quantitative studies of the french revolution

Upload: amandaalisha

Post on 10-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/8/2019 Quantitative Studies of the French Revolution

    1/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION

    GILBERT SHAPIRO, JOHN MARKOFF, and SASHA R. WEITMAN

    I. INTRODUCTION: ORIENTATION AND RATIONALERevolutions have fascinated moralists, political theorists, and historians forcenturies. Since 1789, the prevalence of revolution and, perhaps more impor-tant, the threat of revolution have been pervasive forces in contemporarysociety. Social scientists and other students of the human condition havelaunched inquiries with an extremely wide diversity of objectives and methods;and we feel it necessary, in order to delineate our own approach, to dis-tinguish among the variety of activities commonly collected under the titlesof "comparative revolution" or the "sociology of revolution."'

    Some writers have attempted to discern, beneath the violence and chaos ofrevolutions and the diversity of revolutionary experiences, a more or lessuniform pattern of development, advancing a "natural history" of revolution.Such studies concentrate on the course of events in revolutions once they havebegun, rather than, for example, on the patterns of recruitment of leaders andfollowers into revolutionary movements, or the conditions of social structurebefore revolutions break out, or the consequences of revolutionary episodes forthe history of society. Such work is invariably and necessarily comparative.In the "natural history" approach, revolution is regarded as a species, ofwhich the French, the Russian, and the American might be selected specimens.The object about which generalizations are desired is the class of all revolu-tions, whose common taxonomically defining elements are sought.2Other theories and comparative effortsgo beyond such attempts to generalizeover the class of all revolutions, in order to seek common elements in suchbroader categories as "internal war" or "conflicts within nations."3 While we

    1. The most adequate surveys of this field of study and its problems are ChalmersJohnson, Revolutionary Change (Boston, 1966), Revolution and the Social System(Stanford, 1964), Lawrence Stone, "Theoriesof Revolution," World Politics 18 (1966),159-176, and Isaac Kramnick, "Reflections on Revolution: Definition and Explanationin Recent Scholarship,"History and Theory 11 (1972), 26-63.2. Representative works include: Crane Brinton, The Anatomy of Revolution (NewYork, 1938), Lyford Edwards, The Natural History of Revolution (New York, 1965),Pitirim Sorokin, Social and Cultural Dynamics, Vol. III (New York 1937-41), HenriSee, Evolution et revolutions (Paris, 1929), and Rex D. Hopper, 'The RevolutionaryProcess,"Social Forces 28 (1950), 270-279.3. Internal War: Problems and Approaches, ed. Harry Eckstein (Princeton, 1964),

  • 8/8/2019 Quantitative Studies of the French Revolution

    2/29

    164 SHAPIRO, MARKOFF, AND WEITMANwish unqualified success to all who seek to elucidate the human condition, weare hardly sanguine about the possibilities of discovering interesting, important,and reliable uniformities even over the class of all "revolutions," let alone of"internal wars." Perhaps a sub-category, such as "modernizing revolutions,"has some chance of defining a category for which reliable and theoreticallysignificant generalizations can be discovered, but our real interest here is notin attacking the strategies of others, but in clarifying our own. While thecomparative study of revolutions in an effort to generalize about them hasdominated the "sociology of revolution," and we too are sociologists studyingrevolutions, we are not engaged in that quest (however worthwhile), and wedo not evaluate the relevance of hypotheses, in our own work on the FrenchRevolution, on the basis of their applicability to, say, the Russian, American,English, or Mexican revolutions.

    Still other scholars have been intrigued by the extraordinarymen who cometo the fore as leaders, and the ordinary men who participate as followerseither in revolutionary movements or in riots or other transient events. Theyhave, consequently, studied the processes of recruitment into various politicalgroups, associations, and crowds, and their social composition.4 Others areinterested in the political in-fighting that emerges when a society's usualrestraintson the means employed in political conflicts do not apply. These lastproduce studies of military coups, foreign involvement, and party strategywith a view to determining which actor (or actors) wins in a revolutionarysituation. Such studies may be of importance in themselves, and some areeven designed, as are ours, to throw light on the social forces operative inrevolutionary situations. But their research strategies, if not their objectives,are sharply differentfrom the one which we are suggesting.

    and "On the Etiology of Internal Wars,"History and Theory 4 (1965), 133-163. Ray-mond Tanter, "Dimensions of Conflict Behavior within and between Nations, 1958-1960," Journal of Conflict Resolution 10 (1966), 41-64, and "Dimensions of ConflictBehavior within Nations, 1955-1960: Turmoil and Internal War" in Papers, PeaceResearch Society, Vol. III (1965), 159-183. R. J. Rummel, "Dimensions of ConflictBehavior within and between Nations," General Systems Yearbook, Vol. VIII (1963),1-50, and "Dimensions of Conflict Behavior within Nations, 1946-1959," Journal ofConflictResolution 10 (1966), 65-73. Gratitude is expressedto John Max, for stimulat-ing ideas on the theory of revolution, and a careful critical reading of this paper.4. While each of the following works goes considerably beyond studies of recruit-ment, the analysis of the social composition of crowds and groups is neverthelesscentral to their method. George Rude, The Crowd in the French Revolution (Oxford,1959); Albert Soboul, Les Sans-Culottes parisiensen l'an II (Paris, n.d.); Crane Brinton,The Jacobins:An Essay in the New History (New York, 1930). Of course, many otherscould be cited.5. Here we would include the vast literature on revolutionary strategy and tactics,such as the well-known work of Lenin and Che Guevera, as well as such scholarlystudies as Robert R. Palmer, Twelve Who Ruled: The Year of the Terror in theFrench Revolution (Princeton, 1941) and large parts of the work of Soboul, cited above.

  • 8/8/2019 Quantitative Studies of the French Revolution

    3/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 165Our efforts are not directly concerned with what happens once a revolution

    has begun, but rather in the definition of the features of a "revolutionary situa-tion." At least since the work of Marx and Tocqueville, this has been a centralfocus for many of the efforts to develop "theories of revolution." Althoughthere are more or less close ties among these various subject matters, this questshould not be confused with the search for a natural history of revolutionaryprocesses, studies of the recruitment of revolutionary participants, or thestrategy and tactics of revolutionary action. While the events of August 4,August 10, or the 9th of Thermidor are of the greatest significance for diag-nosing the quality of the Revolution, we - like Tocqueville - in our questfor its sources in the social structure must invest more of our time and effortin the analysis of the social structure of the Old Regime.In contrast to the perspectives described above, the basic subject matter ofour studies is not revolution, but society - a case study of a trajectory in thelife of a particular society. Because the issue of statistics vs. case studies is aperennial methodological debate in sociology (and because historians arebeginning to be infected by the disease), we feel it necessary to indicate somemethodological grounds for an interest in a single case.

    The French Revolution of 1789 was a unique event; but understanding it,we contend, is of importance if one assumes that there are principles of theorganization, the integration, and the operation of societies which may manifestthemselves more clearly in some historical circumstances than in others. Inrevolutions, we suggest, one finds writ large the processes of social change,which are so much more difficult to discern in slow and gradual historicaldrifts. Revolution is of interest also because we expect to find the conditionsof stability, that is, of the integration and operation of social structures,moreapparent in extremis. It is a point too often neglected that one cannot under-stand stability without understanding change. One can no more study thedeterminants of persistence without studying its opposite than test hypothesesabout heat without examining both cold and hot objects.6 Hence, insofar as"functional"sociology is concerned with the study of persistence and integra-tion, the study of revolutions must be at the center of its attention; it is notan odd, if interesting, sideshow.

    Thus through the intensive examination of a single case one hopes to ob-serve manifestations of the operation of general processes. There is, of course,

    6. Interestingly, comparative studies of revolution commit a similar, if, in a sense,an opposite error: they ignore the study of stability. In seeking the uniformities inrevolutionary situations, they rarely examine nonrevolutionary situations. Hence,Brinton confesses that between successive editions of The Anatomy of Revolution, the"alienation of the intellectuals"could not properlybe regardedas a clear-cutdistinguish-ing sign of a society on the verge of revolution, as he had argued in the earlier version,because it was a general characteristic of all modern societies. (See preface, revisededition [New York, 1965], vi.)

  • 8/8/2019 Quantitative Studies of the French Revolution

    4/29

    166 SHAPIRO, MARKOFF, AND WEITMANalways a question of the limits of the generalizability of results, a problemwhich leads some social scientists to reject the case study altogether. What isoften not realized is that there is always some uncertainty as to precisely whatuniverse one may generalize to; in a profound sense, all studies are casestudies. It is not the principles of statistics that tell us whether the results of astudy of a probability sample of Americans in 1972 can be reasonably expectedto hold true in 1973.7 It is the task of the case researcher to frame his ques-tions, and his explanatory concepts, in such a way as to permit similarquestions to be asked of other cases. In this way, case studies are, ideally,cumulative; and a proper answer to the sociologist who objects that we arestudying "only one" revolutionary situation is that we are not the onlyresearchers in the world. The usual comparative study of revolutions, byconcentrating on those features in which revolutions resemble one another,has neglected the distinctive features of particularrevolutions. Yet, we believe,it is partly through studying the distinctive features of the French Revolutionin the context of French society that one can hope to discover sociologicaluniformities of a different kind: the principles of social organization, change,and development.

