Method in the Madness of History

"Pienso en las cosas," wrote Jorge Luis Borges in Historia de la Noche, "que pudieron ser y no fueron": I reflect on things that could have been but never were. His poem mentions the treatise on Saxon mythology that chronicler Bede never wrote, history without Helen of Troy's face, the American Confederacy's victory at Gettysburg, the unicorn's other horn, and more. Without obvious effort, Borges' s poetry vivifies anagrams, nursery rhymes, shopping lists, and descriptions of ordinary objects -- trivia, one might wrongly say. Wrongly, because things that never happened are profoundly important.

We could add to Borges' s list a Mongol empire that conquered the rest of the 1 2th-century world, surgeons condemned to their academic cousins' genteel poverty, cities without wheels, households in control of their own armies, fertility decline in the absence of mortality decline, money bereft of commodities. This roster differs from Borges's: All these things have actually half-happened somewhere. The Mongols did conquer, however temporarily, much of their known world. Surgeons in Mexico or Cuba occupy economic positions more similar to professors than they do in the United States, cities of the Ottoman Empire before 1900 existed in the near-absence of wheeled vehicles, and so on down the list.

The half-happening of these events provides the rationale of history, at least in its explanatory mode, as well as of social science. Social science is the systematic study of what could happen, what could have happened, what will possibly happen, in human social life, and why. History places that systematic study firmly in time and space, insisting that where and when something happen affects how it happens. Both differ from poetry, science fiction, and philosophy by grounding their thought about the nonexistent in verifiable analyses of what has so far been possible, or even occurred. But it still concerns that which is not, was not, or will not be.

The genius of historical explanation concerns principles of change and variation that show nonexistent conditions to have been possible, and perhaps still to be possible. Barrington Moore, Jr. titles Chapter 11 of his book Injustice "The Suppression of Historical Alternatives: Germany 1918-1920." The chapter asks under what conditions a durable democratic revolution could have occurred in Germany. If the notion of suppressed historical alternatives, he says,

Precisely. In the spirit of Barrington Moore, let us reflect on methodological problems in the actual preparation of historical explanations. In order to take counterfactual explanation seriously, we must identify the fields of variation within which the phenomena occur, invoke explanatory principles appropriate to those fields of variation, then locate the concrete phenomena under examination precisely within those fields, taking into account the effect of time and place on the causal processes involved. In this paper, I will briefly distinguish different levels at which methodological problems complicate historical explanation. Then I will then concentrate on just two of the many problems we might discuss: first, the seductions of skepticism, second the perils of invariant explanatory models.

The problems appear at three levels we might label as Metamethod, Methodology, and Methodical Maneuvers. Metamethod refers to the presumptions, often spoken, on which historical analysis rests. We might think of it as responses to questions such as these:

To what extent, and in what ways, can historical analysis generate reliable explanations of social phenomena?

To what degree must empathetic interpretation, as distinguished from causal accounting, lead us to such explanations?

At what scale(s) do coherent historical processes operate: individuals, eras, whole societies, all of human experience?

What are therefore the proper starting-points for historical analysis: individuals, groups, social relations, civilizations?

In what regards, and to what extent, then, do time and place limit the scope of possible historical explanations?

I will claim that historians divide sharply in answering these questions, with significant consequences for their day-to-day practice.

I will also claim that both historians and social scientists have commonly made serious errors in supposing that explanations must begin with the presumption that self-motivated monads -- whether individuals, groups, or societies -- constitute the essential historical actors, in making explanation chiefly or exclusively an act of interpretation, and in imagining that explanatory principles consist of invariant, timeless generalizations about categories of structures or processes; they have, in short, adopted the following principles:

I will counter these principles by recommending historically-variable causal accounts of social relations.

By Methodoloy we might understand the basic logics of historical inquiry that make sense within the limits set by metamethods: hermeneutic construction of motivated narratives, case comparison, invocation of general causal principles in the construction of particular accounts, searches for principles of variation in time and space. Texts, courses, and polemics on historical method often touch on Metamethod and Methodical Maneuvers, but they commonly concentrate their fire on such methodological questions. Under the heading of Methodical Maneuvers we can group the variety of empirical and analytical techniques historians employ in gathering, analyzing, and presenting their evidence: collective biographies, systematic catalogs of events, formal analyses of sequences, treatments of textual content, formulation and estimation of econometric models, network analyses, and so on. Although like other historians I spend much of my daily effort on Methodical Maneuvers, attempting to squeeze reliable information from stubborn sources, here I concentrate on Metamethod, since it bounds the possible worlds within which particular methodologies and analytic techniques make sense.

