being a collective entity, is itself a pluralistic structure and, under modern conditions, functions less as a single, cohesive unit than through a large number of committees and subcommittees, formal and informal, permanent and ad hoc; normally this structure works slowly, by patient bargaining and consultation, like all polycentric structures, but the myth of the fusion of authority can accommodate on occasion swift and autonomous action, by a prime minister or an inner cabinet. So also the monolithic character of the disciplined mass parties is balanced by a highly pluralistic universe of active and influential pressure group, acting, by and large, directly upon government, rather than through the instrumentalities of the parties.

Not least, British politicians strike a most remarkable balance between

dogmatism and pragmatism in their behavior. They tend to be neither absolute ideologists (in the manner of most Weimar politicians) nor utterly unprincipled opportunists (in the manner of the "trasformists" of preFascist Italy); parliamentary behavior in Britain is characterized neither by unstable maneuvering for power nor by intransigent insistence upon ideals. K. B. Smellie has said that British politicians, while not unprincipled opportunists, are nevertheless "hard to work up to the dogmatic level"—suggesting that they are, on the average, men of principle, but lukewarm, phlegmatic, and tentative in the principles they hold. This, however, is not what I want to suggest at all. What is involved here is not a matter of temperament. The crucial point is rather that dogma and principle are concentrated upon some aspects of government and kept almost entirely out of others. In essence, the British invest with very high affect the procedural aspects of their government and with very low affect its substantive aspects; they behave like ideologists in regard to rules and like pragmatists in regard to policies. Procedures, to them, are not merely procedures, but sacred ritual. Neither the Tories nor the Whigs appear to have the slightest difficulty in stealing one another's clothes; yet in Britain any procedural reform, even the slightest and most sensible, seems to run into the most intractable and irrational obstacles. Massive evidence for this point is provided by the heaps of fruitless proposals on parliamentary reform, piled up both by private individuals (from Jennings to Crick)^[22] and by Select Committees. Herbert Morrison's Government and Parliament contains some particularly arresting examples, especially in regard to the complex, timedevouring, and needlessly repetitious financial procedures in the House of Commons. This explains not only why the British fit neither the extreme model of the Weimar ideologists nor that of the Italian "trasformists," but also why they do indeed sometimes behave as if they were ideologists in regard to policy. They do so when policy problems appear in procedural guise, as in the case of the House of Lords crisis of 1909–1911, and more recently (and not on the governmental level) in the Labour party disputes over nuclear weapons.

The theory of congruence gives us one reason, but only one of several, for the assertion that such balanced disparities are necessary if democracy is to be stable. Others are perhaps even more obvious. Democratic governments require a healthy element of authoritarianism not only for the sake of congruence between government and other aspects of society, but for the even simpler reason that a representative government must govern as well as represent—must satisfy two values that, on the evidence, are not easily reconcilable. This point can in fact be stated in a much less truistic form and tied in with the motivational basis of the theory of congruence sketched above. That strains in the definition of political roles can produce governmental instability seems manifest; but can governmental stability

also be affected by nonpolitical strains? There is at least a possibility that acute anomie, in whatever segment of life it may appear, will have destructive consequences upon government, and not only because it may alienate people from their general condition of life. If intolerable strains exist in some social relations, but not in authority patterns, it is particularly likely that attempts will be made to manage the existing strains through governmental means; it seems natural that individuals should attempt to modify unacceptable relations through whatever relations they accept. But if government, under these conditions, is incapable of realizing important special changes through political direction, it will itself become identified with the unbearable aspects of social life. Hence, a particular need for some measure of authoritarianism in the governmental pattern, or, to put the same point slightly differently, a high capacity to control social relations toward desired ends. This capacity is always necessary—it is inherent in the governmental function; but it is particularly necessary if society needs a solvent for nonpolitical tensions.

We could rest this point, and others, on an even simpler psychological basis: human beings just are not one-sided. They have superegos and libidos as well as egos. For that reason, as Bagehot so clearly saw, a government that provides no food for the sentiments, a purely businesslike, perfectly rationalized government, is unlikely to get very much support and likely to engender a rash of hysterical social movements to satisfy the instincts it neglects. And as Bagehot also clearly saw, a government that provides no clear and focused leadership, no easily recognized source of directives for conduct, is equally unlikely to succeed. The superego requires such a definite source of directives, even if only an imaginary one, just as the libido requires that it be invested with high affect; and a political system also requires it for operational purposes, above all in times of crisis. These two points may explain why constitutional monarchies seem to work better than parliamentary republics, and why presidential government seems preferable wherever monarchy no longer exists—two very old theories of political science.