    We have referred to such a project as a case study. But this is so only inthe sense that all particular researches are case studies. Consider a manstudying a social structure, such as the contemporary United States, NewYork, or a labor union, conceived statically at one point in time. He may,with the union, for example, regard his sample size as, say, two thousand orone, depending on whether, at the moment, he is generalizing to the entiremembership of that union or to the population of unions in general. Identically,we might be regarded as studying 100,000 grievances, 1200 documents, a fewhundred electoral districts, or one society in revolution.

    The quantitative methodology employed is, at bottom, the search for andthe interpretationof systematic covariations; our study, too, is comparative inthat we compare Bretagne and Languedoc. No doubt an audience of historianswould not take us to task for studying only one country. Indeed, somehistorians of eighteenth-century France, realizing the degree to which thatcountry varied from place to place, have wondered at studying as much as onecountry. Yet it is precisely such variation that permits the employment ofquantitative techniques of hypothesis testing. One can examine covariationonly where there is something that varies.8 The incredible diversity of eight-

    7. For a detailed discussionsee Lee J. Cronboch,Goldine C. Gleser, HarinderNanda,and Nageswari Rajaratnam, The Dependability of Behavioral Measurements: Theoryof Generalizability for Scores and Profiles (New York, 1972), 358-383.8. The incest taboo may well have provoked the largest number of explanations ofany social phenomenon and unquestionablythe largest number of hare-brainedtheories.What prevents decisive testing of these explanations and rejectionsof most, if not all- is precisely its universality. Since all known societies have it, one cannot study the

  • 8/8/2019 Quantitative Studies of the French Revolution

    5/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 167eenth-century France, a society characterized by one nineteenth-centuryhistorian as being unequal in its very inequalities, means that we will, in fact,compare regions that experienced revolutionary excitement, regions producingor supporting counter-revolution, and regions where little happened.Among revolutions, the great French Revolution of 1789 appears to be themost strategic choice for such a study, not because of any unique historicalrole that it plays in world history, but because of the unique richness ofquantitative and quantifiable materials available. We have sought - and arecontinuing to seek - to bring together a mine of data with which we mayexamine and test a wide variety of hypotheses. We are exploring the use of aninstrument for the study of social change in eighteenth-centuryFrance, and sothe theoretical propositions that have been tested, and that may be tested inthe future, range quite widely.

    Some of the finest minds of the world of scholarship have devoted them-selves to the study of the French Revolution. From their works one can cull avast number of interrelated, sometimes contradictory, and sometimes comple-inentary views, a surprising proportion of which may be tested with availablequantitative data. Indeed, we may put this more strongly. The multiplicity ofwell-argued explanations of the French Revolution cries out for some newkind of sifting of evidence. And of course we have our own ideas which wewish to explore. It is important, then, that the materials we have been as-sembling be subject to multiple uses, as we shall illustrate in describing ourempirical investigations to date.

    II. THE CAHIERS DE DOLJANCESAlthough hardly the only data gathered, the most important single resourcewe have generated is a content analysis of the celebrated cahiers de doliencesof 1789. These remarkable documents, as is well known, were produced inthe course of the convocation of the Estates General. The election of 1789allowed a very wide suffrage and saw the representation, to one degree oranother, of virtually the entire kingdom and most of the significant socialgroupings of the old regime. Properly to appreciate the potentialities of thesedocuments, one must know something about the central features of theelection.9conditions under which it does not occur. What can be tested are theories about thoseaspects of incest that do vary, such as which relatives are forbidden to marry. See, e.g.,G. P. Murdock, Social Strictcure(New York, 1949), Chapter 10.9. For a more detailed general descriptionthan given below see Beatrice F. Hyslop,A Guide to the General Cahiers of 1789 (New York, 1933). This seems a good placeto express our gratitude for the very generous assistance of Hyslop during the periodwhen this project was being organized.

  • 8/8/2019 Quantitative Studies of the French Revolution

    6/29

  • 8/8/2019 Quantitative Studies of the French Revolution

    7/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 169and socially, is remarkable enough. What is more remarkable still about thecahiers, and gives them their special interest to students of social change, isthat the French Revolution is the only major revolution at the beginning ofwhich the entire nation, in effect, gathered in public assemblies and recordedits grievances, aspirations, and demands for change. Since nothing wouldappear as patently significant in the study of a revolution as the range, in-tensity, and distribution of grievances among groups in the population, thecahiers de doleances are absolutely unique in importance as a documentarysource.

    Moreover, the cahiers have particular virtues (if some widely-discussedvices) as a tool for the study of public opinion. Many historians have criticizedthem for a wide variety of alleged shortcomings as public opinion data. Forexample, certain individuals are supposed to have been too influential, andothers (like the Duke of Orleans) deliberately circulated "model cahiers"which were too closely imitated by some of the actual cahiers. Moreover, thecollective nature of the assemblies makes it difficult to know, it has too oftenbeen argued, just whose opinions are expressed in these documents. We feelthat much of this criticism is misplaced.-' The modem opinion poll gives one astudy of the opinions held by a sample of unrelated individuals. The relationof this collation of individual opinions to any form of political action isLa France d'apres les cahiers de 1789 (Paris, 1897), 21, suggests that there weremore than fifty thousand.Beatrice Hyslop, a more recent authority, guesses "more thantwenty-fivethousand,"A Guide to the General Cahiers, ix-x. Albert Soboul offers sixtythousand in Precis d'histoire de la revolution franfaise (Paris, 1962), 103-104. Theestimate of the total number of cahiers actually written is an extremely hazardoustask.In the first place, there were roughly forty thousand rural communes; but in many in-stances the "parish" n the convocation sense encompassed a number of such districts.The number of preliminary cahiers of clergy, while undoubtedly enormous, is totallyunknown, for these documents have rarely been studied, reprinted,or even catalogued.The other cahiers are too few in comparison with these to drastically affect the total.We are engaged in further studies of the number of parishes in each bailliage whichdid probably meet and draw up a document.11. Let us consider, for example, the charge that the cahiers reveal little about theviews of assemblies that drafted them because they contain material copied from othercahiers, or from propagandadesigned specifically to influence them. But a comparisonof the bona fide cahierswith such electoral propagandamaterialsreveals that choice wasexercised in selecting among available models; that frequently only a few articles werecopied; that models were rarely if ever copied in toto; that even when totally copied,new demands were usually added. (See, for example, the revealing analysis by PaulBois, Cahiers de doleances du tiers etat de la senechaussee de Chateau-du-Loirpour lesEtats generauxde 1789 [Gap, 1960], ChapterIV.) In short, everything suggests deliberateselection. And why not select a more articulate, expressive, and forceful statement ofone's own genuinely held demands? Numerous other charges and objections have beenraised against the cahiers, discussion of which must be reserved for another place.More detail may be found in John Markoff, "Who Wants Bureaucracy? French PublicOpinion in 1789" (unpublished Ph.D. dissertation, Johns Hopkins, 1972) and SashaWeitman, "Bureaucracy,Democracy, and the French Revolution" (unpublished Ph.D.dissertation,WashingtonUniversity, Saint Louis, 1968).

  • 8/8/2019 Quantitative Studies of the French Revolution

    8/29

    170 SHAPIRO, MARKOFF, AND WEITMANentirely problematic. For the purposes of historical sociology, opinion pollsare altogethertoo democratic, since in the realm of political action the opinionsof some are far more influential than those of others. At the same time, pollsare too individualistic, since significanthistorical action derives less often fromisolated men than from collective interactions. And if this is true of con-temporary democracies, it is bound to be truer still of the corporate-aristocraticregime of pre-revolutionary France. The public opinion that historical andpolitical sociologists ought to be studying is not just a set of attitudes and notjust a set of opinions; what they ought to be studying is the expression ofopinion qua political action. It is opinion that has been articulated throughavailable institutional channels - such as, in modern society, mass media,party platforms, petitions, and demonstrations. Given the choice between thecahiers and a Gallup poll, for the study of public opinion on the eve of therevolution, we would opt for the cahiers. The endorsement, if not the drafting,of a cahier was a collective political act, an integral part of the political process.That those who were more articulate, more persistent, more concerned, moreinfluentialwere more likely to get their views into the texts is not a weak pointof the cahiers as a research tool; on the contrary, we regard it as a very greatstrength, for our interest is in relating the contents of these documents toantecedent conditions, subsequent events, and processes of historical change.The collective nature of these documents also permits an avenue of escapefrom the explanations of social phenomena strictly in terms of aggregate psy-chology to which studies using modem opinion polls are almost necessarilywedded. If we want to understand the effects of differing levels of socialmobility, as sociologists it is more important to know the political attitudes ofregions of high and low mobility (or of groups within regions) than ofindividual persons of high and low mobility. Yet the data on mobility's effectson attitudes in contemporary sociology are generally of this latter kind. Evenif, say, mobile individuals become conservative, the net historical effect ofmobility still may not reduce the threat of revolution, because the non-mobilemay be resentful. (See the research by Shapiro and Dawson, discussed below.)The almost total neglect of this distinction in the literature on mobility as afactor in the French Revolution (aside from the work of Tocqueville), as wellas the more general lack in sociology of data on collective attitudes - despitethe central role of such elements in, for example, the theories of Durkheim -suggests how important the study of such data as the cahiers may be. It is forsuch reasons that we contend that the cahiers would be of interest to sociolo-gists even if there had been no revolution.