To consider such an effort worthwhile, we must suppose that it will create reliable knowledge. The postmodern wave that has washed over history during the last decade or so has raised serious doubts about the possibility of cumulative, reliable knowledge, especially on the ground that all purported knowledge passes irretrievably through language's arbitrary conventions. Postmodern skepticism has made realists like me more aware of the philosophical foundations of our ideas concerning agency, identity, language, social processes, and historical knowledge. I keep hoping, indeed, that we will have a genuine confrontation between fully-formed arguments for and against historical realism before the great wave simply recedes, leaving nothing but rubble on the beach.

Class analysis provides one possible setting for such a confrontation. In 1963, E.P. Thompson roared onto the terrain of class analysis like an invading army. Descending from the heights of literary criticism and biography, he daringly attacked on two fronts, machine-gunning mechanistic Marxism at the same time as he cannonaded conservative condescension. At least for England from the 1780s to the 1830s, he swept the field, persuading a wide range of readers that something he called the "making" of a working class occurred through a sustained series of struggles and convincing the rest that they now had a new, seductive leftist thesis to combat. With a literary historian's panache, Thompson mustered an extraordinary range of evidence for his thesis, drawing connections between political philosophy and popular culture, enormously broadening the conception of relevant texts, giving popular utterances and crowd actions a literary standing they had rarely achieved before. His victorious vision of class formation in England inspired numerous historians of other western countries to search for parallel constructions in their own territories and periods, so much so that the phrase Making of the Working Class acquired the immortality of a cliché

Like European appropriation of Asian and African territories, Thompson's conquest of British class analysis laid down a terrible burden for his successors. Just as anticolonial leaders once felt obliged to advertise their own democratic commitments while condemning the actual operation of French or British democracy, in order to demonstrate their own advance over previous understandings today's leftist historians feel compelled to reject Thompson's account of class formation without ceding popular history's terrain to Whig self-congratulation or Tory disdain.

Declining confidence in the capacity of organized parties and militant workers to check the power of states and capital, much less to bring about just and prosperous regimes, encourages the same historians to turn inward for consolation, seeking hope in improved critical understanding rather than expanded capacity for collective action. As faith in revolution fell, faith in deconstruction rose.

Both Patrick Joyce and James Vernon, for example, have sought refuge from Marxist realism in linguistic analysis, Joyce fretful and Vernon with shrill bravado. Each proposes his own interpretation of English popular culture and its creeds as an alternative to the Thompsonian history of class formation. Joyce concentrates on Lancashire and the North between 1848 and 1914, eventually concluding with great unease that something like widely-shared class consciousness began to emerge not in Thompson's 1790s but toward World War I. Vernon's Politics and the People, for its part, takes on all of England from 1815 to 1867, but uses as recurrent points of reference close studies of public politics in Boston, Lewes, South Devon, Tower Hamlets, and Oldham.

Despite avoiding direct confrontation with Thompson's treatment of 1780 to 1832, Joyce and Vernon both seek self-consciously to displace Thompsonian analysis of class formation. They do so by means of three maneuvers: denial that economic experience shapes class consciousness, insistence on the variety of economic and social experience, embedding of all meaningful experience in language. In so doing, each makes two further moves he does not quite recognize, and therefore does not bother to defend. The first is to adopt radical individualism, an assumption that the only significant historical events or causes consist of mental states and their alterations. The second is to doubt the inter subjective verifiability of statements about social life. Together, the two moves take them into the swamp of softcore solipsism.