We might speak in this connection of a second kind of "strain" that can afflict human behavior—strain that results from overconcentration on any pattern of behavior satisfying to only one aspect of the human psyche. Certainly we can regard the compulsive ritualism of political movements in the Weimar Republic as attempts to satisfy needs frustrated by the insipid matter-of-factness of the Weimar government; and we can regard the American penchant for giving officers of private associations stately titles (high potentate, grand marshal, sultan, and the like) in the same light. Indeed, one could argue that strain, in one sense or the other, is a condition inherent in human life, for anything that decreases strain in the first sense tends to increase it in the second. And if this is true, then it is clear that

a stable government is never one lacking in strains, but one that strikes a tolerable balance between disparate strains, between strains resulting from the existence of inconsistent norms and strains resulting from overconsistent norms. This lends support to the still older theory of political science that mixed government is the most stable form of government—and puts it also in a new light.

As for the need to balance pragmatism and ideologism in politics, an explanation not quite so manifestly relevant to the present theory is required. A high degree of pragmatic behavior in politics is obviously desirable because it makes possible political integration, either on the party or the parliamentary level. Because of the great multiplicity of interests existing in any modern society, both a two-party system and stable coalition government presuppose a great deal of compromise, and compromise in turn presupposes a relatively low degree of affect and dogma in political competition. Pragmatic attitudes in politics also help to keep political problems in proper focus and on a proper level of significance, and thus facilitate rational and relevant decision making. Political ideologies tend to elevate even the most minor matters to major importance, so that energies tend to be dissipated upon insignificant questions while great problems go begging for solution; they invest everything with "world significance," to translate literally an untranslatable German word. But if too much principle can kill a government, so can the lack of any principle at all. Spiro has pointed out,^[23] quite rightly, that too unprincipled an approach to politics will lead to a lack of consistent and long-range policies, to an eclectic flitting to and fro from policy to policy, which may prove destructive to a government by setting it internally at cross-purposes; what is wanted is "programmatic" government, and this must of necessity blend principle with pragmatic adjustment. One can add to this another, perhaps still more serious point. Pragmatic politics, absolutely unburdened by affect and principle, can lead to only one kind of political behavior: sheer opportunism—mere maneuvering lacking any higher purpose. That is what Italian politics was like before Giolitti's grant of universal suffrage, and, judging from the works of Mosca and Pareto, opportunism did more than anything else to discredit democratic institutions in Italian minds. Nor should one overlook that the most unprincipled and adaptable politician is likely also to be the most cynical (and effective) manipulator of symbols for unscrupulous ends; Richard Rovere's fascinating analysis of McCarthy provides a perfect case in point. And just as ceremony feeds the emotions, so principle feeds the superego—narrows the range of decisions requiring "rational" adjustment, and thus makes pragmatic reasoning a more tolerable burden. In this last sense, a proper balance between commitment and moderation is part and parcel of those manifold counterweights that preserve a tolerable balance of strains in social life.

Religion, Economic Development, and Mass Society:
an Appraisal Of Other Theories

No attempt will be made here to test the theory I have sketched directly by broad-scale, comparative analysis; but, as pointed out at the beginning, one can to some extent test a theory indirectly by viewing it in light of other, well-substantiated theories. If the latter can be subsumed to the theory proposed, that is certainly a powerful argument for it. If one can show on the basis of a proposed theory why certain other correlations between variables hold, that is a still more powerful argument for it. And if one can show that the theory not only explains such correlations but also explains why they are not better correlations, then one has as cogent a case as can be made for it, short of direct testing.