    The cahiers de doleances, then, are the results of a reasonably uniformprocedure12 whereby Frenchmen spoke of what bothered them in 1789 -

    12. There are countless minor and countable major variations from the standard

  • 8/8/2019 Quantitative Studies of the French Revolution

    9/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 171and their complaints range from the intricate nuances of the system of indirecttaxation all the way to frustration produced by the loud church bells in thenext parish. They represent a set of collective representations that sociologyought not ignore. Moreover, they have been, surprisingly,largely unexploited.To be sure, historians have invested a great amount of energy in publishingthese fascinating materials, in reading them for insights into the Old Regime,in quoting them as illustrative material, and in closely analyzing them inregional and local studies. But the questions that appear most exciting to ushave hardly been asked before. How did the complaints of the French peoplein 1789 vary across the map of France? What aspects of the social, economic,and political structure of eighteenth-century France are associated with dif-ferent sorts of demands? With what kinds of events during the revolution aredifferent kinds of demands correlated? An appropriate research technique forsuch questions is, first, to subject the cahiers to content analysis, and then toperform multivariate statistical analyses. Only such methods, fully utilizingavailable electronic data-processing technology, can allow for the detailedexamination of such a huge mass of material for the entire nation. While anumber of imaginative and skillful efforts have recently been published byothers, the full potential of these data for social science has hardly beenapproached.13procedure, which pose interestingproblems of comparabilityand headaches for the manwho designs a sample. The parish of Garrebourg in the bailliage of Sarrebourg etPhalsbourg, for example, drew up two distinct cahiers. When we find some suchdeviation from the norm in a parish, the explanation is usually unknown to us, whereasthe variant procedures used at higher levels in the electoral process may have researchedhistories. For example, Third Estate of the city of Metz petitioned for directrepresentation at the Estates General, being unsatisfied, as were a number of othercities, with representation through the bailliage to which it had been assigned. Thepetition was granted;but as it was now on the verge of the opening of the Estates, thecity was convoked by neighborhoods,rather than by corporations,to save time. A towncahier to be sent to Versailles was then drawn up. Some corporations, however, pro-tested. They claimed they had been insufficientlyrepresented by this aberrantprocedure.The Estates General considered the issue and on July 11, 1789 declared the electionnull and void and ordered new elections. These new elections, which did not take placeuntil the fall, far later than the overwhelming majority of elections, produced a secondcashier. (See the article by Lesprand, "Election du depute direct et cahier du tiersetat de la ville de Metz en 1789," Anniuaire de la societet d'histoire et d'archeologielorraine [1903].) There is no end to such oddities.13. A number of recent studies by Paul Bois, Charles Tilly, and Philip Dawsondemonstrate theoretically meaningful covariation of cahier contents and social indicatorsin a local or professionally circumscribed context. Complementingthe studies describedhere, national-level content analyses are under way at the University of North Carolina,by George V. Taylor, and at the Centre de Recherches Historiques of the Sixth Sectionof the Ecole Pratique des Hautes Etudes in Paris, under the direction of Frangois Furet.Plans have been made to incorporate, ultimately, the last-mentioned work (which isproceeding by methods quite consistent with our own) and our codes into a single datafile. We are grateful to Tilly, Dawson, Taylor, and Furet for their many forms ofassistance in this work.

  • 8/8/2019 Quantitative Studies of the French Revolution

    10/29

    172 SHAPIRO, MARKOFF, AND WEITMANIII. THE CONTENT ANALYSIS

    The procedure of content analysis we have used, to which we have giventhe name "Concrete Analytic Coding," attempts to break new ground in anumber of ways. We have, first of all, rejected the idea of entering the fulltext of the document into the computer, in order to search for particularword-combinations listed in a "dictionary." Instead, we have continued to usehuman coders in the electronic era as intermediaries between the text and themachine. In previous applications of code-aided content analysis, the categoriesused for the coding are also the categories used for the substantive analysis.This, applied to the cahiers, would have meant that coders would haveutilized categories such as "Egalitarianism,""Radicalism," "Pro-Bureaucracy,"as these abstractions reflect our current research interests. The unfortunatecoders would then have had to struggle, not only with the complexity andmuddle of the Old Regime and its terminology, but also with the complexityand muddle of our concepts and their terminology. First of all, this wouldprobably have resulted in very poor reliability. But, even more importantly,the many other rich resources of the cahiers de doleances would go unexplored.A few studies bounded by our a priori notions of what is interesting wouldbe produced, but not much more. If our theorizing improved after the coding- in other words if we learned anything in the course of this enterprise-we would need to code the cahiers all over again to test our new insights.Moreover, other researchers, with different substantive and theoretical inter-ests, might well wish to make use of these data, and would not be able touse such data.

    To avoid this dilemma, we attempt to separate radically the activity of thecoder from the activity of the scientist who will use this material for the studyof eighteenth-century France. By (figuratively) joining the human coder toa computer we can establish such an appropriate division of labor in a two-step process. The human coder isolates a grievance in the text, and translatesit into a standard, mnemonic set of symbols which both humans and com-puters can read. The scientist, on the other hand, through the use of com-puter programs, subsequently specifies conditions for retrieval and scoring,and uses his sociological and historical judgment as to what kinds of concretegrievances are relevant to his various purposes.

    The coder's job is to translate, as one translates, say, from French to En-glish. In this case he translates from French to a highly limited vocabulary,a syntax and grammar simplified for ease of computer programming. If therewere millions of ways in which to express a particulardemand in the originalFrench, we expect, in the ideal, that there would be only one way of ex-pressing the same demand in our code. Hence the task of the computer, orrather of the scientist instructing the computer to find some grievance in the

  • 8/8/2019 Quantitative Studies of the French Revolution

    11/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 173collection, is rendered possible. Of course there is some distortion and someloss of' meaning in any translation. But in principle, as little analysis as pos-sible is built into the codes.

    Our code does not indicate the presence of demands for such an abstracttheoretical concept as "bureaucratization." What it does indicate is the pres-ence of such demands as those for the abolition of venal offices in the judi-ciary, roughly on the same level of concreteness as the demand appears inthe original document. It is the job of the scientist to call this a demand forbureaucratization if he so desires. Of course there may indeed be codes whosemeaning is relatively abstract, but these are used only when that abstractionis in the nature of the original statement, as in "All men are created equal."In sum, our aim is to avoid placing demands into categories of a higher levelof abstraction than the level of the original formulation. As is always thecase, in scholarship or in life, we cannot pretend that we have accomplishedour aim completely.

    This strategy has the twofold advantage of permitting the coders to oper-ate with relatively reliable low-level categories while, at the same time, alarge body of material - about 100,000 demands - is made available tomany analysts with varied interests, each of whom can group the materialin whatever abstract categories best serve his interests. Such rewards do notcome without corresponding costs. There is a considerable labor to be doneby the analyst, in deciding how to use this material to measure what he isinterested in, for the coders have not done this job for him. Experience in-dicates that the cost in time, however, is not excessive.14

    In the ideal, the set of symbols written by a coder could be translated backinto the original form with little change in the substantive meaning of thedemand. Given the nature of the materials, with their bewildering varietyof demands, complaints, suggestions, recriminations, arguments, and evenstatistics, such a task appears formidable. How is one to capture this diversityin a concrete code? And even more importantly, how can such a concretecode be made usable by coders who, after all, cannot be expected to mem-orize thousands of symbols? The solution to this problem was to create acode that (a) is analytic and (b) has mnemonic properties. The goal wasto have an orderly and rational arrangement of the diverse categories, sothat a coder does not have to remember the precise set of symbols neededto translate a particular demand, but can quickly find them in a manual,or even reconstruct them by a process of inference.

    The code manual is divided into sections representing major institutionalcategories of eighteenth-century France: Government, Economy, Religion,

    14. Computer programs have been developed to provide the analyst with completeBoolean freedom, in either retrieving grievances or constructingscale scores.