Hardcore solipsism, a venerable philosophical doctrine, denies the possibility of any knowledge beyond that of the knower's own individual experience. According to hardcore solipsism, all efforts to communicate, persuade, explain, much less accumulate collective knowledge, face insuperable barriers. No consistent believer in hardcore solipsism could pretend to write authoritative historical analyses. Joyce and Vernon opt for softcore solipsism by recognizing(however uncertainly) collective actors, by claiming to know something about what 1 9th century workers thought, by treating language as subject to systematic analysis, and by persisting in the effort - useless according to strict solipsistic doctrine -- to teach others their interpretations of British history. Fixed on the task of refuting Thompson and his ilk, furthermore, they center their analyses on questions of consciousness, on knowing what different groups of ordinary people actually thought at various times in the l9th and 20th centuries. In the process, they abandon agency, cause, and effect except in so far as conscious deliberation causes individual action.

Vernon's and Joyce's occultation of agency separates them from conventional historical narrative, in which limited numbers of well-defined motivated actors, situated in specific places and times, express their ideas and impulses in visible actions which produce discernible consequences, those consequences typically being the objects of explanation. Conventional narrative entails not only claims to reasonably reliable knowledge of actors, motives, ideas, impulses, actions, and consequences but also a) postulation of actors and action as more or less self-contained, b)imputation of cause and effect within the narrative sequence. Softcore solipsism makes most of these elements difficult; denial of agency makes them impossible.

Vernon and Joyce also rule out alternative modes of social-scientific analysis, which require less access to other people's consciousness as well as allowing actors, actions, and environment to interact continuously, but demand even stronger conceptions of causal connection. Either solipsism or the denial of agency suffices to command rejection of these forms of social analysis. In short, the Joyce-Vernon philosophical position obliterates any possibility of historical explanation. It also undermines any grounds they might propose for accepting the validity of their analyses in preference to Thompson's or anyone else's.

As a practical matter, Joyce and Vernon pour much of their effort into twinned enterprises: 1) identifying alternative discourses to those of class; 2) finding new sorts of evidence to illustrate those discourses. Neither enterprise, however, advances any rationale for believing its results. Hardcore or softcore, solipsism lays on its advocates the burden of proof that what they are saying deserves more attention than the chattering of birds.

If historical analysis consists of nothing but language games, of course, one game is as good as another. Thus the objective of academic effort reduces to the provision of multiple perspectives on ultimately indeterminate events. In this view, Vernon and Joyce break sharply with the realist epistemology and ontology of E.P. Thompson. (In the face of this sort of negation, Thompson told me a few years before his death that he had long disapproved of my penchant for social science but now saw that despite my failings we both stood on the same -- realist -- side of a widening, dangerous divide.)

Joyce and Vernon nevertheless remain captives of Thompson to a far greater degree than they acknowledge. First, they focus on the explication of plebeian consciousness in a very Thompsonian manner. Second, they rely on the assembly of numerous texts -- now defined with the great breadth to which Thompson accustomed us -- to substantiate that explication of consciousness. They engage in Thompsonian hermeneutics. Thompson must take the credit or blame for the sheer power of his argument and practical example, as well as for his own tendency to center his rare methodological discussions on the relationship between experience and consciousness. Thompson thereby undermined one of his own most important teachings. For it was Thompson above all who argued that class was not an individual state of mind, not even the collective mentality of a single group, but a dynamic, contested relationship among sets of people.

Anyone who adopts language as the analytic base for the treatment of class should, in fact, immediately recognize the significance of Thompson's teaching. Language is a deeply social medium, heavily dependent on interpersonal negotiation and creation. In solitary confinement, humans never learn to speak. The minimum set for the study of language consists not of a single thinking individual, but of two persons in communication with each other. To the degree that the linguistic turn brought historian toward solipsism, it led entirely in the wrong direction.

Historians do not wander alone through the epistemological and ontological wilderness. Social scientists and historians alike have frequently made the same mistake: interpreting social relations as if they were individual attributes. The program many social scientists call methodological individualism makes a virtue of just such a procedure. In the analysis of work and labor markets, economists have commonly supposed that people's jobs and incomes resulted directly from their individual human capital through the impersonal operation of something mysterious called the market. But the organization of jobs, work, and compensation actually centers on constantly renegotiated relations between workers and employers. Ethnicity and nationality likewise consist not of individual characteristics but of labeled connections among people. The individualization of identity causes great confusion in social analysis.

