On Lawfulness in History and Historiography

History and Theory 51 (May 2012), 172-192 © Wesleyan University 2012 ISSN: 0018-2656

ON laWfUlNeSS IN HIStOry aNd HIStOrIOgrapHy

Bert leUrIdaN aNd aNtON frOeyMaN1

aBStract

the use of general and universal laws in historiography has been the subject of debate ever since the end of the nineteenth century. Since the 1970s there has been a growing consensus that general laws such as those in the natural sciences are not applicable in the scientific writing of history. We will argue against this consensus view, not by claiming that the underlying conception of what historiography is—or should be—is wrong, but by contending that it is based on a misconception of what general laws such as those of the natural sciences are. We will show that a revised notion of law, one inspired by the work of Sandra d. Mitchell, in tandem with Jim Woodward’s notion of “invariance,” is indeed applicable to historiography, much in the same way as it is to most other scientific disciplines. Having developed a more adequate account of general laws, we then show, by means of three examples, that what are called “pragmatic laws” and “invariance” do in fact play a role in history in several interesting ways. these examples—from cultural history, economic history, and the history of religion—have been selected on the basis of their diversity in order to illustrate the widespread use of pragmatic laws in history.

Keywords: pragmatic laws, invariance, causation, historical explanation, stability, laws in historiography

I. INtrOdUctION

the use of general laws in historiography, understood as the scientific study of the past, has been the subject of debate ever since the Methodenstreit at the end of the nineteenth century and the writings of, among others, Wilhelm Windelband in the beginning of the twentieth century. In analytical philosophy, this debate rose to a height in the 1950s and 60s, following the controversial publications of carl gustav Hempel and ernest Nagel.2 It also played a large part in the general discussion about Marxist and post-Marxist historiography, as well as the second

1. We would like to thank Sandra Mitchell, elisabeth Nemeth, and lindsey Vandevoorde, as well as the anonymous referees of this journal, for their very helpful suggestions, criticisms, and com-ments. Bert leuridan is postdoctoral fellow of the research foundation – flanders (fWO). anton froeyman is phd fellow of the research foundation – flanders (fWO).

2. carl g. Hempel, “the function of general laws in History,” Journal of Philosophy 39 (1942), 35-48; ernest Nagel, The Structure of Science: Problems in the Logic of Scientific Explanation (london: routledge and Kegan paul, 1961). for an overview of this debate, see Theories of His-tory: Readings from Classical and Contemporary Sources, ed. patrick gardiner (london: allen and Unwin, 1959), and Philosophical Analysis and History, ed. William H. dray (New york: Harper and row, 1966).

On lawfulness in HistOry and HistOriOgrapHy 173

generation of the Annales and the New Economic History, especially with relation to the use of quantitative methods in historiography.3 Since the 1970s, however, there has been a growing consensus that general laws such as those in the natural sciences are not applicable in the scientific writing of history. We will argue against this consensus, not by claiming that the underlying conception of what historiography is—or should be—is wrong, but by contending that it is based on a misconception of what general laws such as those of the natural sciences are. the claim of this paper is that a revised notion of “law,” one inspired by the work of Sandra d. Mitchell, in tandem with Jim Woodward’s notion of “invariance,” is indeed applicable to historiography, much in the same way as it is to most other scientific disciplines. We do not aim here to introduce new basic methodological principles. rather, we will argue that historians unknowingly have been talking about general “pragmatic” laws on a large number of occasions. More specifi-cally, we will argue by means of a series of examples that historiographical dis-cussions often originate—either explicitly or implicitly—from questions about the stability of a given pragmatic law. Stability will therefore prove to be a central notion in the use of general laws in historiography.

In the next two sections we will first present Mitchell’s concept of “pragmatic law,” and then we will distinguish between causal and noncausal pragmatic laws by means of Woodward’s theory of “invariance under interventions.” In the third section we will show that our resulting conception of “law” is to be preferred over the traditional concept advocated by Hempel and Nagel, not only in histori-ography but in many other scientific disciplines as well. In the fourth section, we will give some initial remarks on how these concepts should be interpreted with respect to historiography. In sections five to seven we will apply these concepts to three different examples. these examples have been selected on the basis of their diversity in order to illustrate the widespread use of pragmatic laws. they come from cultural history, economic history, and the history of religion, and they treat topics in ancient history and early modern history. the first two are recent discussions, whereas the third relates to a classical discussion in the his-tory of historiography. each of them is more or less typical of a wide range of other historiographical discussions. from this it should be clear that the use of pragmatic laws is, and always has been, widespread in historiography, and that historiographical discussions are often based—either explicitly or implicitly—on disagreements or discussions about the stability of a pragmatic law.

II. pragMatIc laWS

the problem of lawfulness, that is, the question whether there exist general and universal laws, is not confined to historiography or even to the social sciences or

3. See Hervé coutau-Bégarie, Le phénomène “nouvelle histoire”: Stratégie et idéologie des nou-veaux historiens (paris: economica, 1983), 31-125; georg g. Iggers, Historiography in the Twentieth Century: From Scientific Objectivity to the Postmodern Challenge (Middletown, ct: Wesleyan University press, 1997), 51-96; david Harrison, “the ‘New Social History’ in america,” in Making History: An Introduction to the History and Practices of a Discipline, ed. peter lambert and phillipp Schofield (london: routledge, 2004), 109-120; Michael roberts, “the annales School and Histori-cal Writing” in lambert and Schofield, ed., Making History, 78-92.

bert leuridan and anton froeyman174

the humanities in general. Historians often believe that the use of universal laws of nature in the natural and the life sciences is straightforward and uncontested. In the following paragraphs it should become clear that this is not the case.

With the laws of physics such as Newton’s laws of motion in the back of their minds, philosophers of science in the second half of the twentieth century tried to define the concept of law. Within the logical empiricist tradition, a set of criteria was developed that gave rise to what now can be labeled “the received view of laws of nature.”4 We will list only a few of these criteria here. lawful generaliza-tions were, for example, defined as universal conditionals (sentences of the form “if something is A, then it is B” that hold for every case of a) that are (nonvacu-ously) true, and hence have no genuine exceptions.5

to distinguish lawful generalizations from merely accidental ones, several extra criteria had to be invoked. after all, it may happen that the universal condi-tional “all the screws in Smith’s car are rusty” is nonvacuously true and excep-tionless (if all the screws in Smith’s car do happen to be rusty), but no logical empiricist would have been prepared to call it a law. to count as a law, a gener-alization should be general or nonlocal, which means that it should not refer to particular objects or spatio-temporal locations such as Smith’s car. In addition to this, a law should also be projectible: its supporting evidence should not exhaust all its instances. In other words, the generalizations should be useful for genuine prediction. presumably this is not the case with “all the screws in Smith’s car are rusty”: to know that this generalization is true, we must first check every screw in Smith’s car, but then there is no screw left for which we can predict that it is rusty. finally, lawful generalizations, as opposed to accidental ones, somehow should be necessary and they should support counterfactuals (statements of the form “if A were the case then B would be the case,” when A in fact is not the case). the fact that all the screws in Smith’s car are rusty, if true, is just a contin-gent matter of fact and seems to have no modal force.

the received view is still influential in contemporary philosophy of history, albeit in a negative sense. Mark day’s introduction to the philosophy of history may serve as an illustration here.6 On the one hand, he writes that “[t]he legacy of positivism [of which the received view is part] still weighs heavily on philo-sophical discussion of the social sciences and of history,” and he rehearses the traditional criteria for lawfulness. On the other hand, he stresses that no scientific laws are to be found in history.7

4. the loci classici of the received view of laws of nature are Nagel, The Structure of Science, 29-78; carl g. Hempel, Aspects of Scientific Explanation and Other Essays in the Philosophy of Science (New york: the free press, 1965), 231, 265-268; and Nelson goodman, Fact, Fiction, and Forecast (Indianapolis, IN: Bobbs-Merrill, 1973), 17-27, 73-81. these authors had quite different views both regarding the precise interpretation of these criteria and regarding their relative impor-tance. Our presentation of their views hence is a simplified one.