  • 8/8/2019 Quantitative Studies of the French Revolution

    12/29

    174 SHAPIRO, MARKOFF, AND WEITMANthe Judiciary, the Stratification System, the Constitution. Each division isrepresented by a mnemonic symbol. Government, for instance, is symbolizedby "G." These categories are then divided and subdivided into finer cate-gories, which, in turn, are represented by their own mnemonic symbols. Thecoder zeroes in on the precise code required by a search. To follow throughone possible search: Government is divided into Finances, Taxation, Ad-ministrative Agencies, the Military, the King, and Regional and Local Gov-ernment; Taxation is divided into Tax Advantages, New Taxes, ExistingDirect Taxes, Existing Indirect Taxes, Direct Tax Agencies, Indirect TaxAgencies and Tax Administration; the Existing Direct Taxes are dividedinto the Capitation, Vingtiemes, Taille, Impots Accessoires 'a la Taille, Im-pots Reels, and Imp'ts Personnels; and the Taille is subdivided into TaillePersonnelle, Taille Tarifee, and Taille Reelle. In looking for the code for aparticular institution or practice, the coder flips through the pages of thecode manual, following a branching process, until he locates the precise coderequired. For example, a coder encountering a demand concerning the taillepersonnelle would code it as G TA DI TP -standing for Government-Taxation-Direct Taxation-Taille Personnelle. Thus the code manual is ana-lytically arranged, and the final combination of mnemonic symbols -thatis, the code -stands for a concrete demand. The hierarchically nested setof categories we have constructed is an indispensable aid in the translationof the prose of the cahiers into code. Our intention is to avoid delimiting thestructure of analysis, for the analyst is free to interpret the concrete code ashe wishes. In the example given, he need not group the taille personnelle forhis purposes with other governmental functions at all, since the computer isable to identify such grievances specifically, and not merely as "taxation"demands. On the other hand, if an investigator is interested in all demandsin the category of taxation, he may wish to utilize our classification of theconcrete taxation demands. If so, he need merely ask for retrieval of thosecodes beginning "GTA" regardless of the more detailed specification thatfollows.

    A distinct set of symbols was created to translate the action specified ineach demand. This action may be quite precise ("Re-establish") or it maybe extremely vague ("Do something about."). As there are about one thou-sand institutions and about fifty actions for which we have codes, the num-ber of combinations defining distinct possible codes is enormous. In additionto "Standard Action Codes," which might occur in any institutional context,such as "Abolish," "Equalize," "Modify," "Simplify," "Improve," "Elim-inate Abuses," "Render Responsible," we also allow a large number of "Spe-cial Action Codes" which are institution-specific. For instance, the demand"Only members of the Third Estate may be deputies of the Third in the

  • 8/8/2019 Quantitative Studies of the French Revolution

    13/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 175Estates General" is a frequent and important demand. Since the action de-manded is specific to the Estates General, we created a special symbol for it.

    The flexibility of our coding system is important, and can be illustrated bythe choices open to a student of grievances regarding taxation. He may wishto retrieve the location of all demands of a defined category, so he can con-sult their exact wording in the documents themselves, using our codes as anindex. Or he may wish to produce a quantitative score, to study the relation-ship between a defined class of grievances and some other variables (whichmay be other grievance scores, or characteristics of the Old Regime, or theincidence of revolutionary events, or all of these). In constructing a score,he may wish to increment the score if a document mentions taxation at all,or only if it mentions a direct tax, or only if it mentions the vingtieme desbiens fonds (a particular direct tax). Furthermore, he may wish only to addto the score, or retrieve the location of the grievance, if the demand is ex-pressly for a reduction of taxation, or for increased efficiency in its admin-istration, or for a standardization of its rate over France, or if privileges ofexemption are to be abolished, or established, or maintained. Finally, he canlink any such specification with any number of others by the terms "and," "or,"9and "not": for example, retrieve those demands for either a reduction or anabolition of either the vingtieme des biens fonds or the capitation.'5There is one other special feature of our coding procedures that must bementioned. It was clear at the outset that our ideal of perfect concretenesswas not going to be realized. Therefore, we built a capacity for self-improve-ment into our system. Suppose a coder encounters a reference in the textto a direct tax that is not yet included in our list of direct taxes and for which,hence, there is no specific symbol. The coder is then instructed to indicatethat the text refers to a "miscellaneous" direct tax. (He writes "G TA DI 0":

    15. The measurementof the inter-coderreliability presents severe problems with thiskind of code, and we have found the standard techniques to be inapplicable, but wehave evidence that agreementbetween coders is high. Moreover, analyses already carriedout have yielded meaningful results. See Gilbert Shapiro and Philip Dawson, "SocialMobility and Political Radicalism: The Case of the French Revolution,"Dimensions ofQuantitative Research in History, ed. Robert Fogel (Princeton, 1972), and Markoff,"Who Wants Bureaucracy?"There are a number of complexities in these documentsthat prevent the method justdescribedfrom yielding anything approaching complete success. Especially troublesomeis the possibility that grievances may be coupled semantically. In such a demand as,"Either expel the Jews from the Kingdom, or treat them just like other Frenchmen,"the interdependenceof the two parts of this statement needs to be taken into account.We cannot explore fully such issues here. And it must be other scholars, and notourselves, who pass final judgment on our success or failure, based on the substantivestudies which we are now beginning to complete and a far more detailed statement ofour content analytic techniques and their associated problems which is currently beingprepared.

  • 8/8/2019 Quantitative Studies of the French Revolution

    14/29

    176 SHAPIRO, MARKOFF, AND WEITMANGovernment, Taxation, Direct, Miscellaneous.) In addition, the coder indi-cates as precisely as possible the identity of this new item in a space reservedfor "Remarks." If, at some later time, we are not satisfied with the precisionof our code in the Direct Taxes category, we may instruct the computer toretrieve all instances of miscellaneous direct taxes, so we can read the coder'sremarks and invent new symbols if we so desire. Other coding difficulties arehandled analogously. For example, if a coder does not understand the text,because of some obscure historical reference, he writes a special code to alertus to a "historical problem" and then copies that portion of the text in thespace reserved for "Remarks." The flexibility afforded by computer retrievalscontinually enables us to improve our coding to the desired level.

    Along with the code for the grievance content, we have recorded a set ofsymbols to indicate its source. Not just the document, but the page number,and the location on that page as indicated by a special scale, is entered intoour electronic file. If we or others ever wish to go beyond the coder's remarks,we can return to the original text for further details with little difficulty. Forexample, should one subsequently decide that our classification of demandsdealing with the law of commercial bankruptcies was inadequate for someresearch purposes, one could locate these demands in the cahiers quickly,and recode them in greater detail, thereby avoiding the major expenditureof time and labor that would be otherwise required to locate the relevantpassages in the text.

    This means that an important by-product of our work is the creation ofa highly detailed index to a large number of cahiers. By sorting our electronicfile according to the codes for grievances, and then printing the location in-formation along with the grievance, we can provide a valuable bibliographictool. Even scholars with no interest whatsoever in quantitative researchmight well want to be able to locate cahiers containing interesting discussionsof some aspect of French society. If a particular scholar wants, say, to findthose cahiers of the Nobility which contain discussions of the relation of theFrench Church to Rome, our index enables him to find at once the precisetextual locations of all demands dealing with this question in the coded docu-ments.

    IV. THE SAMPLE OF DOCUMENTS

    Given limited resources and the enormous number of documents produced,it was, of course, necessary to sample. We wished (a) to obtain as completeas possible a picture of French public opinion at the beginning of the Revo-lution; (b) to utilize those classes of cahiers for which we could compute(and thereby correct) biases in the sample; and (c) to utilize published docu-

  • 8/8/2019 Quantitative Studies of the French Revolution

    15/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 177ments where possible (for the costs of obtaining and coding manuscripts arefar higher). These three criteria led to a collection consisting of virtually'6all the extant "general cahiers" of the Nobility and the Third Estate (that is,those documents produced by the Nobles and commoners at the final stageof the electoral process, and carried to Versailles) and a sample of 826 parishdocuments clustered in 47 bailliages.17 The first two categories constitutequite a dense sample: they make up 83% of those produced in 1789;18 theparish sample is far less dense, but our preliminary analysis, summarizedbelow, indicates that it is only slightly biased. Our study of the characteristicsof sampled bailliages as compared with the characteristics of all of Franceyields as a by-product (discussed more thoroughly below) an analysis of thedifferential rates of survival of historical documents, and a measure, for thiscase, of the bias introduced into historical study from this source. But themajor purpose of this analysis is to correct biases in our sample, for to dis-cover biases is to discover, at the same time, how to eliminate them throughthe assignment of appropriate weights.

