"Style" for Historians and Philosophers

This essay is a substantial development of the ideas of Chapter 11. It was first presented in Corfu in May 1991, at a conference on "Recent Trends in the Historiography of Science," which benefited from the wonderful leader-ship and hospitality of Kostas Gavroglu and Aristide Baltas.

Relations between the history and the philosophy of the sciences are often debated and sometimes contested. My interest here is collaboration. I shall describe a new analytical tool that can be used by historians and by philosophers for different purposes. It is a specialized, indeed technical, version of an idea often used or abused elsewhere: "style." The historian of science A. C. Crombie had been writing about "styles of scientific thinking in the European tradition" since the mid-1970s, and his work finally came to fruition in three volumes (1994). I heard him lecture on the topic in 1978, and in chapter 11 adapted the idea to metaphysics and epistemology, changing the name slightly to "styles of reasoning." The two uses, by historians and philosophers, are complementary but to some extent asymmetric. The historian may conclude that the philosopher's use of the tool is bunk, irrelevant to understanding the past. But the philosopher needs the history, for if the tool does not provide a coherent and enlightening ordering of the re-cord, then it has no more place in sound philosophy than would any other fantasy.

Crombie's idea is less about the content of the sciences than about their methods. The focus is on how we find out, not on what we find out. It is out of step with present fashion, which teaches us so much about the intricate details of incidents and relationships. It derives from a conception of the entire Western scientific tradition; we cannot help but recall that

Spengler (1918, 1922) too spoke of the "Western style." His use of the word Stil is so generous that his translator says, "The word Stil will therefore not necessarily be always rendered `style"' (1926, 108). Be prepared, however, for surprise translations. For example, die Expansionskraft der abendländischen Stil (1922, 55) becomes, in translation, "the expansion power of the Western Soul" (1926, 46). Crombie's ambitious analysis should remind us more, however, of the cabinet than of Spengler's bandstand, for it draws upon a lavish array of citations spanning three millennia, plus dense references to secondary studies—the lifetime collection of an erudite.

One root source is, of course, the History of Art. Arnold Davidson (2001c) traces the passage from art history to epistemology. Phrases like "style of thinking" or "reasoning" occur naturally enough without specialist connotations. This is to be expected with a word like "style" that already has so many connotations. In chapter 11 I mentioned the cosmologist Stephen Weinberg and the theoretical grammarian Noam Chomsky. Both authors attribute their idea of a Galilean style of reasoning to Husserl. I. B. Cohen gave a more detailed account of the same kind of reasoning; he called it "the Newtonian style," a way of combining "two levels of ontology," the mathematical and the measurable.

A case could perhaps be made out [he added] that this style is Galilean or Keplerian, rather than an invention of Newton's. In fact, Edmund Husserl has written at large concerning the "Galilean" style, essentially the mode of modern mathematical physics; from this point of view, the Newtonian style can be seen as a highly advanced and very much refined development of the Galilean. (Cohen 1982, 49).

Cohen and Weinberg were referring to §9 of Husserl (1970, part 2). In this very long section Husserl certainly did write, as Cohen puts it, "at large" about Galileo as the discoverer of a new kind of science, but I do not think that he used the words "Galilean style." In fact, I do not think he used the word "style" in the way any of those three writers do, or as I do. For example, the word is used six times on one page (Husserl 1970, 31), twice with emphasis in the original German, but in each case to refer to a feature of the "empirically intuited world."

Literary critics have long distinguished a "generalizing" and a "personalizing" use of the word "style." There is a Balzacian style and there is Balzac's style. Equally, in swimming, there is the Australian crawl and freestyle, as opposed to the style of Patti Gonzalez, that can be imitated but is inimitably hers. It is entirely natural to talk of the style of an individual scientist, research group, programme or tradition. Kostas Gavroglu, although taking the word "style" from myself and by derivation from Crombie, has quite legitimately put the word to its personalizing usage, for he contrasts the "style of reasoning" of two low temperature laboratories, and indeed of two men, Dewar and Kaemerlingh Onnes (Gavroglu 1990). Crombie and I instead intend something more attuned to Cohen, Chomsky, and Weinberg than to Gavroglu. And even if we put aside all obviously personalizing uses of "styles" of thinking, there are plenty of generalizing uses in the history or philosophy of science that differ from Crombie's. For example, Freeman Dyson's third Gifford lecture "is concerned with the history of science. It describes two contrasting styles in science, one welcoming diversity and the other deploring it, one trying to diversify and the other trying to unify" (Dyson 1988, 13).