The propositions I should like to consider in light of the present theory are the following: (1) that countries with large, or predominant. Catholic populations do not produce stable democracy; (2) that a high degree of economic development correlates positively with stable democracy; and (3) that stable democracy correlates positively with vigorous associational life in a society and negatively with "mass societies," societies in which people are highly individuated or in which very strong primary group attachments exist. The first proposition has been stated in a great many sources, but perhaps most strongly and unequivocally by certain French writers, for example, André Gide and Raymond Aron. The second has been argued most strongly by Lipset.^[24] The third comes from Komhauser's The Politics of Mass Society .^[25]

These three propositions hold up very well; certainly they are the best propositions about the basic conditions of stable democracy available at present. Other propositions—for example, those relating stable democracy to integrated party systems or to a high degree of political consensus—may hold up even better, but deal with the subject on a much more superficial (or at any rate, a different) level. But however cogent, each of the three propositions leaves something to be desired when properly tested. For this, there may be several possible explanations. One possibility is that the phenomenon which the theories are meant to explain may be "multicausal" and that certain additional factors must be invoked in each case to make the theories entirely satisfactory. Another is that imperfections in the correlations arise because of the operation, in some cases, of accidental circumstances, the son of circumstances which, I have argued, led to the longevity of the French Third Republic. Here is also a third possibility, namely, that the conditions singled out in the propositions are not directly responsible for the stability or instability of democracies, but are linked to these only through some other condition—a condition that they either help to bring about or of which they are frequently a mani-

festation. My argument is that the last is indeed the case: that religion, economic development, and a society's structure of participation affect the stability of democracy only insofar as they affect or reflect relations among its patterns of authority. This is the preferable explanation of the imperfections of the three theories for several reasons: first, because in each case it can in fact be shown that the theories hold to the extent that they are special cases of congruence theory and fail to hold to the extent that they are not; second, because we can never tell when our subject matter intrinsically requires complex explanations (as distinguished from complex explanations that are required merely because we cling to highly imperfect hypotheses); and third, because it is very unlikely that fortuitous circumstances could account for the whole combination of conditions here identified as stable democracy, even if they can account for the mere survival of a political order.

Democracy and Catholicism

The proposition that countries with large, or predominant. Catholic populations tend not to be stable democracies certainly contains some truth, but the correlation between stable democracy and Catholicism is highly imperfect. For one thing, it is by no means established that predominantly non-Catholic countries tend to be markedly more stable democracies; it may be that stable democracy is simply very rare, no matter what a society's religious composition, in which case the failure of most predominantly Catholic countries to produce it indicates a condition pertaining to humanity, not just to such countries. Even granted that the most conspicuously stable democracies have existed in largely Protestant countries, some uncomfortable facts remain. Among the more stable democracies are several with very sizable Catholic populations, for example, Australia, Canada, Switzerland, and the United States. At least one almost exclusively Catholic country, Ireland, is a highly stable democracy. Present-day West Germany, with a government verging upon hyperstability, is much more highly Catholic than was the unified Germany of the Weimar Republic. And there are even some reasonably stable Latin American democracies, like Costa Rica and (until very recently, and perhaps still today) Uruguay.

Furthermore, most of the familiar arguments used to link Catholicism with unstable democracy (arguments used to explain the supposed correlation) simply do not stand up to close examination. Catholicism, being a highly dogmatic religion, is supposed to make for ideological intransigence in politics, yet nothing more pragmatic or accommodating can be imagined than political behavior in Italy before World War I or the attitudes of the French Center of the Fourth Republic described by Leites. Catholicism, being authoritarian in structure, is supposed to engender a preference for highly authoritarian government, yet in France it appeared

rather to engender a preference for near anarchy. Catholicism, by intruding clerical issues into political life, is supposed to prevent the consensus required in a highly integrated political system, yet nothing more consensual could be imagined than Italian politics in the postunification era.

It is reasonable to suspect, therefore, that if the imperfect correlation between Catholicism and unstable democracy does indeed represent some son of causal relation—if it is not just a spurious correlation—then Catholicism must frequently be associated with some other factor, which can also exist in non-Catholic countries, which need not invariably exist in Catholic countries, and which leads more invariably to unstable democracy than Catholicism itself. The moment the question is viewed in this way, it becomes apparent that the theories proposed here can make very good sense both of the correlation and of its imperfections.

Catholicism, by subjecting its adherents to a highly authoritarian relationship, and one which involves extremely powerful psychological sanctions, certainly does nothing itself to reinforce democracy. Potentially it always threatens to introduce an incongruity into a democratic society, so that in any democracy there exists a fair probability that Catholicism will have dysfunctional consequences. But no more than a fair probability. If a democracy is sufficiently hedged about with authoritarian elements, the incongruity between government and church need not be very great. Moreover, if the church does not play a very important role in the associational life of a society, or if between church and government there exists a multiplicity of less authoritarian associations, even associations with a religious tinge, the incongruities will be diminished, in the sense of being mediated and reconciled. Whether or not Catholicism will inhibit or undermine stable democracy depends, therefore, not only on the inherent characteristics of Catholicism, but also on the nature of the governmental pattern and the vigor (and particular characteristics) of associational life in a society. This is why there neither is nor could be a very dose correspondence between stability of democracy and size of Catholic population, and yet why there is some such correspondence.