    V. EXTERNAL DATAWe have come to call "external" all data other than content codes character-izing the cahiers themselves. These data bear either upon regional variations

    16. A small number of general cahiers were not coded due to special circumstances.For example, the cahier of the Third Estate of Nemours is so long as to be prohibitivelycostly to code. The cahier of Dauphine was drawn up by the provincial Estates ofDauphine, a procedure differing greatly from that by which the vast majority ofdocuments were produced. For details see Markoff, "Who Wants Bureaucracy?",Appendix.17. To construct the sample, we began with all those bailliages covered in the officialseries, published under the auspices of the French Ministry of Education by the Com-misson de Recherches et Publication des Documents Inedits sur l'Histoire Economiquede la Revolution Frangaise. To these we added the volumes put out by a number ofdepartmentalcommittees which followed reasonably similar criteria of editing, and arethus comparable.From this list of bailliages, we excluded all those having fewer thanten published parish cahiers, leaving a total of 47 bailliagesin our sample. We are grate-ful to Marc Bouloiseau for his advice and aid in collecting information essential tosamplingdecisions.18. This is not quite the same as saying that 83% of electoral districts are repre-sented. The reader will appreciate that the complexities of the electoral procedure donot permit a single figure to represent "proportion of assemblies whose documents areincluded in our sample." That the three orders had the option of drawing up cahiersin common, that special rules created some circumscriptions for the Third Estate thatdid not exist for the Nobility (e.g., the town of Valenciennes), that an order thatparticipatedin the drafting of a joint document might also produce one on its own (asin Amont-a-Vesoul) - all these are considerations which make it futile to attempt tosummarizethis problem with a single number. It is clear, nevertheless, that the degreeof representation n our sampleis high, howeverreckoned.

  • 8/8/2019 Quantitative Studies of the French Revolution

    16/29

    178 SHAPIRO, MARKOFF, AND WEITMANin conditionsof social, economic,and political structure n the Old Regime,or upon the distributionof revolutionary r antirevolutionary cts of somesort, duringthe revolutionary eriod. Three sets of criteriahave guided oursearchfor external data. First and foremost we are interested n data thatare significant, n light of our purposes. Second, we are interestedonly indata presented n the form of regionalbreakdowns or the whole of France.While a few individualobservationsmightbe missing (just as data are oftenmissing for isolated units in almost any statistical study), we would notrecordany source which provides data for only one region of the country.Finally, we have considered he sources and methodsof each item critically,and rejectedsome and acceptedothers on groundsof quality.It should bepointedout that in assessing he qualityof a source of data, we do not re-strict ourselvesto an examinationof the methods,if known, by which theinformationwas gathered.We can also check to see if the data in questionare in plausible agreementwith otherrelevant data. In the terminologyde-velopedfor the evaluationof the qualityof research nstruments,we wouldsay that we look into "criterion alidity"as well as "face validity."'9We donot necessarily eject a sourceof data if we know nothingwhatsoeveraboutits methodsof collection.For some crucialvariables,wheremorerespectableinformation s totally lacking, an informedguess of a strategically ituatedeighteenth-centurybserver is a good deal better than nothing. Moreover,our fundamentalmethodof researchstrategy, he searchfor meaningfulco-variations,will simplyturn up nothingin the face of much randomerror.If a data source is systematicallybiased,we can only protectourselves andour readers romfalse findings hrough he use of such historicaland socio-logical insightas we happento possess and comparisonsamong independentsources.But if dataare subject o randomerror,we will simplyhave no cor-relations o scrutinizeor report.It follows that conclusionsbased on consis-tent, observed statisticalrelationships an, when such questionabledata are

    19. Armand Brette, whose great work on the convocation of 1789 puts all laterworkers in his debt, claims that the population figures collected by the central govern-ment for the purpose of allocating deputies equitably to the bailliageshad little relationto reality, because local officials, polled by the central government for this purpose,falsified their reports to the point of ludicrousness (a frequent conclusion of Brettewith regard to Old Regime administration). This is to attack the face validity of thesefigures: on the face of it they look silly. We have discovered, however, that the num-ber of deputies assigned to the electoral districts is in excellent agreement with thedistribution of population for France as a whole, which leads us to very differentconclusions than Brette's. In other words, we have verified the criterion validity of thesame data. Of course an individualdata point may be wildly inaccurate, but the generalpattern of the series as a whole, its image of regional variation, cannot be. (Put moretechnically: correlational methods do not require the exact measurement of individualdata points. They require, more weakly, that the distributionof the observations be alinear transformation of the "real" values.) See the Introduction to Armand Brette,Recuell de documents.

  • 8/8/2019 Quantitative Studies of the French Revolution

    17/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 179involved, be given more credence than conclusions based on the absence ofrelationships.

    The extraordinary quantity of data of apparently acceptable quality, andthe great many facets of the Old Regime and of revolutionary behaviorrendered measurable thereby, was a pleasant surprise. Some data derive fromthe statistics gathered for administrative purposes by the eighteenth-centuryroyal government itself, notably tax assessments and yields - tax assess-ments are marvelous data, being simultaneously an important burden andan indicator of the activity being taxed - population figures, and indicatorsof the extent of commercial activity.20 Other data are provided by the de-tailed monographic research of historians, for example Greer's statistics onthe victims of the terror (classified by social class) and on the emnigres,Louisde Cardenal's material on the location of clubs populaires, Gustave Bord'senumeration of the Freemason lodges on the eve of the Revolution, or JeanSuratteau's detailed studies of voting in the elections of l'an IV. Still otherdata, while not quantified, are available in quantifiable form. Brette's researchinto the occupations of deputies to the Estates-General was, through thecreation of suitable categories, turned into quantitative data classifying bail-liages by the occupational characteristics of their delegates. Other data stillcome from historical researches and publications carried out by variousFrench ministries in the nineteenth century for comparative purposes: forexample, the important researches on literacy of men and women over threecenturies that Maggiolo carried out for the Ministry of Education, or themany data series put out by the Service de Statistique. Sometimes maps con-stitute important sources of data. A road map of the major eighteenth-centuryroutes royales, when overlaid with a transparent bailliage outline map, per-mitted the coding of the length of roads - and the number of intersections-in each bailliage.A good deal of this material was found - sometimes sought for and some-times discovered accidentally - in our reading of the available literature.Some of our finds we owe to the friendly assistance of several historians;others were run across during our own archival explorations. In all, we haveassembled literally hundreds of data series for the Old Regime and the Rev-olution.21

    20. Besides information collected routinely, as an integral part of the everydayprocess of administration,a number of important special inquiries were launched bythe controllers-generalin the eighteenth century, which provide highly significant in-formation. See Bertrand Gille, Les Sources statistiques de l'histoire de France: Desenqu'tes du XVIIe si'cle a 1870 (Paris and Geneva, 1964).21. Before coding these data sources, it was necessary to establish a master list ofeach of the ecological units for which we have data. Extensive difficultiesand compli-cation arose which made necessary considerable investigation into the historical facts.Many bailliages, for example, were known under more than one name. Cities and

  • 8/8/2019 Quantitative Studies of the French Revolution

    18/29

    180 SHAPIRO, MARKOFF, AND WEITMANOne difficult problem in the use of this material could generally be avoided

    in studies of post-revolutionary societies, but is virtually impossible to avoidhere. Unfortunately, almost none of our data are broken down regionallyby bailliages, as are the cahiers. Most available data series are presented forother administrative divisions of France: generalites, dioceses, cities, departe-ments, provinces22 and the jurisdictions of various kinds of mixed admin-istrative-judicial bodies (e.g., the greniers a sel), or subdivisions of any ofthe above. The boundaries of these units generally do not coincide with eachother, and even less with those of the bailliage. Nor are these boundariescompletely stable. The division of France into generalites underwent severalchanges in the course of the eighteenth century. The map of departmentssimilarly underwent changes: thus the department of Rhone-et-Loire wasdivided in two to weaken the influence of its rebellious principal city, Lyon,while 1808 marked a change in the boundaries of several other departments.

    Some technique for assigning to one type of geographical unit approximatescores based on data series which we have recorded for another type of unit isclearly required if we are to be able to correlate our different data series. Forexample, if we wish to study the relationship between some indicator of rev-olutionary behavior, such as the incidence of the Terror, and cahier contents,such as noble conservatism, we face the difficulty that the former are pre-sented to us by department, while the latter are available only by bailliage.We require a procedure which will transform either the departmental datato the bailliage level, or the bailliage data to the department level.

    We have encountered two different, cross-cutting complications in oursearch for a solution to this difficulty. First, data in the form of proportionsmust be handled quite differently from absolute figures. Second, some vari-ables seem appropriately transformed on the basis of the proportion ofpopulation shared between two ecological units, while others seem moresmaller units with identical names abound. Lists of data for generalitis may identifythemselves as data for provinces. Borders changed. Some authors give cumulative figuresfor the province of Normandie rather than separate figures for the three generalizesof which it was composed (Caen, Alengon, Rouen). In all such cases we have developedeither special procedures for coding data, or for the later computer-processingof datawhich reduce the various lists of differentsources to a reasonable common ground.22. The province is an elusive area within which local customary law applied. Inspite of the historians'consensus that there were no fixed, or even readily approximableboundariesto the provinces, one finds quantitativestatistics from the Old Regime pre-sented in this way. There is, for example, a little known source of regional variationsin the number of nobles and it is by province (Bonvallet-Desbrosses, Richesses etressources de la France: Pour server de suite aux m6yens de simplifier la perceptionet la comptabilite des deniers royaux [Lille, 1789]). This is somewhat analogous topresenting data on the number of women in "New England" or on the "west coast."It is far from obvious to the reader just which states are included in the categories.First priorities go to approximating,as best we can, the map intended by the author.