5. a universal conditional(∀x)(Ax ⊃ Bx) is vacuously true if there is no x that is A. for example, the conditional “all unicorns are white” is vacuously true—but so is “all unicorns are not-white.”

6. Mark day, The Philosophy of History: An Introduction (london: continuum, 2008). day starts his discussion of scientific laws in history by referring to Isaiah Berlin’s famous rejection of the deductive-nomological model of explanation in history (Isaiah Berlin, “History and theory: the concept of Scientific History,” History and Theory 1, no. 1 [1960], 1-31).

7. day, The Philosophy of History, 60, 62-69.


But day and others set too high a standard for history. although the traditional criteria for lawfulness have dominated philosophy of science for decades, they have been questioned more and more since the 1980s and the 1990s. It became clear that if one stuck to the standard criteria of lawfulness, there would be few or no laws whatsoever present in science. this holds for chemistry,8 biology,9 psychology,10 and the social sciences.11 It even holds for physics, in spite of the paradigmatic role this discipline has played in philosophy of science.12 In particular, generalizations put forth in scientific practice are often fraught with exceptions, so they cannot be true universal conditionals. they often hold only in certain circumstances or under certain conditions, so they are neither universal nor necessary. likewise, they are often idealized representations of the world, so they can at best be vacuously true (since the objects to which they refer do not actually exist). they often refer to particular places and times (be it directly or only indirectly), so they fail to be general.13 It follows, and this is good news for historians, that history need not be worse off than other sciences in this respect.

It has become clear that scientific research rarely if ever results in generaliza-tions that satisfy the traditional criteria for lawfulness. this is problematic for two reasons. first, many generalizations have received the honorific status of “law of nature” even if they don’t satisfy these criteria. examples are Newton’s laws of motion, Mendel’s laws of inheritance, and gresham’s law in economics.14 Why would philosophers of science defend a set of criteria that denies these gener-alizations, and perhaps any existing piece of knowledge, the honorific status of “law”? Second, laws of nature are traditionally deemed indispensable for predic-tion, explanation, the assessment of counterfactuals, and/or manipulation and intervention. If this is correct, we must give up the hope that scientific knowledge may serve prediction, explanation, and intervention unless we come up with a more reasonable account of laws of nature.15

8. Maureen christie, “philosophers versus chemists concerning ‘laws of Nature’,” Studies in the History and Philosophy of Science 25, no. 4 (1994), 613-629.

9. John Beatty, “the evolutionary contingency thesis,” in Concepts, Theories, and Rational-ity in the Biological Sciences, ed. gereon Wolters and James g. lennox (pittsburgh: University of pittsburgh press, 1995), 45-81; John Beatty, “Why do Biologists argue like they do?,” Philosophy of Science 64, no. 4 (1997) S432-S443; robert N. Brandon, “does Biology Have laws? the experi-mental evidence,” Philosophy of Science 64, no. 4 (1997), S444-S457; elliott Sober, “two Outbreaks of lawlessness in recent philosophy of Biology,” Philosophy of Science 64, no. 4 (1997) S458-S467.

10. Steven Horst, Laws, Mind, and Free Will (cambridge, Ma: MIt press, 2011).11. clive Beed and cara Beed, “Is the case for Social Science laws Strengthening?,” Journal for

the Theory of Social Behaviour 30, no. 2 (2000), 131-153; John t. roberts, “there are No laws of the Social Sciences,” in Contemporary Debates in Philosophy of Science, ed. christopher Hitchcock (Malden, Ma: Blackwell publishing, 2004), 151-167.

12. Nancy cartwright, How the Laws of Physics Lie (Oxford: clarendon press, 1983).13. these arguments can be found passim in the works cited in footnotes 8 to 12.14. See, for example, William Bateson, Mendel’s Principles of Heredity: A Defence (london:

cambridge University press, 1902) for the case of Mendel’s laws.15. Several authors have proposed a revised account of laws of nature. In this footnote, we will

briefly discuss the most prominent figures in the debate and indicate why we prefer to start from the works of Mitchell (and Woodward; see section III) to build an account of lawfulness in history and historiography. Of course, these alternative accounts all have shown their philosophical merits in con-texts other than the one we focus on here. (for an introductory overview of revised accounts of laws of nature, see Stathis psillos, Causation and Explanation [chesham, UK: acumen, 2002], chapter 7.)

In his book, Natural Laws in Scientific Practice (Oxford: Oxford University press, 2000), Marc


this is precisely what Mitchell does.16 Instead of rejecting the concept of law of nature, she sets out to refine it and proposes a pragmatic approach to the ques-

lange offers an intricate account of laws of nature: laws are generalizations that are inductively confirmed, where this inductive confirmation guarantees a special kind of stability. the set of laws is the unique set of truths that is nontrivially stable under all counterfactual suppositions that are consistent with each of its members. lange’s account is interesting and intriguing, but also very technical and more a piece of metaphysics than an account of how natural laws function in scientific practice (despite the title of his book). Hence we deem it less useful as an account of the use of laws in historiography.

Nancy cartwright claims that no laws, not even laws in physics, are universal and exceptionless (cf. footnote 12). they hold only ceteris paribus, in the context of nomological machines. Nomo-logical machines are “fixed (enough) arrangement[s] of components, or factors, with stable (enough) capacities that in the right sort of stable (enough) environment will, with repeated operation, give rise to the kind of regular behaviour that we represent in our scientific laws” (Nancy cartwright, The Dappled World [cambridge: cambridge University press, 1999], 50). Nomological machines are very similar to “mechanisms” (which we discuss below), but where mechanisms are deemed to occur in nature, nomological machines are usually engineered by humans (ibid., 49). In cartwright’s framework, it is “not laws that are basic, but capacities. It may be questioned, however, to what extent capacities can be coherently individuated independently of lawlike regularities (see psillos, Causation and Explanation, 192ff.)

according to peter Menzies, laws of nature are robust or resilient under actual and hypothetical experiments, whereas accidental generalizations are not (peter Menzies, “laws of Nature, Modality and Humean Supervenience,” in Ontology, Causality and Mind: Essays in Honour of D. M. Arm-strong, ed. John Bacon, Keith campbell, and lloyd reinhardt [cambridge, UK: cambridge Uni-versity press, 1993], 207). this view is similar to Woodward’s, but contrary to Woodward, Menzies closely ties the notion of lawhood to human agency. Moreover, contrary to Mitchell, he sticks to the idea that laws are exceptionless (cf. psillos, Causation and Explanation, 188). In our opinion, these features make his account less interesting than Woodward’s and Mitchell’s.

finally, in a recent book, Steven Horst also advocates an account of laws of nature that reacts against the empiricist notion of strict laws (Horst, Laws, Mind, and Free Will). focusing on the laws of psychology, and their apparent conflict with free will, he adopts a cognitive pluralist view according to which laws are “idealized claims that pick out potential partial causal contributions to real-world behavior” (ibid., 8). given that Horst’s laws are idealizations, and given that their ideal-ization conditions are essentially open-ended, the existence of (psychological) laws does not imply (psychological) determinism (ibid., 9).