Catholics certainly have no difficulty in acting as constructive democratic citizens in countries offering the chance to participate in a large variety of nonecclesiastical associations, particularly in countries like the United States and Switzerland, perhaps even Britain, where they account for a large but not predominant part of the population. It has been argued, of course, that in these countries Catholics are a minority (although a very large minority) and that this makes all the difference in the effect they can have on politics. Perhaps so, though obviously a group comprising from 20 to 40 percent of the population can affect the political life of a nation a great deal; but this hardly explains why Catholics themselves should be

relatively unaffected by Catholicism in these countries—why, for example, they should not themselves be markedly more authoritarian or ideological in political behavior, but fit instead so easily into the overall political culture of their society. Nor are Catholics a crashingly discordant element in what we might call "authoritarian democracies," like the Kanzlerdemokratie of the Bonn Federal Republic and the cabinet system of Britain. The real trouble lies in societies that combine a large Catholic population with a very pure democracy and in which nonsectarian associational life is poorly articulated. France of the Third and Fourth republics, for example, had not only a predominantly Catholic population but also one of the world's most unmitigated democracies and, relatively speaking, an unintensive associational life; such as it was, moreover, a good deal of its associational life was directly centered upon the Church. From this standpoint, the Fifth Republic ought to be a good deal more stable than the Third and Fourth, just as the German Federal Republic has been more stable than Weimar Germany. Certainly Gaullist France should be more stable than postwar Italy, a system that, under the understandable impression that pre-Fascist Italy was destroyed by Mussolini and not by its own weaknesses, had hedged against the possibility of excessive executive authority much more than it had provided against an insufficient amount of it. It is, of course, too early [in 1965] to assess the stability of the Fifth Republic, but just because of that—and because of the inauspicious circumstances of its beginning in near revolution and charismatic rule (always the most transient form of rule)—it provides a good test case for the future assessment of the theory.

A Catholic country, in short, requires a democracy more modified by authoritarianism, and perhaps also by unquestioned "principle," than countries in which certain other religions predominate. Yet Catholic countries are also less likely to possess such modified democracies than others. Where the Church has been powerful enough to resist reform, it has generally also been powerful enough to delay liberal government and to assure that revolution would be required to establish it. Nowhere has the Church been able entirely to withstand the liberal wave. But in many places, by delaying change, it has contributed to the abrupt imposition of highly advanced democracy upon highly traditional societies and, by requiring revolutionary change, prevented that blending of modern and premoderm forms that is the secret of stable democracy in Britain; also, by requiring revolution, it has prevented the persistence of unquestioned principle in the course of major change and instead made everything subject to "rational" debate. Catholicism thus has tended to prevent precisely those conditions that might make it compatible with stable democracy. This, even more than the inherent disparity between Catholic and democratic patterns, explains why Catholic countries so often (and yet with such signal exceptions) are unstable democracies.

Democracy and Economic Development

If there is a close connection between democracy and economic development—if stable democracy and a high degree of economic development are positively correlated—then still another link may exist between the stability of democracy and the religious composition of society, for the connection between religious attitudes and economic development seems now to be well established. But is there really a close connection between democracy and economic development?

The evidence suggests strongly that there is. Upset, for example, found a positive correlation of sorts between stable democracy (in his sense of the term) and almost every conceivable index of economic development. Average per capita income in European and English-speaking stable democracies he found to be (in 1949 and 1950) $ 695, in unstable democracies only $ 308. Stable democracies had one doctor for every 860 persons, unstable democracies one for every 1, 400 persons. There were 17 persons per motor vehicle in one case, 143 in the other; on one side, there were 205 telephones, 350 radios, and 341 newspapers per 1,000 persons, on the other only 58 telephones, 160 radios, and 167 newspapers. Stable European democracies had a male agricultural population of 21 percent, European dictatorships one of 41 percent. No less than 43 percent of the population in stable European democracies lived in urban areas (cities over 20,000), but only 24 percent did so in European dictatorships. And so on. These findings are supported by Coleman (in the Conclusion of The Politics of the Developing Areas ), countries with "competitive" political systems in the developing areas coming out much better, on the indices used also by Upset, than "semi-competitive" and "authoritarian" countries.^[26] And Deutsch's^[27] work also lends credence to the thesis, since the great majority of the countries located at the highest extreme of economic development in his "country profiles" are also stable democracies—for example, Switzerland, Australia, New Zealand, Sweden, Norway, Denmark, the United Kingdom, and the United States.