  • 8/8/2019 Quantitative Studies of the French Revolution

    19/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 181appropriately transformed on the basis of the shared area. Thus we havedeveloped four different transformation procedures.

    In order to perform appropriate transformations, it was necessary to se-lect the most accurate and reliable maps depicting the borders of eighteenth-century regional units.23 With map transparencies and a ruled grid, wemeasured the total area of each geographic unit and of its areas of overlapwith others. Cities were attributed to larger units by consulting a variety ofgeographical dictionaries, maps, and other sources. This information is usedby a set of computer subroutines that may be applied to transform any dataseries to a different kind of area unit. We have also tested these procedures,by studying data which we have recorded independently for different units.The results of these tests were highly encouraging.24All this material taken together -the content analysis of the cahiers, thecomputer programs for manipulating the cahiers data and constructing scales,the coding manual, the coded external data series on Old Regime social struc-ture and revolutionary behavior, the proportions used to assign approximatescores to compare data series gathered by different ecological units, the com-puter programs enabling one to do all this with relative facility-will con-stitute a resource for the study of social upheaval, for the study of publicopinion, and for the study of French history that will be made available toother scholars for their own purposes. As a data archive, and a body ofprograms to manipulate that data, this project will, we hope, make a con-tribution to research, beyond our own explorations of this material. Thepreparation of this archive, while not yet complete, is far enough advancedfor us to announce its availability.25In fact, enough material has been as-

    23. For department, bailliage, and generalite lines, we found the maps in Brette(based upon Cassini) the best available.24. For example, from department population statistics, approximate populationscores for generalitys were estimated. These were then compared with independentfigures deriving from sources reporting generalite data directly. The approximationand the "real" data show correlations up to .97. Further information on this test willbe found in Markoff, "Who Wants Bureaucracy?"25. Tapes, containing the coded cahiers data, the external data series, the programsrequired to obtain information (retrieve, score, transform into other ecological units,etc.) will ultimately be deposited at the Inter-university Consortium for PoliticalScience Research at the Institute for Social Research, University of Michigan, theSocial Science Information Center of the University of Pittsburgh, and the Centre de

    Recherches Historiques in Paris. Ultimately, we hope to join our files with studies ofpolitical upheaval in nineteenth- and twentieth-century France directed by CharlesTilly, and with the codification of the Statistique generate de France, currently con-ducted by I.C.P.R. in collaboration with the Centre de Recherches Historiques, to pro-vide an instrument for the study of social history and political upheaval over twocenturies. The conditions of access and procedures for use of the material will be an-nounced by these depositories in the light of their respective administrativepolicies.

  • 8/8/2019 Quantitative Studies of the French Revolution

    20/29

    182 SHAPIRO, MARKOFF, AND WEITMANsembled so that some studies have been completed, and we are currentlyworking on others.

    VI. SOME ILLUSTRATIVE STUDIESSome of our findings seem to be of sufficient general interest to warrant asummary here, even though more elaborate reports would be required fora full, systematic presentation of our methods and results.20

    1. Representativeness of Document CollectionsThe first set of findings emerge from our efforts to decide upon a "sam-

    pling frame" for our study of the parish cahiers: that is to say, a collectionof documents from which we would draw the sample to be coded. Whilethe study was launched for this particular technical purpose, we feel thatthe findings bear interestingly on general historiographic issues as well.

    Three populations of parish documents have been evaluated: the collectionof extant and catalogued manuscripts, the collection of published documents,and the collection of those published in the official series.27Since national data cannot be found characterizing the parishes of France,and most publications reflect the convocation procedure in publishing togetherthe parish cahiers of a particular bailliage, we were almost compelled toform a sample "clustered" by bailliage. In other words, we have attemptedto characterize each bailliage with regard to the frequency of demands to befound in the cahiers of the parishes represented at its assembly.

    After discovering (or estimating) the total number of parishes authorizedto meet in each bailliage assembly of the Third Estate, we were able to com-

    26. The interpen.etrationof our various efforts in the establishment of the archiveof cahiers-codes and external data makes it impossible to attribute any part of thiswork to any one of us. But, while we have drawn upon the others' advice and informa-tion for our individual efforts at data analysis, the substance of the findings reportedmust remain the responsibilityof the authors individually, in the works cited in foot-notes 11 and 15 and in articles concerning the sampling procedure and its historio-graphical lessons and regional variations in eighteenth-centurypopulations which wehope to publish shortly.27. Since the nineteenth century, the Commission de l'Histoire Economique etSociale de la Revolution Franraise has published dozens of volumes of cahiers in theseries, Collection de documents inedits sur l'historie economique et sociale de la Re-volution fra7icaise. Because this is at least one mouthful, we have opted to followBeatrice Hyslop in referringto this series simply as the "D.I." series. The rules of theCommission require editors to publish all unpublisheddocuments of a given bailliage,to publish the text completely, and not summariesor excerpts, and to provide necessarybackgroundinformation on the convocation procedure, local socioeconomic conditions,and political events.

  • 8/8/2019 Quantitative Studies of the French Revolution

    21/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 183pute the proportion, in each bailliage, of parishes represented in (1) the col-lection of surviving manuscript documents; (2) the collection of publisheddocuments; and (3) those in the official ("D.I.") series. These proportions,characterizing the bailliages, could then be correlated with other data derivedfrom the general cashiersor from external sources to evaluate the extent towhich the rate of survival (or of publication) of cahiers is related to the ex-pressed sentiments of the higher assemblies or the social character of thedistricts.

    We have already mentioned one finding, regarding the representative char-acter of the surviving documents. To expand briefly upon this theme, wehave found that, generally speaking, a preliminary analysis shows only rathersmall correlations between these proportions and our indices of social struc-ture, revolutionary behavior, and of content of the general cahiers.28Appar-ently, the accidents of history have operated more or less at random andarchivists and editors have not generally permitted similar commitments toaffect the selective retention or selective publication of parish cahiers. Fur-thermore, the official (D.J.) series shows no greater deviation from nationalnorms than the other two populations. Its greater conveniences as a samplingframe-standard criteria of inclusion, modern orthography, integral repro-duction of documents, extensive background information, elaborate analyticalindexes, and availability -make this an occasion for celebration.

    Some of these small correlations however, are of great interest. Amongother findings, we discovered that the bailliages most highly represented inthe three document collections were more highly concerned with participa-tion in the electoral process: they were more likely to have their cahierprinted, to attach an imperative mandate to it, and to establish committeesof correspondence to sustain relationships between elected delegates and theirconstituencies. (We do not know if these correlations are due to a greaterpreference of archivists or editors for the preservation or publication of suchcahiers, or to the fact that such activity was more likely to produce multiplecopies of the cahier in 1789, thus decreasing the likelihood of loss.)

    Executions during the Terror, and emigrations, are both significantly lowerin those departments heavily represented in the D.I. volumes, which, if Greer'sinterpretations are correct, would indicate that they are more pro-revolu-tionary.29The D.L. series also seems to be slightly biased in favor of more

    28. In the early stages of this research we utilized the content analysis of the generalcahiers made by Beatrice Hyslop for her book French Nationalism in 1789 Accordingto the General Cahiers (New York, 1934). These data, very generously furnished byHyslop, were the basis for the findings mentioned here as well as for Weitman.29. Donald Greer, The Incidence of the Terror During the French Revolution: AStatisticalInterpretation (Cambridge,Mass., 1935) and The Incidence of the EmigrationDuring the French Revolution (Cambridge,Mass., 1951).