there is also another possible solution to the problem described above that deserves to be men-tioned. Instead of trying to revise the notion of law, philosophers of science have also tried to develop lawless accounts of science (cartwright could be included here as well). for example, ronald giere advocates an understanding of scientific practice of which model, and in particular theoretical model, is the fundamental concept (ronald N. giere, Science without Laws [chicago: University of chicago press, 1999, 5]), and Bas van fraassen has promoted the notion of symmetry instead of laws (Bas van fraassen, Laws and Symmetry [New york: Oxford University press, 1989]). recently, the con-cepts of “mechanism” and “mechanistic explanation” have become very popular in philosophy of science. Some of the major proponents of the mechanistic literature have explicitly, and justifiably, presented their views in opposition to the empiricist notion of strict laws and deductive-nomological explanation (see, for example, peter Machamer, lindley darden, and carl f. craver, “thinking about Mechanisms,” Philosophy of Science 67, no. 1 [2000], 1–25; and William Bechtel and adele abraha-msen, “explanation: a Mechanist alternative,” Studies in History and Philosophy of Biological and Biomedical Sciences 36 [2005], 421-441). yet these accounts do not provide a real alternative to the more relaxed account of laws we adopt here (for a discussion, see Bert leuridan, “can Mechanisms really replace laws of Nature?,” Philosophy of Science 77, no. 3 [2010], 317-340; for a discussion of historical explanation from a mechanist perspective that, given the conclusions of the paper just cited, is fairly consistent with our present arguments, see Stuart S. glennan, “ephemeral Mechanisms and Historical explanation,” Erkenntnis 72 [2010], 251-266).

16. Sandra d. Mitchell, “pragmatic laws,” Philosophy of Science 64, no. 4 (1997), S468-S479; idem, “dimensions of Scientific law,” Philosophy of Science 67, no. 4 (2000), 242-265; idem, Biological Complexity and Integrative Pluralism (cambridge, UK: cambridge University press, 2003).


tion whether there are laws. She focuses mainly on lawhood in biology, but the concept of “pragmatic law” suits other scientific disciplines as well. as she puts her approach: “the pragmatic approach focuses on the role of laws in science, and queries biological generalizations to see whether and to what degree they function in that role.”17

What is the role of laws or generalizations in science? “the function of sci-entific generalizations is to provide reliable expectations of the occurrence of events and patterns of properties.”18 to the extent they do so, scientific gener-alizations can be used for, for example, prediction, explanation, and manipula-tion.19 yet given that these generalizations will seldom be completely universal (let alone necessary), we need to know in which contexts they hold and in which they do not. the important question, therefore, is not whether scientific generalizations are contingent—most certainly, they are; the truth of scientific generalizations is always contingent on certain conditions; the important ques-tion is how and to what extent they are contingent. this means that, if we want to use a generalization, we need to assess the stability of these conditions. In particular, what are the conditions upon which the regularity under study is contingent? How spatio-temporally stable are these conditions? and what is the relation between the regularity and its conditions (is it deterministic, probabi-listic, or . . .)? Stability is a very important ontological parameter for the evalu-ation of a generalization’s usefulness.

It is important to note that stability is a gradual parameter. all regularities are contingent in that they rest on certain conditions. the laws of physics are contingent on the way the universe originated,20 Mendel’s laws of heredity are contingent on the existence of sexually reproducing organisms, and so on. and all of these conditions are historically shaped and are to a greater or lesser extent spatio-temporally stable. thus it is not only the laws of physics that deserve to be called stable. the laws of physics, chemistry, biology, and so on all are to some extent contingent. Only if contingency is interpreted gradually, Mitchell claims, will our conceptual framework be rich enough to account for the diver-sity of types of regularities and generalizations and for the complexity found in the sciences.21 this continuum of contingency/stability replaces the traditional dichotomy of necessary laws versus accidental generalizations.

Mitchell also distinguishes another ontological parameter: strength. gener-alizations may be deterministic or probabilistic; they may allow for only one or instead for multiple possible outcomes starting from the same set of initial conditions; they may have few or no exceptions, or quite a lot of them.22 the

17. Mitchell, “pragmatic laws,” S469; original emphasis.18. Ibid., S477.19. typically, pragmatic laws are used for manipulation in the applied sciences (technology,

biotechnology, policy contexts, and so on).20. this is not to say that there are no important differences between physical and biological

generalizations; see Mitchell, “pragmatic laws,” S472–S473. 21. Ibid., S469–S477; Mitchell, “dimensions of Scientific law,” 250–259.22. Mitchell, “pragmatic laws,” S477. We will see that Woodward likewise allows for

nondeterministic invariant (and hence explanatory) generalizations. Note that deterministic laws are not a prerequisite for prediction. probabilistic laws allow for probabilistic predictions (or the prediction of probability distributions).


pragmatic account of scientific laws thus does not bring along a commitment to determinism.

finally, Mitchell also distinguishes several representational parameters, such as degree of accuracy, level of ontology, simplicity, and cognitive manageability. We will not elaborate on these as they aren’t relevant to our present purposes.

these ontological and representational parameters together form a multidimen-sional framework by means of which one can compare scientific generalizations; they also provide the means for evaluating the usefulness of such generaliza-tions.23 this evaluation is of course context-dependent. for example, whether a generalization (such as some mathematical formula) is accurate enough, or instead too accurate, depends on what we use it for. likewise, in order to use a generalization for the purpose of prediction, explanation, or manipulation in a certain context, we must make sure the conditions on which it is contingent hold in that context. In this paper, we will focus mainly on the notion of stabil-ity (especially in sections V to VII), since that parameter is the most relevant to historiography.

Before we go on to the next section, let us make some short terminological remarks. first, we will use “law” in the sense of a generalization. Hence laws are epistemic: they can be “true,” “apply to some context,” “be used for . . .,” and so on. laws describe regularities. regularities are ontic: they “hold,” or “exist.” Second, despite the fact that the label “pragmatic” applies primarily to laws, we will also apply it derivatively to models and theories. We will treat models and theories as coherent collections of laws plus background conditions. given that the laws in question are pragmatic, the models and theories will be pragmatic as well. they are contingent on certain conditions, fail to be necessary, may have exceptions, and the like.24 third, it may be objected that up till now no definition has been given of “pragmatic law.” this is true, but it is no oversight. Mitchell rejects any normative approach that starts from a definition. She prefers “to characterize . . . the variability of types of generalization within the empirical sciences,” and “to direct the entire discussion away from the question of what we should call a law towards an understanding of how scientific generalizations of various types function in inferences to satisfy the pragmatic goals of science.”25 the variability of types of generalization is evident from her multidimensional framework. generalizations that are used for explanation or for prediction or for intervention vary both along ontological (stability, strength) and representational (degree of abstraction, simplicity, . . .) lines. as she expresses it: “the complexity that is reflected in the diversity and plurality of claims in the sciences reflect[s]

23. Ibid., S477.24. Note that our characterization of models and theories sidesteps many philosophical nuances.

for example, models and theories are often distinguished from each other, where models are treated as mediators between abstract theories and the world (see Mary S. Morgan and Margaret Morrison, Models as Mediatiors: Perspectives on Natural and Social Science [cambridge, UK: cambridge Uni-versity press, 1999]). We also use “models” in the sense of representations, not in the sense of things represented. Where the first sense is most common in philosophy of science, the latter is used in logic and philosophy of mathematics (see Wolfgang Balzer, c. Ulises Moulines, and Joseph d. Sneed, An Architectonic for Science: The Structuralist Program [dordrecht: reidel, 1987], 2-6).