This is a lot of evidence—but far from conclusive. If we look, not at overall averages, but at specific cases and at ranges of economic differences among countries in the same governmental categories, it becomes apparent that we have here a rather weak correlation, not a strong one. A few countries that have been anything but paragons of democratic stability rank quite high in economic development. France, Venezuela, and the USSR (not to mention Kuwait) rank high now, and Weimar Germany did in its own day. It is true that the half-dozen countries that are most highly developed are also stable democracies, but difficulties seem to arise only a very little below their level. And the ranges between the most and the least developed stable democracies, as well as the most and the least de-

veloped unstable democracies (and nondemocracies), are enormous. The least developed stable European democracy had, according to Upset, an average per capita income of $ 420, the most developed European dictatorship one of $ 482. Stable democracies have as many as sixty-two persons per motor vehicle, dictatorships as few as ten; one has as many as 46 percent of males employed in agriculture, the other as few as 16 percent; in one case as few as 28 percent of the population live in cities, in the other as many as 44 percent. Such overlaps, varying in size, but all large, are found for every index.

It follows that between the great extremes of economic development and economic underdevelopment is a large no-man's-land where apparently any governmental order, from stable democracy to totalitarianism, can exist—and at the great extremes there are not many cases. That stable democracy is not found at the very lowest levels is, of course, not surprising. At that level of underdevelopment the more obvious requisites for any democracy at all, stable or unstable—a certain minimum level of literacy, information, and communications—are not satisfied. But beyond that extreme, in the higher and intermediate ranges of economic development, everything is puzzling and indeterminate, if one assumes a simple correlation between democracy and economic level.

Lipset himself seems well aware of this, for his theory does not stop with economic development, but also uses a large number of other factors to explain the stability or instability of democracies: rapidity (rather than level) of economic development; "legitimacy" (that is, acceptance of a regime as right for a society), with special emphasis on the support of conservative groups; religion; historical development; and governmental structure. By adding all these conditions to qualify the correlation with economic development, the theory can be made to come out right in every case, but so could any theory whatever; the whole procedure smacks of the familiar methodological fallacy of "saving the hypothesis." Moreover, the theory that finally emerges from all these modifications is extremely complicated, to say the least. It does not blatantly violate the rule of parsimony, for it is not established that a simpler theory with equal explanatory power is available; but it is very cumbersome to use and the significance of any of its variables almost impossible to establish, owing to the number of variables alleged to be independently capable of producing unstable democracy.

For these reasons alone one ought to look for another way to deal with the imperfections of the correlation, and such a way is provided by the present theory. Like religion, we can regard economic development as an aspect of society that correlates with stable (or unstable) democracy only insofar as it has an impact upon the congruence of authority patterns in

society and the balancing of disparate patterns in government—as it does frequently (hence the positive correlation) but certainly not always (hence its imperfections).

That this is probably the right way to proceed is suggested by the most important modification Upset himself makes in his theory. Not only the level of economic development seems to him to be associated with the performance of democracies, but also the rate of economic development; very rapid economic development, he argues, has consequences inimical to democracy, but more gradual economic development supports it. On the evidence, this is a very strong argument—and exactly what one would expect on the basis of congruence theory. Rapid industrialization from a traditional base, whatever its long-run effects, obviously introduces into society profound incongruities, particularly if it is imposed by direction, in one form or another. If economic development is relatively slow, strains between the economic and other sectors of life can be kept at tolerable levels by the gradual adaptation of other social institutions to the slowly evolving industrial economy. Nor is this the only reason why rapid economic development has political consequences different from gradual economic development. In the course of rapid economic development, the social order (appearing, as it must, as a barrier to development) is usually attacked in all its aspects, and in that process even aspects of social life which are, if not perfectly, then at least sufficiently, compatible with a modernized economy may be uprooted. Even if social life is not coercively uprooted by those who want to industrialize rapidly, it may nevertheless be uprooted by revolutionary violence resulting from the great strains attendant upon rapid industrialization. In either event, the outcome is the same: society and government will not consist of that mixture of modern and premodern patterns, that blending of disparities, which is characteristic of the more stable democracies. Furthermore, rapid industrialization usually occurs, for obvious reasons, at a relatively late stage in the history of industrialism; because of this, the desire for rapid economic modernization is likely for manifest historical reasons to be accompanied by expectations of relatively advanced democracy. Rapid industrialization in nontotalitarian countries will therefore tend to be accompanied by the creation of especially pure democracies, and these are always more tenuous than impure democracies; and if democracy is achieved by revolution, the result is, of course, the same. Worst of all, the sudden creation of an advanced democracy will abruptly liberate men politically, while the exigencies of rapid industrialization will subject them, in the short run, to unprecedented disciplines and compulsions. In this way, rapid industrialization not only unsettles the social order in general, but tends to create particularly great strains between government and other aspects of social life.