Identities in general reside in interpersonal relations, which is why the possession of multiple identities -- highly problematic in an individualistic perspective -- poses so little practical difficulty to most human beings. (The only people I have ever met who had more or less unitary identities were either psychiatric patients or fanatics, or both.) To be a daughter is to live in a certain relationship to a parent, to be a slave is to endure a certain relation to a master, to be a citizen is to hold certain rights and obligations vis a vis a specific state, to belong to a working class is to share with other people a certain relation to capitalists. Precisely: when he insisted on class as a relation among groups, E.P. Thompson rejected its reduction to individual characteristics, including individual consciousness. Alas, historians did not hear him well, any more than social scientists in general have understood the centrality of transactions, not individuals, in social life. Language, culture, identity, and class all reside not in single minds but in dynamic, contingent, negotiated relations among human beings.

Both Thompson and his critics contributed to another widespread problem in historical explanation: the assumption that to explain we must invoke an invariant causal law -- if X, then Y -- and show that the case at hand conforms to that law. In the instances we have been discussing, the causal laws concern sufficient conditions for class consciousness. But such thinking pervades history and the social sciences.

The most egregious examples of invariant thinking appear in comparative-historical analyses where nations, states, or societies serve as the objects of comparison. Even methodological individualists, nevertheless, frequently follow the same logic, albeit at a smaller scale. They model the necessary and sufficient conditions under which a rational decision-maker (or, in other versions, the follower of a unitary vision, illusion, or impulse) would take steps to create a state, make war, rebel, secede, vote, join a social movement, or carry on some other well-defined political performance.

Similar reasoning appears frequently in studies of nationalism, democratization, the disintegration of empires, social movements, transformations of states, wars, revolutions, and other large-scale political phenomena. In the case of nationalism, available theories range from primordialist to constructivist, from realist to subjectivist, but a surprising proportion of them claim not to account for the variable degrees or qualities of nationalism but to place most or all nationalisms in the same box. The study of social movements offers more promising recent trends, since a number of scholars have taken to relating variation in the organization of movements systematically to differences and fluctuations in political opportunity structure. Yet even there much theorizing has proceeded as if all social movements fell into just two internally homogeneous categories: Old and New. In analyses of war, democratization, and state formation, similar simplifications prevail. Invariant models concerning self-motivating social units continue to wind like honeysuckle through the study of large-scale political processes.

Looked at closely, the standard practice makes little sense. Coherent, durable, self-propelling social units -- monads -- occupy a great deal of political theory, but none of political reality. Ostensible general conditions such as revolution, nationalism, or war always turn out to fall not at a single point but along a whole range of positions on some intersecting set of continua. The employment of invariant models, furthermore, assumes a political world in which whole structures and sequences repeat themselves time after time in essentially the same form. That would be a convenient world for theorists, but it does not exist.

Three recent inquiries of my own made me aware of the difficulty, which had bothered me for years without my being able to articulate it well. The topics differed greatly: European revolutions from 1492 to the present, American immigration during the l9th and 20th centuries, British popular politics between the 1750s and the 1830s.

The first inquiry was a book on European revolutions from 1492 to 1992. On agreeing to write the book, I had implicitly assumed that it would be easy, almost a potboiler, an exercise in locating the best model of revolution around -- perhaps Skocpol's, Kimmel's, or Goldstone's - - polishing it up a bit for my own purposes, then fitting it to a number of European revolutions: the old game of Improving Karl Marx that self-appointed theorists among us all play so confidently. (These days the thinker Improved is more likely to be Max Weber, Anthony Giddens, Jurgen Habermas or, heaven help us, even Talcott Parsons rather than Karl Marx, but the rules remain the same: Explicate the model; single out one or two elements for criticism; correct those elements; glue the updated model back together; congratulate yourself; publish the result.) But I wanted to connect the analysis of revolution to those of state formation and collective action in which I had been dabbling for some years.

At length I realized that I was yoking a lion and a hippopotamus together for plowing; I was starting mayhem rather than the neat cultivation of a field. Why? Because my favored models of state formation and collective action concerned continuous variation rather than recurrent invariant phenomena. Although I had perpetrated invariant models of both earlier in my career -- we remain creatures of our educations so long! -- through protracted struggles with historical material I had first rejected one-track models, then begun to formulate accounts of variable trajectories by searching for deep causal mechanisms. Meanwhile, the available models of revolution, at least in their most general forms, all purported to specify the necessary and sufficient conditions for revolution, conceived of as a relatively invariant bundle of structures and processes. The misfit soon became obvious.