25. Mitchell, “pragmatic laws,” S469.


both ontological differences among causal structures in the domains studied and in other features of the representational medium.”26

III. caUSatION aNd explaNatION

among other things, pragmatic laws figure in explanation. there is growing consensus in philosophy of science that explanation requires causation. the old theories of explanation that attribute no distinctive role to causation in expla-nation (such as Hempel’s deductive-nomological model), or that make causal notions derivative from explanatory notions (such as Kitcher’s unificationist view27), have become outmoded. yet generalizations may have stability and strength without being causal (for example, the generalization relating the read-ing of a barometer and the occurrence of storms, discussed below). therefore, we need an account of causation that allows us to distinguish causal pragmatic laws from noncausal ones. among the causal accounts of explanation, Woodward’s manipulationist theory of causation,28 which characterizes causation as “invari-ance under interventions,” best fits Mitchell’s pragmatic account of lawfulness;29 moreover, it dovetails nicely with explanation and causal discovery in the special sciences (such as the biological, the social, and the behavioral sciences).30 Hence we are convinced that historians and philosophers of history may benefit from Woodward’s theory.31

26. Mitchell, “dimensions of Scientific law,” 259.27. philip Kitcher, “explanatory Unification and the causal Structure of the World,” in Scientific

Explanation, ed. philip Kitcher and Wesley c. Salmon, Minnesota Studies in the philosophy of Sci-ence, vol. xIII (Minneapolis: University of Minnesota press, 1989).

28. It is interesting to note that edgar Zilsel, the austrian mathematician, philosopher, historian, and sociologist of science, who explicitly endeavored to find historical laws, made use of the concept of manipulation to distinguish between causes and effects in a way that bears some resemblance to Woodward’s account. See elisabeth Nemeth, “edgar Zilsel on Historical laws,” in Explanation, Prediction, and Confirmation, ed. d. dieks, W. J. gonzalez, S. Hartmann, t. Uebel, and M. Weber. Vol. 2 of the series The Philosophy of Science in a European Perspective (Basel: Springer, 2011), 521-533, for a discussion of Zilsel’s views on historical laws.

29. Jim Woodward, Making Things Happen: A Theory of Causal Explanation (New york: Oxford University press, 2003); idem, “explanation and Invariance in the Special Sciences,” British Journal for the Philosophy of Science 51 (2000), 197-254. for a more elaborate combination of the concepts of “invariance” and “pragmatic law,” see leuridan, “can Mechanisms really replace laws of Nature?”

30. Woodward’s account of causation is closely connected to statistical techniques such as struc-tural equation modeling (see Making Things Happen, chapter 7), and the use of directed acyclic graphs for causal discovery (see Judea pearl, Causality. Models, Reasoning and Inference [cam-bridge, UK: cambridge University press, 2000], and peter Spirtes, clark glymour, and richard Scheines, Causation, Prediction and Search [cambridge, Ma: MIt press, 2000]). pearl’s framework has been incorporated in the potential outcome model for observational data analysis, which in turn may be of interest to statistically oriented historians (see Stephen l. Morgan and christopher Winship, Counterfactuals and Causal Inference: Methods and Principles for Social Research [cam-bridge, UK: cambridge University press, 2007], and erik Weber and Bert leuridan, “counterfactual causality, empirical research, and the role of theory in the Social Sciences,” Historical Methods 41, no. 4 [2008], 197-201 for a review of this book).

31. Karsten r. Stueber refers briefly to Woodward’s theory in his discussion of the role of folk-psychological generalizations in historical narratives. See Karsten r. Stueber, “reasons, generaliza-tions, empathy, and Narratives: the epistemic Structure of action explanation,” History and Theory 47, no. 1 (2008), 31-43.


Woodward’s main question is whether there are explanatory generalizations in the special sciences. He answers affirmatively. for Woodward,

explanatory relationships are relationships that in principle can be used for manipula-tion and control in the sense that they tell us how certain (explanandum) variables would change if other (explanans) variables were to be changed or manipulated. the qualifica-tion “in principle” means that what matters for the purposes of explanation is not whether the manipulation in question can actually be carried out but rather whether the putative explanatory relationship correctly describes what would happen on the (possibly counter-factual) supposition that the manipulation is carried out.32

explanatory generalizations relate (explanandum) variables to other (explanans) variables. a variable (also known as a random variable) is a quantity or magni-tude that can take more than one value. for example, temperature can take any value higher than -273° celsius; sex can take the values “male” and “female.” to be explanatory, such a generalization need not satisfy the traditional criteria for lawfulness. rather it has to be a causal generalization or, in Woodward’s termi-nology, a generalization that is invariant under interventions.

Heuristically, one may think of an intervention as an idealization of an experimental manipulation carried out on some variable x for the purpose of ascertaining whether changes in x are causally or nomologically related to changes in some other variable y. . . . the idea we want to capture is roughly this: an intervention on some variable x with respect to some second variable y is a causal process that changes x in an appropriately exogenous way, so that if a change in y occurs, it occurs only in virtue of the change in x and not as a result of some other set of causal factors.33

the phrase “in an appropriately exogenous way” is intended to rule out cases where the intervention causally influences y via a route that does not go through x, or cases where the intervention is correlated with any of y’s other causes.34

let us give two well-known examples to illuminate the notion of invariance.35 Hooke’s law, F = -Ksx, describes the relation between the extension x of a par-ticular spring s (with spring constant Ks) and the restoring force F it exerts. this

It should be noted that Woodward’s manipulationist theory is not the only theory of causation that fits history and historiography; see anton froeyman, “concepts of causation in Historiography,” Historical Methods 42, no. 3 (2009), 116-128 for a defense of causal pluralism in historiography. But it is the one that best suits our present purposes, namely, to address the problem of lawfulness in history by means of Mitchell’s account of pragmatic laws.

32. Woodward, “explanation and Invariance in the Special Sciences,” 198; original emphasis. a strongly similar view is expressed by carl Hammer, “explication, explanation, and History,” History and Theory 47, no. 2 (2008), 183-199. yet Hammer presents his ideas as if they were new. He does not refer to Woodward’s theory of invariance nor to any other manipulability account of causation (such as von Wright’s, gasking’s, collingwood’s, or Menzies and price’s; for references, see Making Things Happen). the following quote by Hammer shows that his views are parallel to Woodward’s: “What we are after in giving such explanations is, among other things perhaps, control, where con-trol is taken to be the possession of knowledge that could be used to practically manipulate elements of the world in order to bring about other elements. It is a kind of intellectual power. Note that it is important that the power of control is intellectual and only practical in a hypothetical sense” (Ham-mer, “explication, explanation, and History,” 188; original emphasis).