The theories of congruence and balanced disparities can thus account comprehensively for the deleterious impact upon democracy of rapid industrialization. But what of the positive correlation between economic development and stable democracy in general? If rapid industrialization is indeed inimical to democracy, and if it leads, as it must, to the attainment of relatively high levels of development in relatively little time, then surely this correlation is rather puzzling—until one remembers that the most advanced industrial societies (at present, or at any rate when Upset collected his data) are precisely those which developed industry most gradually but still lead the field because of their early start. In these countriesGreat Britain, the United States, Switzerland, Sweden, and Canada—industrialization was associated with the relatively slow growth of populistic democracy and the gradual adaptation, never complete displacement, of preindustrial patterns, or else the forming of society itself simultaneously with the growth of industry and democracy. When this is realized, it becomes apparent that level of economic development, the moment one goes beyond extreme underdevelopment, matters only because speed of economic development matters; and from this it also follows that the present theory can account for Upset's general proposition, no less than for the supposed modification of it which he introduces. This is because the modification is, in fact, the really crucial theory.

Economic development thus correlates very imperfectly with stable democracy because certain forms of development produce precisely those incongruities and imbalances that endanger democracy. That is why some very highly developed countries are unstable democracies or not democracies at all. But economic development need not produce these deleterious consequences and has not produced them in the (currently) most highly developed countries—that is why there is, nevertheless, a positive correlation between level of economic development and stable democracy. Indirectly, in the ways Upset mentions, industrialization may even create conditions favorable to democracy, provided it is achieved in the right way. It is even more likely, however, that the two are associated simply because their roots lie in the same historical conditions—that is to say, because the processes of industrialization and democratization coincided and did not, while both were being achieved gradually, frustrate one another.

If this is so, then it would clearly be unwise to base on the correlation any optimistic predictions about the future impact of industrialization upon the stability of democracies. The only predictions that seem to follow from this analysis are that late industrialization is, in the typical case, likely to have political consequences directly opposite to those of early industrialization and that rapid industrialization is particularly dangerous to democracy if, paradoxical though it may seem, it is accompanied by rapid democratization.

Democracy and Mass Society

The third theory I wish to consider has now become associated mainly with William Kornhauser—although Kornhauser himself acknowledges the great role played in its formulation by such diverse people as Jacob Burckhardt, Gustave Le Bon, Ortega y Gasset, Karl Mannheim, and, above all perhaps, Emile Durkheim, Emil Lederer, and Hannah Arendt. This theory attributes unstable democracy primarily to "mass society," a technical term denoting societies without a vigorous corporate life, societies in which elites have "direct access" to nonelites, and nonelites to elites, owing to the weakness or absence of organizations mediating between government and primary social relations. Such societies leave people "naked" before the state, and the state naked before people, unlike "pluralistic" societies, in which sovereign and subject are separated by many autonomous social formations, through which each must act upon the other.

With this theory we can deal rather more briefly than with the others, not because it is transparently bad, but, on the contrary, because it is so very well established. The theory of mass society is by all odds the best substantiated of the theories we have considered and just because of that requires the least reinterpretation. Kornhauser supports it with truly massive evidence—evidence which shows not only that the more unstable democracies fit the model of mass society but also that the most apathetic and extremist antidemocrats, even in stable democracies, are found among the most "socially isolated" members of all classes and walks of life, among the "nonparticipants" who exist, in greater or smaller numbers, in any society. Kornhauser, in other words, not only links variables on the social level, but also, by looking at individual behavior in a great variety of contexts, shows that these relations have a clear and definite motivational basis. All that is necessary here, therefore, is to show that the theory of mass society is intimately related to the theory I have sketched—indeed, it is almost the same—and that the very slight imperfections which exist in it are attributable to the very slight ways in which it departs from the present theory.