  • 8/8/2019 Quantitative Studies of the French Revolution

    22/29

    184 SHAPIRO, MARKOFF, AND WEITMANrural areas.But none of these correlationsare very high, indicating,we re-peat, the most basic finding: that all threedata collectionsprettyadequatelyrepresent he nation in 1789.2. The Importance of Urbanization

    One majorfindingthat has imposeditself upon us (often unexpectedly)in a numberof distinctstudieshas been the importanceof urbanization.Thefirstcase was a preliminary tudy,carriedout usingHyslop'scontentanalysis.For a preliminaryook at the way the cahiersbehavestatisticallywe chosethirty-sixof Hyslop'scategorieswhich seem to reflectsupport or or opposi-tionto democratic eformof the society. Ourcriterionwas essentiallywhetherthe grievancedemanded omepoliticalor socialequalities or underprivilegedgroups, or the eliminationof absolutistrestrictionson freedomof expressionor politicalaction.For example,demands or religious oleration,equalityoftaxation,the abolitionof feudal dues, and universalsuffrageare included.In this, our first pass at the problem, we attemptedto operationalize heclassicconceptionof the FrenchRevolutionas the most dramatic nstanceofthe democratic evolutionof the eighteenth entury,expoundedmostrecentlyand eloquently n RobertPalmer's wo volumes.Sevenvariablescharacterizinghe old regimewere analyzed:total popula-tion; the rate of the dime, or tithe, an involuntarychurchassessmentuponagriculturalproduce which varied throughFrance; the proportionof thepopulation hat is "urban" ccording o ArthurYoung as correctedby HenriSee; the total of all taxes paid by the unit, as estimatedby JacquesNecker;the vingtiemede biens fonds, a tax on agriculturalwealth widely regardedas the least unfair of the old regime;the vingtiemed'industrie,a similartaxon artisanand industrial nterprise; nd a measureof tax privileges (one of

    the major sources of grievances n the old regime) from the work of theComit6de Contributions f the ConstituentAssembly.Seven indices of revolutionarybehaviorwere selected for this study: thecontribution atriotique, n interimvoluntary ontribution etweenthe periodof the old andthe establishment f the new basis of taxation; he proportionof clergywho took the oath to supportthe Civil Constitutionof the Clergywithoutreservations;he GreatFear of the summerof 1789, widespreadout-breaks of panic sparkedby rumorsof a plot by the aristocracy o set loosebrigands o murder he people;the so-calledMassacresof September, laugh-tersin the prisons n 1792, againsparkedby rumorsof conspiracy, ollowingseriousmilitaryreverses; he total numbersexecutedduringthe Terror;thenumberwho emigrated;and the number of elected delegatesdesignatedbySydenhamas "Girondins."30

    30. These thirteen variables were selected for their theoretical interest or historical

  • 8/8/2019 Quantitative Studies of the French Revolution

    23/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 185Various relationships appear between democratic sentiment and old regime

    variables. The Clergy seem to have been more democratic in populous de-partments; the Nobility in more urban areas, and where taxation was heaviest.Most interesting, perhaps, the Nobility seem to have been more democraticwhere they enjoyed the greatest taxation privileges. As for revolutionary be-havior, the Nobility seem to have been more democratic in those areas wherethe Great Fear broke out, a paradox only for those who have not read Tocque-ville on how the ruling class, in adopting the views of enlightenment, set revo-lutionary models for the masses, and undermined their conventional faith inthe social structure.

    Examining our selection of variables characterizing the old regime, onestands out as explaining more of the variance in other old regime variables,and in revolutionary sentiment and behavior, than any other: the proportionof the population classified as urban by Henri See and Arthur Young. Thiswas unexpected, despite the warnings we might have heeded from CharlesTilly and Paul Bois. This urbanization measure is correlated with the totaldirect taxation, the vingtieme des biens fonds, the vingtieme d'industrie, andtax privileges, among the old regime variables; and with the total numberexecuted in the Terror, the contribution patriotique, and the presence ofGirondin deputies among the revolutionary behavior set.31importance from among those data for which processing had been completed at thetime of the trial run reported here. Our plans for the future include a more thoroughstudy along similar lines, using the much larger set of data we have collected.Our sources for these data were: Population and total taxes: J. Necker, De l'Ad-ministration des financesde la France (Paris, 1784); P. Gagnol, La Dime ecclesiastiqueen France au XVIIIe siecle (Paris, 1910); Urbanization: Arthur Young, Voyages enFrance, ed. Henri See (Paris, 1931); Vingtiemes and Tax Privileges: J. Mavidal andE. Laurent, Archives parlementairesde 1787 a 1861 (Paris, 1867- ), lere serie,Vol. XXVI, 513ff; Contributionpatriotique:ibid., Vol. XXIV, 4; Clergy Oath: P. Sagnac,"Etudestatistiquesur le clerge constitutionnel et le clergy refractaire,"Revue d'HistoireModerne et Contemporaine 8 (1966), 97-115; G. Lefebvre, La Grande peur (Paris,n.d.); P. Caron, Les Massacres de septembre (Paris, 1935); Greer, Incidence of theTerror,Incidenceof the Emigration.31. A number of other findings from this study are of interest. (1) The literatureon the revolution seems to be divided between those who see it as a bloc, and thosewho see it as a succession of distinct events, each led and participated in by differentsegments of the population. Thus far, the data of this study support the latter view.Relationships between revolutionary behavior variables are extremely few, and small,only one appearing significant: between the Great Fear and the total of emigration.(2) We have already outlined the effects of urbanization. Despite the positive corre-lation between democratic sentiment among the Clergy in the spring of '89 (when thecahiers were written) and the population of the departments,there is a negative corre-lation between the Clergy's willingness to take the oath in support of the civil constitu-tion, and population a few years later. This, like the absence of many significantrela-tionships among various revolutionarybehavior indicators, should warn us against anyoversimplifiedview of the historical process as a simple two-person game, with per-manent, easily identifiedopponents. (3) The Girondins came from more heavily taxed,

  • 8/8/2019 Quantitative Studies of the French Revolution

    24/29

    186 SHAPIRO, MARKOFF, AND WEITMANOur subsequent work has borne out this impression of the power of urban-

    ization as a significant historical force and as a factor in the events of theFrench Revolution. The studies, of which some more detail is indicated be-low, of Shapiro and Dawson on radicalism in 1789, of Weitman on demo-cratic sentiment, and of Markoff on the surprising tendency of the cahiersto demand a more bureaucratic government, are all agreed, despite the diver-sity of what they seek to explain, that urban growth is a factor of majorimportance. The Shapiro and Dawson study, for example, though commencingas an examination of the widely-held thesis that blocked upward mobilitywas a leading cause of bourgeois radicalism in 1789 found, incidentally (butnot insignificantly), that the effects of urbanization had far greater effects onradicalism than did mobility rates.3. Population Data

    For two reasons, we have been led to an extensive analysis of eighteenth-century French population information. In order to compare ecological unitsof differentsizes, it would obviously be necessary to remove, and hence first tomeasure, the effect of sheer scale. In addition, the pressure of populationincrease is widely regarded as one of the important social forces of thecentury.

    We have collected together over twenty-five estimates of the populations ofregions of France, some arrived at by extremely different means. Our firstanalytical objective was to discover if it made a difference, and if so, howmuch, whether one estimate or the other was accepted. The results of aprincipal-axis factor analysis shows that the various estimates are close linearfunctions of one another, and, although they would then provide differentabsolute populations for a given region, their variations from region to regionare in effect identical. As a result, we feel comfortable in using any of thesepopulation estimates or, most sensibly, an average of them.4. The Effects of Ennoblement

    In collaboration with Philip Dawson, Shapiro has attempted to use ourmaterial to put to the test the common claim that the bourgeoisie in eighteenth-century France was led to revolutionary action by the frustration of beingdenied access to noble status. The alternative hypothesis was most clearlyexpressed a century ago by Tocqueville: that increased opportunities for en-noblement in the eighteenth century actually increased the radicalism of theand hence wealthier areas. (4) The emigration was strongest where pre-revolutionarytax privilegeswere greatest.

  • 8/8/2019 Quantitative Studies of the French Revolution

    25/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 187middle classes. The issue regarding the sources of the French Revolution isonly one instance of a more general, theoretical problem in social dynamics:the relationship between upward and downward mobility and political radical-ism or conservatism.

    To test these conflicting hypotheses, the opportunities for ennoblement inthe eighteenth century were measured for each bailliage, and this independentvariable was then related to each of four radicalism scores derived from thecoded general cahiers of the Third Estate. (For purposes of comparison, theanalysis was repeated with the cahiers of the nobility.) The scales of radical-ism in the cahiers were, first of all, the total number of grievances expressed;second, the number of grievances reflecting demands for equality; third, thedegree of similaritybetween the contents of the cahier and the Declaration ofAugust 4; and finally, the similarity of the document to the Declaration of theRights of Man.

    It is apparent from the results that, regardless of the measure of radicalismused, the Third Estate in those bailliages with any ennoblement opportunitieswas more radical than the Third Estate in bailliages with no such oppor-tunities. Furthermore, where there were opportunities, there was some slighttendency for the Third Estate to be more radical where opportunities weregreater. And, finally, no such tendencies appear when we examine the noblecashiers.Hence it would seem that the hypothesis of Tocqueville is more inaccord with the data than the more popular theses of Dollot and Taine, orElinor Barber. This is probably because, as Tocqueville emphasized, thehistorical effect of an act of ennoblement upon its audience - those leftbehind - is greater than its effect on the presumably grateful, newly-privi-leged family, if only because there are so many more of the former. But theimportant finding of this study which we mentioned above is that opportunityfor ennoblement seems to have had extremely little effect of any sort onradicalism, especially as compared with the downright enormous effects ofurbanization.