33. Woodward, “explanation and Invariance in the Special Sciences,” 199-200.34. this is spelled out technically in the notion of an intervention variable; see ibid., 201, and

Woodward, Making Things Happen, 95-99.35. See Woodward, Making Things Happen, 269 for Hooke’s law, and 14-15 for the barometer

example.


generalization is invariant under some range of interventions that change the extension (such as stretching the spring) in that it correctly describes what the restoring force would be under these interventions. (It may fail to be invariant under extreme interventions where the spring is extended such that it breaks. that is not a problem. Invariance is not an all-or-nothing feature of generalizations. What it takes for a generalization to be invariant is that it holds under some speci-fiably wide range of interventions, not under all possible interventions.) consider, by contrast, the following example. a barometer’s reading (B) is strongly cor-related with the occurrence of storms (S).36 this correlation may well serve as a pragmatic law. It is relatively precise (strong) and the causal relations from which it derives (with atmospheric pressure as a common cause; see below) are based on conditions that are relatively stable. as a result, we may use the correlation to predict the occurrence of a storm from the barometer’s reading. But the correla-tion between B and S is not itself causal. Suppose that we change the barometer’s reading (and nothing else) based on the outcome of a coin flip (if heads, we switch it to the left; otherwise we switch it to the right). this would count as an appropriately exogenous change to B. yet the relation between B and S would fail to hold under any such intervention. By manipulating the barometer’s reading we cannot induce or prevent a storm. the correlation between B and S is not invariant under interventions on B (nor under interventions on S); hence it is not a causal relation. the correlation that does obtain is the result of a common cause, to wit atmospheric pressure.

Woodward’s theory of invariance fits Mitchell’s pragmatic account of laws for several reasons. first, like stability and strength, invariance is a gradual concept. Some causal generalizations are more invariant than others (they are robust under a larger range of interventions). Second, a generalization can be invariant even if it is contingent on certain background conditions that have only limited stability. thus pragmatic laws can be invariant, and as invariant they can be used in expla-nations. third, to be invariant a generalization need not be deterministic. Statisti-cal generalizations may be invariant as well. for example, smoking causes lung cancer (if we were to intervene in the behavior of a population so that its members smoked less, we would expect lung cancer incidence to drop in the following years), even though not every smoker is doomed to get lung cancer. Hence, like the pragmatic account of scientific laws, Woodward’s manipulationist account of causation is not wedded to determinism.

together, Mitchell’s and Woodward’s frameworks give us just what we need to account for laws in the special sciences. Mitchell provides a multidimensional framework to compare scientific generalizations. Woodward provides the basis for a distinction between causal and noncausal generalizations that nicely fits actual scientific practice in a wide array of disciplines.37

36. let B be a random variable taking the following values: “stormy,” “rain,” “change,” “fair” and “very dry.” let S be a binary random variable taking as values “storm” and “no storm.”

37. In a recent paper that is very close to Mitchell’s multidimensional approach, Woodward dis-cusses a number of parameters or characteristics of causal generalizations that he claims are important in biological contexts: stability, proportionality, and specificity. We take it that these characteristics are important in other disciplines as well. See Jim Woodward, “causation in Biology: Stability, Specificity, and the choice of levels of explanation,” Biology and Philosophy 25 (2010), 287-318.


IV. pragMatIc laWS IN HIStOry

the idea that laws of nature play a role in the historical sciences has regained influence. consider the following remark by Stephan Berry: “during the second half of the twentieth century, historiography proper and neighboring disciplines, such as archaeology, anthropology, or political science, gave rise to new fields of inquiry, where the concept of laws is involved at least implicitly. However, the lessons have been learned, and most scholars today are cautious: you may make generalizations about history—but avoid the ‘l-word!’”38

Berry’s account is interesting in that it acknowledges the role of causal gener-alizations in at least certain areas of history even as he admits that the notion of law is something historians rightfully eschew. However, as is evident from one of his earlier papers, he holds too much to the traditional view of laws of nature. laws of nature (as opposed to mere trends) in his view have to be time-invariant in the sense that they act uniformly over time.39 another difficulty is that he does not address what we think are the most interesting aspects of this debate, to wit, the methodological problems that go with the use of nonstrict laws in history.

another very recent attempt at rehabilitating the concept of “general law” in history can be found in david christian’s plea for Universal History.40 But again, his views differ from ours. christian restricts his attention only to three large-scale law-like patterns (relating to increasing human control over resources, increasing number of human beings, and increasing complexity of human societ-ies), whereas our discussion in principle includes any useful causal generaliza-tion from any scientific discipline (see below). and, like Berry, christian is not concerned with the methodological problems that go with the use of nonstrict laws in history.

the claim of this paper is that the debate on the use of laws in history can be made much more clear and fruitful by applying Mitchell’s concept of “pragmatic law,” supplemented by Woodward’s notion of invariance under intervention, to history and historiography. to be able to do this, we have to answer the questions, first, where do the pragmatic laws used in history derive from? and second, what do prediction, manipulation, and explanation in history amount to?

the generalizations used in history may derive from many different sciences. Historians investigate the actions of individual humans, societies, economic sys-tems, and a host of other entities. for this they may wish to rely on other scien-tific disciplines such as psychology, economics, and sociology, but also biology, climatology, demography, chemistry and physics, and so on. these disciplines all generate generalizations that can “provide reliable expectations of the occurrence

38. Stephan Berry, “the laws of History,” in A Companion to the Philosophy of History and His-toriography, ed. aviezer tucker (Malden, Ma: Blackwell publishing, 2009), 168, original emphasis.

39. Stephan Berry, “On the problem of laws in Nature and History: a comparison,” History and Theory 38, 4, Theme Issue 38 (1999), 129.

40. david christian, “the return of Universal History,” History and Theory 49, no. 4, Theme Issue 49 (december 2010), 6-27.


of events and patterns of properties” (see above) yet are contingent on conditions that are more or less stable. the less stability they exhibit, the more careful the historian must be to verify that they hold in the context under investigation.

the generalizations that are formulated in these sciences rarely are determinis-tic. as we saw in the previous sections, pragmatic laws and invariant pragmatic laws may be statistical generalizations. this is important in view of the long-standing debates about determinism in history. Our claim that pragmatic laws can be used in the study of history (especially if they are invariant under some range of interventions) does not at all imply that history is deterministic, nor is it committed to any form of historicism.41

let us now turn to the use of these laws in history. Strictly speaking, prediction is not part of historiographical practice. Still, historians may engage in retrodic-tion or postdiction. from a set of sources indicating a past course of events, together with a theoretical framework that incorporates one or more pragmatic laws, another past event may be “predicted.” this prediction is then to be con-firmed by independent sources (recall from note 22 that generalizations need not be deterministic to serve in prediction).42

What about explanation? In the case of historiography, we can distinguish between two aspects of explanation, which relate to different kinds of explanan-da. these aspects do not refer to different historiographical methods, but are, each at their own level, present in every historiographical explanation. first, one may try to explain the occurrence of historical events. these may be singular events (for example, why there was a major crash on Wall Street on October 24, 1929). Or they may be general events or regularities (for example, why were all success-ful revolutions in early modern europe supported by at least a part of the social elite?43). to explain both singular and general historical events, historians may invoke invariant pragmatic laws.44

the second aspect of historical explanation concerns not the historical events themselves, but the historical sources. these sources are the data historians have to deal with, just as measurement results are the data of physicists. a body of his-torical sources, such as “plots” and the outlook of playhouses in Shakespearean theater (see section V), is explained if historians can offer a coherent, accurate, and preferably simple account about why the sources are what they are. Histori-ans do this by constructing an image of historical reality. If historians want their image of historical reality to be coherent, accurate, and not overly complicated (which they should if they want to give an explanation), they have to use certain pragmatic laws.