That the two theories are closely related should be manifest without long explication. If it is true that democratic government is inherently and profoundly in tension with the primary and occupational relations of social life, then it follows from the present theory that vigorous interactions upon the "intermediate" (or secondary) levels are required to produce congruity of authority patterns in democracies. Only a wide variety of intermediate associations can keep the inherent contradictions between democracy and life at the primary social levels from directly leading to insupportable strains; only intermediate associations can provide opportunities for learning democratic behavior patterns.^[28] This is even more important than the fact, of which Kornhauser makes a great deal, that such associations act

as buffers against overbearing rulers, for while this fact may explain why vigorous associational life can be a barrier to totalitarianism, it does not explain why it should be particularly compatible with stable democracy; after all, impeding totalitarianism is not at all the same thing as facilitating stable democracy—there are unstable democracies and nontotalitarian autocracies as well. Kornhauser's emphasis upon the protective effect of secondary associations seems therefore to miss the really crucial relation between associational life and stable democracy—except in one sense. Kornhauser argues that associations not only act as buffers shielding the ruled from their rulers, but that they also work in the opposite direction: they provide the rulers with a certain autonomy from the ruled, while still subjecting them to exigent social pressures. If this is so, and it seems cogent, then it is apparent that a vigorous associational life not only permits authority patterns to be congruent, but leads also to one of the balanced disparities that democracy seems to require—a balance between governmental autonomy and dependence, "democracy" and "authoritarianism."

The two theories thus closely fit one another. Do they in any way differ? The answer is that in one way they clearly do: if a society has a vigorous associational life, but if the associations themselves are highly undemocratic, then, upon my theory, democracy should not be stable, and upon Kornhauser's, it should (at any rate, one condition making for unstable democracy—large-scale, illegitimate, antilibertarian movements—should not exist). It is, of course, perfectly possible for intermediate associations to be as undemocratic as primary groups, or even more so. It is improbable, but not impossible—and it has happened. Weimar Germany is itself a case in point, though admittedly there are not many others, for intermediate associations resist absolute authoritarianism nearly as much as they resist utter democracy. But the Weimar example suggests that, insofar as the two theories differ, congruence is the more powerful—for it is just in improbable cases like Weimar Germany that the theory of mass society fails to hold.

In this connection it is worth noting that, in a book teeming with apposite evidence, data relating to Germany are conspicuously sparse and inapropos. Some of the evidence intended to show unintensive associational life in Germany actually relates to the Bonn period;^[29] some of it shows that German associations tend to be engaged in politics to an unusual extent, not that there is a scarcity of associations or of participation in them;^[30] and the rest either shows what is true of all societies, that participants are disposed more favorably to democracy than nonparticipants, or that certain other conditions unfavorable to democracy existed in Weimar days. In actual fact, life upon the secondary levels was very vigorous in the Weimar period—and very incongruent with democracy. One might perhaps argue that large-scale participation in illiberal organizations was

itself a response to social atomization, but this would be transparent sophistry, for it implies that illiberal associations simply do not count when one evaluates the extent to which a society is atomized, a position for which there can be no possible justification. If only mass movements like the Nazi and Communist parties had existed at the intermediate levels in Weimar Germany, such an argument would be sensible enough, but in fact one finds at those levels a great many organizations—most of them of a piece, from the standpoint of authority relations.

Conclusion

However easily other theories can be subsumed to, and even improved by, the theory I have proposed here, it should be admitted that at this stage the theory is still little more than plausible. It is deficient, most obviously, in that no rigorous validation of it has been undertaken; the data used here illustrate the propositions, but do not validate them. Moreover, before the theory can be adequately tested, much more theoretical work is required. Testing the theory will not be just a matter of looking at the relevant data. For one thing, to test the theory we need a much better set of categories for classifying authority patterns than we now possess. Terms like democratic, authoritarian, and constitutionalist may do in the initial formulation of the idea, but for any large-scale comparative analysis a more discriminating and less ambiguous set of categories seems required. Such categories ought to be applicable to any kind of authority pattern, not just to government, a point that rules out many of the typologies now in use in political science. It is also necessary to find ways to measure with some exactness the extent to which different kinds of authority are present in actual cases, so that degrees of correspondence and disparity between authority patterns can be precisely established and, even more important, so that the very idea of congruence can be given a more precise formulation. In addition to measurable categories of authority, we need such categories for different types and degrees of instability; we need to relate types and degrees of instability to types and degrees of incongruence and imbalance; and, not least, we should be able to establish direct connections between incongruence and imbalance of authority patterns and the obvious conditions and system failures associated with unstable democracy.