Attention: I don't claim to have found the Deep Causes of all changes in the character of states or of all variations in collective action. I only claim to have recognized that the regularities lie in the generating mechanisms rather than in the recurrence of whole structures, the repetition of whole sequences, the reappearance of the same unilinear processes. Such a recognition does not preclude typologizing states or collective actions, mapping sequences of conflict, or even tracing long processes of transformation, but it does entail recognizing that those operations do not yield explanations. They simply specify what is to be explained.

In the case of revolution, then, I found that I had to rethink the phenomenon as one zone in a much larger field of variation including many political interactions no one would label revolutionary, then search for clues as to why some peoples, places, and eras spent a lot of their time in that zone while others barely approached it. My first crude device for doing so consisted of analytically separating the conditions for revolutionary situations from those for revolutionary outcomes -- revolutionary situations consisting of open splits within polities, revolutionary outcomes consisting of substantial transfers of power over states. I argue that the two sets of conditions vary and change in partial independence of each other.

My second crude device was to treat each of those conditions as a continuum, for example from no split whatsoever in a polity to a split putting every political actor on one side or another. My third was to treat major changes in the organization of states, state systems, and armed force as determinants of the positions of different states and polities on those continua. -My answers surely contain defects, but they illustrate another way of thinking about revolution than as a one-track phenomenon. They represent a historicizing of the problem.

Any vendor of standard models of revolution, for example, will have trouble selling them to specialists in early modern European history who go beyond fitting their appliances to the English revolution of 1640-1660, the Fronde, and the revolt of Catalonia to asking why so many more forcible attempts to seize state power -- revolutionary situations, roughly speaking -- occurred, and why so many of them actually succeeded. Close study of the circumstances of those centuries' revolutionary situations does not produce a new General Model of Early Modern Revolutions. It does, however, reveal the grounding of revolutionary situations in prevailing conditions of indirect rule, military expansion, and dynastic competition.

Take the factor of dynastic competition: In Muscovy, then in the Russian empire that grew with Muscovy as its kernel, for three centuries after 1492 every time a tsar died without an adult, militarily competent son or brother to succeed him a serious struggle for the throne ensued, often with wide popular support for one faction or another. During the 17th and 18th centuries the serious claimants at different times even included thirty or forty men pretending to be tsars or heirs whom everyone else had believed dead -- often murdered at the behest of the late ruler. Cossack Emelian Pugachev, who led the great peasant-Cossack rebellion of 1773-1775, claimed to be the deposed dead tsar Peter III. A number of the claimants actually made it to the throne: Boris Gudonov, his successor the false Dmitry, Ivan V, his brother Peter the Great himself all became tsar irregularly, outside the standard inheritance rules, through the use of force. Russia was no more extreme in this regard than Poland, Hungary, and a good many other early modern states. Yet by the l9th century militarily contested successions had become rare in European monarchies.

The whole story of that transition would take too long to tell, and would require too many allowances for variations among, say, Iberia, the Balkans, and the British Isles. But one cluster of factors nicely illustrates my general point: the tight interdependence in early modern European states among the organization of great families, the existence of huge patron-client chains attaching officials, servitors, and tenants to those great families, the embedding of military force in those patron-client chains, and the adoption by great families of out marriage strategies accomplishing three purposes: first, giving their heirs claims on aristocratic and royal successions elsewhere, providing local members of the family (including emperors or kings) with some call on military assistance from grandees or rulers outside their own countries, and arranging another place to survive comfortably if life became too dangerous at home.

Together, these circumstances meant that almost every royal succession constituted an opportunity, or at least a hope, for some rival to the most obvious heir, often a foreigner in whom another royal family also had an interest; where the inheritance was unclear or the heir incompetent, the opportunity became a strong incentive to employ autonomous military force, and enlist aggrieved popular support, for a dynastic coup. When Protestant lords invited fellow-Protestant William of Orange to England in 1688 to displace catholic king James II, they did not simply call on an experienced statesman from a distinguished family; they called on a grandson of Charles I and son-in-law of James himself. ,An explanation of the Glorious Revolution requires much more than knowledge of William's family background. Nevertheless, no one will understand it and other revolutions of the time without exploring the mechanisms by which great families attached themselves to each other and to regimes. Such an exploration is deeply historicist. In it, structure and culture interact.