Both in the case of the explanation of historical events and in the case of the explanation of historical sources, the key issue is whether these causal pragmatic laws hold in the historical context in question. as we have seen, this is connected

41. cf. Berry, “On the problem of laws in Nature and History,” especially 124-125.42. Berry calls this “reconstruction” and stresses that it provides a useful possible role for laws in

the historical disciplines (Berry, “the laws of History,” 168).43. See Preconditions of Revolution in Early Modern Europe, ed. Jack greene and robert forster

(ann arbor, MI: UMI, 1996).44. for Woodward’s account of singular-causal explanation, see Making Things Happen, 209-220.


to the notion of the stability of a causal pragmatic law. In many cases, historio-graphical discussions revolve around this notion.

a historical debate about the stability of causal pragmatic laws can take differ-ent shapes. first, a historian can argue that a certain causal law that is commonly used holds only on certain conditions that were not present in the period of inter-est, and can offer an alternative that is contingent on conditions that were present in that period (see section V). Second, a historian might argue that a given law is contingent on fewer conditions than is often thought, and that contrary to the received view it does apply to the period of interest (see section VI). third, and this is one of the more rare cases, a historian can argue that a certain causal law holds only on certain conditions that were not present in the period of interest even though this historian does not provide an alternative. from this it would fol-low that in this type of case not using a certain kind of generality at all provides a simpler and more coherent and accurate image of historical reality, and therefore a more coherent and accurate explanation. this might seem counterintuitive, but in the seventh section of this paper we will show how this is possible.

In the rest of this paper we will analyze three very different examples of his-torical debates in order to show that the stability of causal pragmatic laws is a key issue in historiography. We will present them in the order of the different cases we have just described. examples of the first two kinds are relatively easy to find, but those of the last are not.

But first, let us tackle one more problem. Our examples should help to alleviate an important set of possible worries regarding the use of causal pragmatic laws in history.45 first, it seems as if the notion of general law has become so watered down that in the end all we say is that historians sometimes argue over the rel-evance of some causal generalization or the existence of some regularity, a mes-sage that is hardly surprising. In reply we claim that this is not the sole message of the paper. We also show how historians do (and should) argue over generaliza-tions and regularities, namely, by focusing on their conditions of stability (as our examples will illustrate). a second worry is that some generalizations (causal or otherwise) may hold only in a very narrow historical period or social or cultural setting. does this make them ill-suited for the label “causal pragmatic law”? this worry correctly highlights the fact that there is no context-independent thresh-old for pragmatic lawfulness in Mitchell’s framework.46 But we consider this a strength rather than a weakness of her account. Whether a generalization should count as a pragmatic law in part depends on the questions at stake. a generaliza-tion’s holding only in a limited domain does not preclude it from being useful in that domain. a final possible worry is that the generalizations historians use are often stated in such broad terms that they become virtually vacuous and hence useless (generalizations of the genre “people tend to do what they prefer”). In reply, we invite readers to examine the examples we shall now provide. they ought to notice that the generalizations used or discussed by evelyn tribble, Wil-liam Harris, and lucien febvre are far from vacuous or useless.

45. We would like to thank an anonymous referee for pressing us on this issue.46. there is a threshold for invariance, however. for a generalization to be invariant, it should at

least be invariant under one intervention (see Woodward, Making Things Happen, 243).


V. cOgNItION IN tHe glOBe

Our first example is from a recent paper by evelyn tribble on the process of cognition in theatrical practice in sixteenth- and seventeenth-century england.47 She criticizes the traditional (nonhistorical) models of theatrical practice. In tribble’s view, these are overly individualistic and hence do not apply to Shakespearean theatrical practice: “I will argue that our understanding of the [elizabethan] playing system, particularly of the mnemonic demands that the repertory system made on its participants, has been consistently distorted by a tendency to view cognition as individual rather than social, which has caused us to imagine the workings of complex group structures in mechanistic terms.”48 for example, when she turns to “the neglected question of how actors manage to memorize their parts,”49 she cites a clinical study by tony and Helga Noice.50 this study comprises a large-scale qualitative study of modern actors’ memoriz-ing techniques and a “general model of acting cognition,” “which includes such processes as breaking down the scenes into ‘beats,’ determining a character’s overall goal and its articulation in each ‘beat,’ and then becoming the ‘character in his or her imagination.’”51 tribble does not question the Noices’ work itself, calling it ambitious and stating that it “breaks new ground in both its methodol-ogy and its results.”52 arguably, the Noices’ general model of acting cognition is descriptively adequate regarding present-day theater and it helps to explain how contemporary actors memorize text verbatim. But tribble does throw doubt on the use of this model in the context of theatrical practice in Shakespeare’s time. She does so by listing the conditions upon which it rests and by showing that not all of them were present in Shakespeare’s time:

While this [the Noices’] study breaks new ground in both its methodology and its results, it can tell us very little about how actors in Shakespeare’s time may have approached their roles. this is not because the biological basis of memory has changed, but because the mnemonic structures informing the practice of acting are culturally and historically defined. It does not take much reflection to realize that this “general model of acting cog-nition” is bound very specifically to late-twentieth-century acting practices, which are in turn based on assumptions about character and subtext derived from modern acting theory. Moreover, these practices are the results of institutional conditions such as long rehearsal periods, a relative scarcity of new plays, and, finally, the exigencies of memorizing prose rather than verse. None of these factors can have informed the mnemonic techniques of Shakespeare’s actors. . . .53

47. evelyn tribble, “distributing cognition in the globe,” Shakespeare Quarterly 56, no. 2 (2005), 135-155.

48. Ibid., 136.49. Ibid., 147-148.50. tony Noice and Helga Noice, The Nature of Expertise in Professional Acting: a Cognitive

View (Mahwah, NJ: erlbaum, 1997).51. tribble, “distributing cognition in the globe,” 148.52. Ibid.53. Ibid.


although the biological conditions upon which the general model is contingent are relatively stable and presumably did obtain in Shakespeare’s time, the institu-tional conditions are much more fragile and were provably absent. Instead of hav-ing the required scarcity of new plays and the long rehearsal periods, for example, “[c]ompanies performed a staggering number of plays: six different plays a week, with relatively infrequent repetition and with the additional demands of putting on a new play roughly every fortnight.”54 Moreover, actors did not possess com-plete copies of the play. they were given only “sides” containing their own parts and cues.55 as a result, the Noices’ general model of acting cognition cannot be used to explain how Shakespearean actors managed to memorize their parts.

as an alternative, tribble proposes to draw upon the field of “situated” or “dis-tributed” cognition,56 which comprises disciplines such as cognitive philosophy, education, sociology, aI, and cognitive anthropology.57 according to “distributed cognition,” cognition is not confined to individual cognizers, but distributed over those cognizers and their (cognitively rich) environment. In early modern theater, tribble claims, several tools, artifacts, and practices formed “elements of a cognitive structure” that enabled “an extraordinary level of achievement.”58 What were these tools, artifacts, and practices that made up this cognitively rich environment?59 first, there is the playhouse with its stage doors and playing platform. these, in tandem with a “tacit, invisible, and profound understanding of the stage,” may have helped to orient mental activity and to reduce cognitive demands.60 Second, actors relied heavily on “plots.” “these folio-sized sheets of paper contain scene-by-scene accounts of entrances and, sometimes, exits; necessary properties; casting; and sound and music cues.”61 In tribble’s view, plots functioned as “computational devices.”62 third, the social environment, including the structure of theatrical companies and the apprentice system, played an important cognitive role as well.63 the apprentice system allowed trainee boy actors to play progressively more difficult roles, and it “was also very efficient, because the task of training a boy could be subsumed into the task of memorizing one’s own lines.”64

By referring to these elements and by showing how they form a cognitively rich environment, tribble explains how early modern actors (and in particu-lar young boy actors) succeeded in memorizing plenty of roles without much rehearsal time at all.65 the following can fairly be interpreted as the causal gener-alization that licenses this explanation: “the more thinking that can be off-loaded