Apart from these tasks, a staggering amount of empirical work still seems required before validity can be claimed for the theory. Because of the traditional emphasis on the state in political science, and on nonpolitical matters in other social sciences, studies of authority in general social life are pitifully meager and, for the most part, inadequate. Political scientists have charted little more than the visible part of their world; the massive

infrastructure of political life is still largely terra incognita . This applies even to the data given here about authority in British and German life. By and large, these data are derived from my own and other people's personal impressions and from allusions to the subject in works not directly concerned with the rigorous and intensive investigation of authority relations.

In short, almost everything needed to establish the theory still needs doing, and an enormous amount of work is required if these things are to be done. Hence the claims made for the theory here are modest. Nevertheless I do make some claims for it. I claim that it is at least plausible; I claim that it fits what is known about the more obvious cases of stable and unstable democracy; I claim that it can make still more convincing propositions about stable democracy which are already convincing to a degree; and I claim that it not only relates variables but gives immediate insight into the motivational forces which link them. Not least, I claim that the theory can account for a crucial fact not yet touched upon—perhaps the most important fact with which the study of democracy confronts us. This is that stable democracy is immensely difficult to achieve and has in fact been achieved only in very few cases—that it is unstable democracy, not stable democracy, which is, by any reasonable measurement, the "normal" case.

Every aspect of the essay suggests how very tenuous and improbable are the foundations of stable democratic government. Stable democracy requires congruence of authority patterns between government and segments of social life that resist democratization; it requires balances of contradictory behavior patterns, in such a way that the balances do not lead to undue strain and intolerable anomie; it requires a certain similarity among authority patterns, but not to the extent that basic human needs are thwarted. Surely, if these conditions are indeed required, then the existence of a stable democracy requires much explaining, while unstable democracies practically explain themselves. Perhaps this is why writers like Lipset find so many factors that can of themselves prevent stable democracy and none that can of themselves assure it. What, after all, is a complex multicausal explanation if not an assertion that a great many factors must coincide if a phenomenon is to exist—and that its existence is, for that reason alone, improbable?

Given the evidence of our times, this seems a most appropriate conclusion. The age when we had some reason confidently to expect the universal reign of democracy (and, more than that, the end of fundamental political change, because of a supposedly inherent affinity of human nature to democracy) ended with Bryce—and the trials of Weimar. For our own world we need a more pessimistic approach to democratic government, one not based upon the bland assumption that people are natural dem-

ocrats, but one that directs attention to those calamitously improbable combinations of circumstances which actually make democracy work.

For the present purpose this is certainly the most important deduction that can be drawn from the theory. But still larger implications, which might be briefly indicated, follow from it. One could argue, for example, that the conditions which make stable democracy improbable also make stable totalitarianism improbable: not only is totalitarianism a very "pure" governmental system but also one fundamentally incongruent with certain functionally indispensable social relations. By the same token, semiauthoritarian government—government in which strong, preferably ascriptive, authority is mitigated by adherence to impersonal rules, paternalistic benevolence, institutionalized channels of representation, and a vigorous corporate life—is bound to be the most stable of all possible governments, owing precisely to its impurity and easy congruence with primary group patterns. This corresponds to Weber's view that "traditional" authority is the most stable of his three ideal types.

But still more convincing would be the deduction that, while degrees of stability, or rates of change, may vary, no form of government can be inherently stable—if only because no form of government can escape the dilemma of managing some strains by increasing the probability of others. The present theory leads logically to an inherently dynamic conception of government, in which strains (and, through strains, change) are the rule rather than the exception; stable governments, upon this view, are the product of "accidental" (extremely improbable) conjunctions of conditions that do sometimes, but rarely, occur in actual societies. Such a dynamic view of governmental change—a view positing change as normal and constant, but, unlike Marxism and other historicist theories, specifying no concrete content or goal for the process of change—seems, on the face of it, more in tune with the manifest testimony of history than theories that take stability for granted and attribute change to external, rather than immanent, conditions. But this takes us far beyond the subject of stable democracy and obviously requires a more thorough treatment than can be given here.