Let me underline what this means, and what it does not mean. Considered as wholes, neither lineages nor revolutions had recurrent structures besides those they shared by definition. Invariant models of lineages and revolutions would serve us badly. On the contrary, the regularities lie in the ways that kinship ties affected the formation of alliances, the probability of war, and the claims to succession to supreme positions in dynastic states, which in turn affected the probability and character of revolution. These are not invariant structures or processes, but wide-ranging causal mechanisms whose combinations produced the actual unique histories we observe.

The second inquiry concerns inequality and, American immigration. As Ewa Morawska (1990) has well documented, recent work on immigration has challenged the two dominant models of earlier generations: human capital and assimilation. Human capital models escape my strictures somewhat by deliberately accounting for differential success as a function of variable resources, broadly defined; they deserve suspicion, nevertheless, for their reliance on a invariant model of market-mediated success. Assimilation models clearly qualify as invariant in so far as they posit only one standard path into American life, the chief variation being the speed at which different groups travel that path. As Morawska says, an anti-invariant historicist view helps make sense of the connections between migration and durable forms of inequality, including those forms people organize as ethnicity -- as structured differences according to imputed national or racial origin. In thinking about American immigration as a whole, and about my current collaborative studies of nineteenth-century French silk workers in Paterson, New Jersey, and of twentieth century Italian peasants in Mamaroneck, New York, I find it useful to ask how the social organization of migration constrained the subsequent opportunities of different groups of migrants and their descendants.

In the case of Paterson, I am trying to find out how textile workers from Lyon and its vicinity entered the expanding silk industry of Paterson after 1860, as well as what impact those circumstances had on their experience, and that of their children, in the American labor force. As for Mamaroneck, I am attempting to compare post-1900 migrations from a few villages in the Frosinone, near Rome, to the Lyon metropolitan area, to Mamaroneck and nearby towns, andperhaps eventually to Buenos Aires, São Paulo, and Toronto as well, in order to see how differently the survivors of those migrations turned out at their various destinations.

How well immigrants do in a new country, and whether they return to the old, depends mainly on five factors: the extent to which they integrate on arrival into networks that embrace a wide range of employment opportunities; the opportunities for individual income with which their networks give them contact, especially at the start; the opportunities for collective capital accumulation at the destination; the degree of obligation to support persons and enterprises in the place of origin; and the relative opportunities for reinvestment of accumulated capital at the original and the destination. On the whole, the more the migrant group or its subdivisions serves as an accumulator of capital, the greater the incentive to pass jobs through kin and paesani. Thus durable inequality among immigrant groups and their descendants depends on the initial organization of migration and its capacity for accumulation of collectively-available capital. While it sounds strange to put warm matters so coldly, immigrants and their descendants actually know these principles well; their stories of connections, favors, and ethnic differences reek of them.

Networks transformed or created by migration create and maintain inequality. Members of immigrant groups often exploited each other as they would not have dared to exploit the native-born. Every act of inclusion, furthermore, also excludes. North American immigration produced a remarkable specialization of work by origin, although the precise specializations varied from one locality and migrant stream to another. The characteristic story of Mamaroneck is the present domination of landscape gardening and related fields by Italian immigrants and their heirs, that of Paterson French, British, German, and Italian workers from well-defined industrial locations entering specific branches of Paterson's industry.

Any student of migration can tell similar tales of occupational specialization by regional or national origin. The actual tales refute grand stage schemes of immigration, illustrate the combination of bounded contingency with constraint in social life, and show us powerful causes working consistently as links among events. Generalized, that observation makes my case against monadic ontologies and invariant models, for historicism concentrating on the discovery of mechanisms that generate social structures, sequences, and processes. Again, culture and structure interact.