    5. The Democratization of Public Opinion on the Eve of the Revolution:Marx vs. Tocqueville

    This research, by Weitman, is addressed to two sets of problems. First, wasthe French Revolution a mass phenomenon (as Tocqueville described it) ora class phenomenon (as Marx believed)? Second, which sector of the OldRegime - the economy (again, Marx) or the central government (Tocque-ville) - was primarily responsible for democratizing and revolutionizingFrench society?

    To confront these two theses, the general cahiers of the three estates were

  • 8/8/2019 Quantitative Studies of the French Revolution

    26/29

    188 SHAPIRO, MARKOFF, AND WEITMANsystematically compared.32 The demands were grouped into the followingcategories: demands for equal individual rights; demands concerning estateprivileges, the corporate-aristocratic regime, and the scope of the State; de-mands for relief measures on behalf of the poor; and demands concerningthe structure and operations of the central government.

    Systematic comparison of the cahiers of the nobility and of the ThirdEstate reveal that: (1) regarding the extension of equal individual rights toall, the cahiers of the nobility were, on the whole, as liberal as those of theThird Estate; e.g., 70 per cent were favorable to equal individual tax liability.However, the nobles' libertarianism tended to focus on "active liberties"(i.e., liberties to . . ), while the tendency in the Third Estate was to empha-size "passive liberties" (i.e., liberties from . . .). (2) Of the three estates, thenobility appeared, surprisingly, the least attached to its own status privileges.(3) None of the three orders evinced much attachment to the corporatestructureof the Old Regime. (4) The nobles entertained definite ambitions toplay a distinct political role. (5) Regarding attitudes on the role of the centralgovernment, the nobility was most unambiguously opposed to its expansionand most favorable to decentralization, whereas the Third Estate expressedthe largest number of demands both favoring and opposing expansion. (6)Both the nobility and the Third Estate were highly favorable to the impositionof stringent and extensive controls over the operations of the central govern-ment. Insofar as they differed, however, the Third Estate was more interestedin bureaucratic measures to render the state more efficient,whereas the nobilitywas more interested in constitutional measures to bring the central govern-ment under the rule of law.

    In sum, the Tocquevillian conception of public opinion in pre-revolutionaryFrance as pervaded through and through (i.e., across class lines) by theirresistible pull of democratic ideas and sentiments appears to be more con-cordant with these findings than the Marxian thesis. Insofar as the cahiers ofboth orders presaged the changes subsequently wrought by the Revolution,public opinion in general would appear to have been substantiallyradicalizedprior to the downfall of the old regime.

    The Tocquevillian theory attributesthis democratization of public opinion tothe profound social and cultural effects of over two hundred years of unevenbureaucratic expansion by the central government. The cahiers of the nobilityfrom areas heavily penetrated by the central government (the pays d'e'lec-tions) were compared with those from relatively autonomous provinces (thepays d'e'tats). Some of the findings were: (1) Noble public opinion in thepays d'e'ections was, on the whole, more egalitarian, more statist, more pro-

    32. This analysis was based on notes far more detailed than the content analysispublishedin Hyslop, French Nationalism in 1789.

  • 8/8/2019 Quantitative Studies of the French Revolution

    27/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 189bureaucratic, more constitutionalist, and more jealous of its status intereststhan was Noble opinion in the pays d'e'tats. (2) A comparison of urban andrural bailliages shows, however, that "urbanism"had a similarly democratizingeffect on noble public opinion. (3) Differences between the rural nobles ofthe pays d'elections and those of the pays d'e'tatswere, on the whole, muchmore substantial than corresponding differences between the urban noblesof these two types of regions. (4) Among the urban nobles, those from thepays d'e'lectionwere more democratic than those from the pays d'etat. Onceagain, the research supports Tocqueville's views.

    Secondly, Tocqueville's thesis leads one to expect that the conversion to ademocratic and radical ideology should be less characteristic of those parts ofFrance which still retained some measure of local autonomy (the pays d'e'tats)than of those regions more completely administered by the central govern-ment (the pays d'elections). A comparison of the cahiers of the nobility ofthe pays d'etats and the pays d'elections was carried out. Once again, theresults, generally speaking, support Tocqueville.

    6. Popular Demands for BureaucracyAnother of our studies, conducted by Markoff, concerns the social basesfor bureaucracies. The development of bureaucratic government in Western

    Europe (in the sense in which that term was used by Weber) has beenaccounted for, by Weber and later writers, in several ways. Some scholarshave looked to circumstances thought to render greater rationalization anecessity: for example, the growth of large and costly standing armies. Otherscholars have seen this process as the outcome of the long struggle betweenthe Prince and those with independent powers. Still others have looked to thecharacteristics of the social environment of government in order to try toidentify those features of social structure and culture which provide fertilesoil for the growth of bureaucracies.

    In this light the cahiers de doleances provide ample evidence of an unex-plored aspect of this subject. In 1789, France witnessed an enormous out-pouring of popular demands for bureaucracy. Almost all of what Weberasserted as the defining characteristics of bureaucratic government was de-manded in large numbers of documents. Indeed, examination of the distribu-tion of all grievances in the data collection shows that demands for morebureaucratic government are as prominent as those demands more frequentlysaid to characterize the French Revolution, such as those for liberty, equality,and for abolition of seignorial rights. We have, then, at least one importanthistorical case of widespread popular desire for a more bureaucratic govern-ment, and an important sociological phenomenon to try to explain.

    Markoff argues that first of all, a wide variety of changes had been taking

  • 8/8/2019 Quantitative Studies of the French Revolution

    28/29

    190 SHAPIRO, 1MARKOFF,AND WEITMANplace in France which served to increase the interdependence of distantpeople. This pattern of transformation included increased road construction,rising literacy, urban growth, increased population density, expansion of trade,increased administrative centralization, and (arguably) increased mobility. Itis then advanced that this increased interdependence leads in several ways toa desire for a more bureaucratic government. The increased importance ofdistant decisions, for example, produces a desire for administrators to becontrolled, so that they behave in predictable ways. In addition to thoseeffects which may be expected of any form of increased interdependence,there are special effects which may be expected of certain specific aspects ofincreased interdependence. Markoff proposes, for instance, following Tocque-ville, that the increased authority of the central government produced acultural climate conducive to still greater administrative centralization bymaking it difficult to conceive of any other way to solve problems.

    The documents were assigned scores according to the degree to whichcomponents of the multifaceted concept of bureaucracy were demanded.These included, among others, demands that recruitment and promotion bebased on specialized training, demands opposing the private appropriation ofpublic functions, demands that decisions be based on formal acts in writing,and demands for clear-cut hierarchies of authority. Using our collection ofexternal data, a wide variety of aspects of interdependence was measured.

    For example, the geographical distribution of several degrees of literacywas investigated. The lowest level of literacy was measured by the ability tosign one's name. Regional variations in the number of Jcoles secondaires andin the numbers of their pupils supplemented this information. Finally, weattempted to study the distribution of intellectuals through locating some oftheir characteristic institutions: the Freemason lodges, the various learnedacademies and societes and the eighteenth-century universities.

    Having obtained many measures of interdependence, and of demands fora more bureaucratic government, the covariation of the two types of data wasexamined. Broadly speaking, the data are consonant with the general hy-pothesis: it is indeed the case that indicators of literacy, for example, arecorrelated with bureaucratic demands.

    Since both interdependence and bureaucratizationare complex phenomena,further research is underway to determine the relative importance of differentaspects of interdependence, such as literacy, trade, or administrative cen-tralization, in producing such demands for different aspects of bureaucracyas the abolition of venal offices and the establishment of a governmentalbudget.

    VII. CONCLUSIONWe have directed this paper to describing the general aims and methodologyof our investigations into the French Revolution, and have tried to explain

  • 8/8/2019 Quantitative Studies of the French Revolution

    29/29

    QUANTITATIVE STUDIES OF THE FRENCH REVOLUTION 191the strategic interest of this event for sociological theory and the strategicinterest of the unique available data for sociological research. If we have saidnothing by way of apology as sociologists for treading on historians' turf, it isbecause it is our turf as well. If we have said little on the sometimes acrimo-nious debate about quantitative methodology within the historical profession(a debate into which we have quite inadvertently wandered), it is because wesee little point in more general statements of a programmatic or philosophicalnature. Assertions of the need for cooperation among sociologists and his-torians have become tiresome (if they ever were anything else). Historianswill scrutinize our results and take what they like and reject what they do not.Laments about tearing seamless webs are not very helpful, even if sincere.We hope that we have demonstrated the great potential of the research instru-ments we have been developing and that - although we do not for an instantfeel that these are the only ways to arrive at knowledge -we can beginanswering questions which have up to now been difficult even to ask.University of PittsburghandState University of New York,Stony Brook