54. Ibid., 135-136.55. Ibid., 148.56. tribble’s main source of inspiration is Hutchins’s study of maritime navigation in Cognition

in the Wild (cambridge, Ma, and london: MIt press, 1995).57. tribble, “distributing cognition in the globe,” 139.58. Ibid., 142.59. tribble also discusses the role of the verse (ibid., 149-150) and of parts or “sides” (151-153).60. Ibid., 143-145.61. Ibid., 145.62. Ibid., 148.63. Ibid., 154-155.64. Ibid., 155.65. throughout her paper, tribble also repeatedly shows that her model better explains the sources

than do the alternative, more individualistic accounts of cognition (cf. section IV above).


onto the environment, the more mental energy remains available for those tasks that are primarily internal (memory for the spoken lines, for instance).”66 thus, if we were to change the “amount of thinking that can be off-loaded onto the environment” in an appropriately exogenous way (for example, by changing the use of plots, or by changing the apprentice system), then a change would occur in “the mental energy available for tasks that are primarily internal.” therefore, this generalization is invariant under some range of interventions, and hence causal and explanatory. Its use for explanation does not require that it be a universal, exceptionless, and necessary generalization. It is only required that it holds in the context of interest.

this example indicates that the notion of “pragmatic law,” coupled with an account of causation as “invariance under interventions,” is the right way to understand the explanatory tack taken by tribble; it also suggests the way to tackle the problem of lawfulness in history and historiography.

VI. tHe NatUre Of rOMaN MONey

Our second example concerns the financial system of the roman empire as discussed by William Harris.67 the basic explanandum is the fact that wealthy romans often made seven- or eight-figure payments, despite the fact that one million sesterces must have weighed about a ton. Secondary explananda include, among others, a decrease of the stock of silver coinage in the late republic (after about 79 bce), apparently without any major decline in economic activity, and references in existing sources to different kinds of loans and debts.68 On this basis, Harris argues that the consensus view that all roman money consisted of coins is mistaken, since it cannot give a coherent explanation of the phenom-ena described above. Instead, he offers a different model, one that pictures the romans having money closer to that of our modern economical and financial system. according to this model, the large majority of roman money is, just as in most modern nations, so-called IOU money, “virtual” money that is created by an extended system of loans and that adds to the money supply.69 So Harris’s theory consists primarily of the following two generalizations:

(1) an increase/decrease in the number of loans and/or credits leads to an increase/decrease of IOU money.

(2) an increase/decrease in IOU money leads to an increase/decrease in money supply.

Obviously, generalization (1) is to be understood as causal or invariant. Suppose one were to intervene so as to change the number of loans or credits in a properly exogenous way. then one would expect a corresponding increase/decrease in the

66. tribble, “distributing cognition in the globe,” 144.67. William Harris, “the Nature of roman Money,” in The Monetary Systems of the Greeks and

Romans, ed. William Harris (Oxford: Oxford University press, 2008), 174.68. Ibid., 194-197.69. “IOU” stands for “I owe you.” that IOU money adds to the money supply is known as the

“multiplier effect” (ibid., 178-179).


amount of IOU money. the same holds for generalization (2). In short, it seems that Harris appropriately cites the roman system of loans and credits in order to causally explain the increase in money supply.70

yet there is a problem. the roman empire differed from our modern society in that there were no central banks, clearing banks, or other central financial institu-tions that are the source of IOU money in our modern society. therefore, it is not prima facie obvious that Harris’s model can be applied to the roman empire. Harris answers this objection in the following way, one that indicates the general point he wants to make:

When economists define credit-money, they sometimes, admittedly, make matters more complicated than I have made them in this account, but that is because they quite natu-rally have recent and current conditions in mind, and not the world that existed before the invention of clearing banks. “a credit money system presupposes the existence of the institutions of private property, contracts, enforcement, and clearing,” says one.71 But historically speaking . . . the last of these four elements is a wonderful convenience but not in fact a necessity.72

Harris is explicit about the conditions on which generalization (1) is contingent. there is plenty of evidence of lending in the roman economy, but not all lending adds to the money supply. What is needed is the “capacity of the lender to recover from the debtor;”73 this presupposes private property, contracts, and enforcement, but not clearing (as is traditionally assumed). the way Harris substantiates this nicely illustrates how historians may reason about the conditions and stability of pragmatic laws. although sources relating to the roman monetary system are rather scarce, there is plenty of evidence about early modern england, where there were large amounts of virtual money but no central bank or clearing hous-es.74 this shows that it is possible for a pre-industrial economy to increase its money supply without extra coinage.75

at this point, both the received view (where the total money supply is physi-cally present as coins) and Harris’s view (where the majority of the money con-sists of IOU money) are, in principle, equally possible. the historical sources, the eight-figure payments and the data about the weight of roman coins, then serve as evidence in favor of Harris’s view. In short, the comparison with early modern england shows that Harris’s view is possible, and the sources from the roman era show that it is plausible.

the structure of Harris’s argument is basically the opposite of tribble’s. as we have seen, tribble argues that our modern-day models of theatrical practice depend on certain contingent background conditions (such as long rehearsal peri-ods, scarcity of plays, the availability to every actor of a complete copy of the play, and so on), and that, because of this, the pragmatic laws in question are not

70. although Woodward’s concept of causation is not in general transitive, the causal relations described by generalizations (1) and (2) assumably are.

71. Harris quotes Basil J. Moore, Horizontalists and Verticalists: The Macroeconomic of Credit Money (cambridge, UK: cambridge University press, 1988), 20.

72. Harris, “the Nature of roman Money,” 181; our emphasis.73. Ibid., 180.74. Ibid., 182-183 and 202.75. Ibid., 183.


applicable to the Shakespearean era. as an alternative, she offers a model of cog-nition that is supposed to depend on background conditions that were present in Shakespeare’s time. Harris, on the other hand, argues that a specific set of invari-ant pragmatic laws, which are part of the modern-day model of financial traffic, is contingent on fewer conditions than is typically thought, and hence is more stable than is often thought. In particular, it does not depend on the existence of large banks or financial institutions, but can also arise under conditions such as those present in early modern england and in the roman empire.

VII. atHeISM IN tHe SIxteeNtH ceNtUry

Our third example is again of a different kind. It concerns one of the great classics of twentieth-century historiography: lucien febvre’s Le Problème de l’incroyance au XVIe siècle: La religion de Rabelais. With this book, febvre responded to a statement by abel lefranc made in his Introduction to an edition of the collected works of rabelais. lefranc saw rabelais as a radical rationalist unbeliever, one of the first of a long tradition that culminated in the radical unbe-lief movement of the french enlightenment. febvre reacted against this view, not by arguing that rabelais was not an unbeliever as such, but by stating that the concept of “unbeliever” was simply not applicable to people of the sixteenth century.

at first sight it may seem as though this has nothing to do with laws or law-fulness, nor with causation or invariance under interventions, but only with con-cepts. Nevertheless, concepts are closely linked to pragmatic lawfulness. they denote properties that figure in causal and noncausal regularities; hence, they themselves figure in the causal and noncausal pragmatic laws describing those regularities. In the case of historiography, this amounts to the following. If some property possessed by historical subjects is not in any systematic way linked to other properties, be it deterministically or stochastically, and hence if it does not figure in any (causal) regularities, then it is no use introducing a concept to pick out that property. Why? Because to employ such a concept entails that the sub-ject’s possession of the relevant property cannot be used as an explanans or as an explanandum. In other words, historiographical concepts, such as “bourgeois” or “liegeman,” are meaningful only if they are linked with a (cluster of) pragmatic law(s).