A third area of research that made me think about these topics concerns changes in the forms of collective contention -- for example, why and how sit-ins and similar deliberate occupations of contested spaces rise and fall. For the shared delusions of collective-behavior theorists, sociologists of the 1960s and 1970s generally substituted models of collective rational action: public choice, resource mobilization, political process, and so on. In so doing, however, they(perhaps I should say "we") stuck unwittingly to monadism, assuming that the main problems were a) to explain the behavior of one coherent actor (individual or collective) at a time, and b)to identify a single model of collective action that, with no more than nudges of a parameter or two, accounted in principle for all instances. In the study of social movements, for example, this reformulation rejected earlier portrayals of prohibitionism or feminism as irrational reactions to the stress of social change, but retained the assumption that the social movement was a kind of self-contained group whose behavior could be explained by the group's social situation. Similar, Mancur Olson's injection of collective-goods models into the analysis of what sociologists previously called collective behavior (Olson 1965) sent historians and sociologists scurrying for alternative invariant models that would accommodate identity, loyalty, and self-satisfaction (see Cohen 1985, Gamson 1990).

Let me spare you a detailed critique of standard models for social movements and collective action. Suffice it to say that monadic analyses of contention ignored the strategic interaction among challengers, competitors, and sometime allies that pervades real episodes of contention.(As participants and benevolent observers of social movements, many formulators of monadic models had ample practical awareness of strategic interaction, but failed to draw the appropriate theoretical conclusions from their own experiences.) A combination of influences tipped the balance toward interaction: the infiltration of game-theoretic reasoning from economics and political science; the creation of large catalogs of events as alternatives to the treatment of one group, movement, or action at a time; above all, the historicization of polemology (as francophones call the systematic study of conflict).

In this setting, historicization meant installing time and place as major determinants of contention's character rather than as proxies for other more elusive variables such as modernization or level of grievance. To historicize the study of contention meant recognizing that collective claim-making entails the simultaneous use and recasting of relations, including shared understandings, among local actors. It meant seeing that each locality and each interacting set of claimants, both challengers and authorities, accumulates its own particular experience, memory, understanding, and practices, and accumulation that strongly constrains current contention. My own formulation of these insights adopts the theatrical language of repertoires;contentious actors perform in dramas in which they already know their approximate parts, during which they nevertheless improvise constantly, and of which the exact outcomes remain uncertain.

In this formulation, potential actors choose strategically among available performances, engage other actors, including objects of their claims, in those performances, and improvise their way to some conclusion. The conception is at once deeply interactive -- that is, structural -- and deeply cultural. It reeks of culture, as Arthur Stinchcombe has pointed out, in insisting that shared understanding and their objectifications constrain social interaction (Stinchcombe 1987).Historical Methods: 11My current research on the subject uses catalogs of British "contentious gatherings" between 1758and 1834 to examine how claim-making changed during a period that brought Great Britain the demise of Rough Music, collective machine-breaking, invasions of enclosed fields, and many related forms of interaction, as well as the rise of public meetings, demonstrations, petition drives, popular associations, firm-by-firm strikes and more now-familiar forms of struggle (Tilly 1995a,1995b).

A contentious gathering, for the purposes of this study, is an occasion on which ten or more people gathered in a publicly accessible place and visibly made claims which, if realized, would bear on the interests of at least one person outside their number. The main machine-readable catalog provides detailed descriptions of 8,088 contentious gatherings that occurred in Southeastern England during thirteen scattered years from 1758 to 1820 or anywhere in Great Britain during the seven years from 1828 through 1834. Among other things, my group is analyzing the events in that catalog and complementary evidence to determine whether a strong version of the repertoire model actually holds up to close scrutiny. We are unquestionably seeing profound changes in the texture of British contention, as seizures of grain, invasions of fields, mocking ceremonies, and related forms give way to processions, demonstrations, petition drives, and their kin. The changes pivot on the years of war with revolutionary and Napoleonic France, bearing plausible relationships to the transformations of the state and economy during the war years. That much verifies at least a weak version of the metaphor.

For stronger versions, we must look at innovation and variation within and among contentious gatherings. My analyses yield evidence, for example, of parliament's increased salience as an object of contentious claims and of the role played by public meetings, local assemblies, and popular associations in that shift. One can trace the influence of innovators such as John Wilkes, Lord George Gordon, Francis Place, and Daniel O'Connell on cumulative shifts in contentious repertoires. We have some grounds for claiming that collective actors constantly innovate is small ways, and do so at a faster pace when political opportunities are changing rapidly, but that innovations in the forms of contention only stick when associated with visible success for one actor or another. But many questions remain open.

I won't bore you with other results, technical details, and historical problems. I am trying hereto illustrate how historical thinking, properly conducted, combats monadism and helps reveal the tight interdependence of culture and social structure. For in the analysis of British contentious repertoires, as in the study of revolutions and of immigrant itineraries, we find the cumulative intersection of history, social ties, and shared understandings.

What, then, are these elusive causal mechanisms I have identified as the true locus of regularities in social life? In the case of revolutions, they consist of rapid and visible diminutions of state power, splits in control over the major means of coercion, formation of anti-regime coalitions, and other political shifts that singly neither guarantee revolution nor constitute parts of its definition. In the case of immigration, the crucial causal mechanisms consist of the transmission of information about opportunities within existing ties of kinship or neighborhood, the pooling of capital or credit, the hoarding of access to remunerative work, housing, and social life, the Historical Methods: 12remittance of money and other resources to the place of origin, and other collective actions that shape the structure of opportunities, rights, and obligations; all of these operate outside of immigration, indeed quite outside of residential mobility of any kind.

In the case of changes in contentious repertoires, we must look for causes in the transformation of political opportunities by innovations associated with successful claim-making, in alterations-- incremental or sudden -- of various political institutions' capacity to deliver rewards or punishments, in the creation or rupture of links among potential collective actors, and in similar mutations of shared incentives and organizational resources. If revolutions, immigration, and changing repertoires defy invariant models, that is not because they know no regularities. It is because their regularities do not lie in recurrent structures or sequences but in powerful causal mechanisms that in different combinations produce both those phenomena and a host of others, that in principle could produce events no one has ever seen. At the levels of Metamethod, Methodology, and Methodical Maneuvers, both softcore solipsism and invariant models block our way to the supple, counterfactual search for reliable historical explanations.

Note: I have adapted some material in this paper from my "Softcore Solipsism," Labour/Le Travail 34 (1994), 259-268, "History and Sociological Imagining," Tocqueville Review 15 (1994), 57-74, and various prefaces to the review-essays section of Sociological Forum, 1985-1992.

References

Cohen, Jean (1985): "Strategy or Identity: New Theoretical Paradigms and Contemporary Social Movements," Social Research 52: 663-716.

Gamson, William A. (1990): The Strategy of Social Protest. Belmont, California: Wadsworth.2d edn.

Goldstone, Jack A. (1991): Revolution and Rebellion in the Early Modern World. Berkeley: University of California Press.

Joyce, Patrick (1991): Visions of the People: Industrial England and the Question of Class. 1848-1914. Cambridge: Cambridge University Press.

Kimmel, Michael S. (1990): Revolution. A Sociological Interpretation. Philadelphia: Temple University Press.

Moore, Barrington, Jr. (1979): Injustice. The Social Bases of Obedience and Revolt. White Plains, N.Y..: M.E. Sharpe.

Morawska, Ewa (1990): "The Sociology and Historiography of Immigration" in Virginia Yans-McLaughlin, ed., Immigration Reconsidered. History, Sociolov Politics. New York: Oxford University Press.

Olson, Mancur (1965): The Logic of Collective Action. Cambridge: Harvard University Press.

Skocpol, Theda (1979): States and Social Revolutions. A Comparative Analysis of France. Russia. and China. Cambridge: Cambridge University Press.

Stinchcombe, Arthur L. (1987): review of Charles Tilly, The Contentious French, American Journal of Sociology 92: 1248-1249.

Thompson, E.P. (1963): The Making of the English Working Class. London: Gollancz.

Tilly, Charles (1993): European Revolutions. 1492-1992. Oxford: Blackwell. (1995a): Popular Contention in Great Britain. 1758-1834. Cambridge: Harvard University Press. (1995b):"Contentious Repertoires in Great Britain, 1758-1834," in Mark Traugott, ed., Repertoires and Cycles of Collective Action. Durham: Duke University Press.

Vernon, James (1993): Politics and the People: A Study in English Political Culture c. 1815-1867. Cambridge: Cambridge University Press.