It is precisely this issue that triggered febvre’s reaction against lefranc. the absence of certain regularities, of certain patterns of behavior in the sixteenth century, grounds febvre’s claim that the concept of “unbeliever” cannot properly be applied to that period. Whereas lefranc applied a cluster of pragmatic laws involving the concept of “unbeliever” to the sixteenth century, febvre argues that there are specific conditions under which that cluster applies, and that these conditions were not present in that period. In this respect, febvre’s argument runs parallel to our first example. But the difference is that he does not really give an alternative conception of “unbeliever.” Instead, he advocates a view that refutes the use of “unbeliever” as such to describe the sixteenth century (and, conse-


quently, the category of “believer” as well).76 So, as in the example of tribble, febvre refutes the use of certain pragmatic laws in some specific context, but he does not deem it necessary to give an alternative. to know that the laws in question do not apply in itself already adds to our knowledge. febvre explicitly refers to this as the sole aim of his book: “to believe or not to believe. the naive, simplistic notion that the problem is without mystery, the antihistorical notion that we can deal with it with regard to the men of the sixteenth century in the same way we tend to deal with it with regard to ourselves—this illusion and these anachronisms are what this book has been directed against.”77

let us have a closer look at febvre’s arguments. as we will show, they appeal both to the notion of stability and to the explanatory use of pragmatic laws. first, he gives a rather strict definition of “unbelief.” to be rightly called an “unbe-liever,” it does not suffice just “not to believe.” Instead, one must display certain patterns of behavior:

“Not to believe” is a phrase that is inadequate. What concerns us at the moment is not some kind of abstract belief, the attitude of a man who does not believe in the existence of a god endowed with certain attributes and granted certain epithets. . . . What primarily concerns us is the attitude of a man who, born a christian and totally involved in chris-tianity, could extricate his mind and shake off the common yoke—the yoke of a religion unfalteringly and unreservedly professed by nearly every one of his contemporaries.78

to be called an unbeliever for febvre is “to break with the habits, customs, even the laws of the social groups of which he [is] a part, at a time when those habits, customs, and laws were still in full force,” and there being “material in his knowl-edge and the knowledge of the men of his time either for forming valid doubts or for supporting those doubts with proofs that, on the basis of experimentation, could have the force of real, veritable conviction.”79 these are the causal regu-larities and patterns of behavior and thought febvre has in mind. yet this type of unbelief did not exist in the sixteenth century: “It is absurd and puerile, therefore, to think that the unbelief of men in the sixteenth century, insofar as it was a real-ity, was in any way comparable to our own. It is absurd, and it is anachronistic.”80

Indeed, febvre argues that none of the conditions of “unbelief” were present. None of the possible reasons for being an atheist, of which he deems the historical

76. Of course, this does not mean that there were no pious people or people who did not believe in any god at all in the sixteenth century. In his round-up of the literature following febvre, david Wootton shows that there were undoubtedly people who did not believe in god in that century. feb-vre’s point is not that everybody was religious, but that the specific traits or patterns of behavior that we would now regard as signs of “unbelief” were not generally spread, nor explicit, nor determined in a specifically fixed way, and not as densely grouped together as they are today (david Wootton, “lucien febvre and the problem of Unbelief in the early Modern period,” Journal of Modern History 60, no. 4 [1988], 695-730; see also david Wootton, Paolo Sarpi: Between Renaissance and Enlight-ment [cambridge, UK: cambridge University press, 2002]). In any case, whether febvre is correct or not does not matter with respect to the point we are making. What is important is that for decades febvre’s statement has been regarded as an important contribution to the body of historical knowl-edge, and that this contribution depended on the status of certain pragmatic laws.

77. lucien febvre, The Problem of Unbelief in the Sixteenth Century: the Religion of Rabelais, transl. Beatrice gottlieb [1947] (cambridge, Ma: Harvard University press, 1985), 455.

78. Ibid., 455-456.79. for both quotes, see ibid., 455.80. Ibid., 460.


and scientific reasons to be the most important, were present. Historical reasons were absent, since textual criticism was not yet applied to the gospels. Scientific reasons were absent, since the concept of natural law, which febvre deems neces-sary to replace the actions of an interventionist god, had not yet been developed in the sixteenth century.81

the statement that there was no unbelief in the sixteenth century provides a better and more coherent explanation of the sources, particularly those relating to the apparently inconsistent use of blasphemous and pious formulations by one and the same author. therefore, febvre’s argument qualifies as a historical explanation. If one thinks that a person such as françois rabelais, who often used blasphemous formulations but not explicit statements of unbelief, either was an unbeliever or a believer, one has to assume he had a hidden agenda. either he used this blasphemy for a skeptical or ironic agenda, or he hid his true nature as an unbeliever and tried to make this clear by indirect hints and arguments. again, this is the position against which febvre argues.

they are too casually assumed to have swung at will from aggressive unbelief to the most traditional kind of belief. can it be that the problems about their view that we have declared to be insoluble have been brought into being by ourselves, and by us alone? do we not substitute our thought for theirs, and give the words they used meanings that were not in their minds? thus the problem that was formulated poorly can become a better formulated one.82

the assumption that the concept of unbeliever does not apply to the people of the sixteenth century, at least not in a way similar to how it does now, clears the way for us to read the sources more in their own right, without having to presuppose any hidden agendas. In this way the fact that some concept, and consequentially some cluster of pragmatic laws, does not apply in a certain context can in itself, without an alternative, already serve as an explanation. thus, that febvre does not offer an alternative cluster of pragmatic laws should not be seen as a lack in his argument; on the contrary, it this case it is a strength of his account.

VIII. cONclUSION

the aim of this paper has been to show that, contrary to what is often believed, general laws do play an important role in writing history. the logical empiricist view of natural laws, on the basis of which it has been denied that there are general laws in history, is untenable. the alternative—Mitchell’s concept of pragmatic law, in tandem with Woodward’s theory of causation as invariance under interventions)—offers a view of lawfulness that is much more realistic with respect to the use of laws in different scientific domains, including historiogra-phy. fortunately, too, neither the pragmatic account of laws, nor the invariance account of causal generalizations, brings along with it the bogey of determinism, a bogey that has often served as the basis for a rejection of the idea that laws obtain in history.

81. Ibid., 455-460.82. Ibid., 11.


the notion on which we have primarily focused is stability. Historiographical discussions are often centered around the question of the stability of certain prag-matic laws. We have discerned several different ways in which this can be done. first, a historian can offer a generalization (or model, or theory) as an alternative to an existing one. In this case, he or she should state that the respective general-izations are contingent on different conditions and that the conditions of the first were present in the target context whereas those of the second are not. We have illustrated this with tribble’s treatment of models of acting practice in modern and in Shakespearean theater. this is probably the most widespread kind of his-toriographical argumentation. Second, a historian may argue that some theory is more stable than is generally believed, and that it does apply in the context of interest (contra the consensus view that it does not). We illustrated this by means of Harris’s discussion of the roman financial system. third, the fact that a certain generalization does not hold in a certain historical period can itself already be seen as a valuable contribution to knowledge. this we illustrated by febvre’s Le Problème de l’incroyance.

In all of these cases (and many others too numerous to cite) the standing of various general laws is the issue. far from being nonexistent and inapplicable to history, general laws—properly understood as pragmatic and causal in the ways we have outlined—are in fact at the center of historiographical debate and explanatory practice.

Ghent University

On Lawfulness in History and Historiography

Documents

function of general

history of religionhave

universal laws

scientific writing of

new economic history

examplesfrom cultural

general discussion

historiography isor