For historians and philosophers, the most famous instance of another idea of style is in Ludwik Fleck's fundamental book of 1935, subtitled Introduction to the Theory of the Thought Style and the Thought Collective (Fleck 1979). By a thought style Fleck meant something less sweeping than Crombie, more restricted to a discipline or field of inquiry. Nevertheless, a thought style is impersonal, the possession of an enduring social unit, the "thought collective." It is "the entirety of intellectual preparedness or readiness for one particular way of seeing and acting and no other" (Fleck 1979, 64). Fleck intended to limn what it was possible to think; a Denkstil makes possible certain ideas and renders others unthinkable. Crombie and I fix on an extreme end of the spectrum of such permissible uses, and accordingly enumerate very few styles of thinking or reasoning. This is partly because our unit of analysis is very large in scope. There are many other units of analysis comparable to Fleck's, and which also deal with what it is possible to say. They are thoroughly impersonal, but more restricted in scope, in time and in space. For many purposes they may be, for that very reason, more instructive than something along the lines of Spengler or Co-hen or Weinberg and Chomsky. We think for example of Michel Foucault's episteme and discursive formation, or Nicholas Jardine's not unrelated "questions" (Jardine 1991).

I prefer to speak of styles of (scientific) "reasoning" rather than Crombie's "thinking." This is partly because thinking is too much in the head for my liking. Reasoning is done in public as well as in private: by thinking, yes, but also by talking and arguing and showing. This difference between Crombie and myself is only one of emphasis. He writes that "the history of science has been the history of argument"—and not just thinking. We agree that there are many doings in both inferring and arguing. Crombie's book describes a lot of them, and his very title happily ends not with science but with "Sciences and Arts." He has a lot to say about architecture, clock making, and the doctrine that "knowing is making." Nevertheless, there may still be a touch too much thinking for my pleasure. He titled a 1988 prospectus for his book, "Designed in the mind" (Crombie 1988). Does one not hear the resonance of Crombie's somewhat Koyrean origins? Even my word "reasoning" has too much to do with mind and mouth and keyboard; it does not, I regret, sufficiently invoke the manipulative hand and the attentive eye. Crombie's last word in the title of his book is "Arts;" mine would be "Artisan."

But there's more to my preference for reasoning over thinking than that. It recalls me to my roots—I am talking about what Aristotle called rational, even if my analysis is better suited to the temper of our times than his. "Reasoning" recalls the Critique of Pure Reason. My study is a continuation of Kant's project of explaining how objectivity is possible. He proposed preconditions for the string of sensations to become objective experience. He also wrote much about science, but only after his day was it grasped how communal an activity is the growth of knowledge. Kant did not think of scientific reason as a historical and collective product. We do. My styles of reasoning, eminently public, are part of what we need to understand what we mean by objectivity. This is not because styles are objective (that is, that we have found the best impartial ways to get at the truth), but be-cause they have settled what it is to be objective (truths of certain sorts are what we obtain by conducting certain sorts of investigations, answering to certain standards).

Crombie does not expressly define "style of scientific thinking in the European tradition." He explains it ostensively by pointing to six styles that he then describes in painstaking detail. "We may distinguish in the classical scientific movement six styles of scientific thinking, or methods of scientific inquiry and demonstration. Three styles or methods were developed in the investigation of individual regularities and three in the investigation of the regularities of populations ordered in space and time" (Crombie 1988, 10). These six are (I combine and select wording from several of his expositions):

(a) The simple method of postulation exemplified by the Greek mathematical sciences.

(b) The deployment of experiment both to control postulation and to explore by observation and measurement.

(c) Hypothetical construction of analogical models.

(d) Ordering of variety by comparison and taxonomy.

(e) Statistical analysis of regularities of populations, and the calculus of probabilities.

(f) The historical derivation of genetic development.

I am glad that he includes mathematics among the sciences, which is where they belong, whatever some of my recent philosophical predecessors may have thought. I do not mean that mathematics is empirical—only that it is a science. Note that styles do not determine a content, a specific science. We do tend to restrict "mathematics" to what we establish by mathematical reasoning, but aside from that, there is only a very modest correlation between items (a) through (f) and a possible list of fields of knowledge. A great many inquiries use several styles. For example, the fifth, statistical style is now used, in various guises, in every kind of investigation, including some branches of pure mathematics. The paleontologist uses experimental methods to carbon date and order the old bones. The "modern synthesis" of evolutionary theory is among other things a synthesis of taxonomic and historico-genetic thought.

I start with a canonical list of styles descriptively determined by a historian who, whatever his axes, is not grinding any of mine. As a philosopher I need to discover, from his examples, at least a necessary condition for there being such a "style." We are not bound to accept Crombie's preferred descriptions, nor to conclude with exactly his arrangement of styles. I shall list three related reasons why we may diverge and then give two examples.

(1) Crombie offers an account of the "classical scientific movement" and tailors his characterizations to the long period of time in which that movement was formed and firmed up. He tends to leave a given style at the date when it is securely installed. His discussions of mathematics end with Kepler's revivals of Greek mathematics. His exposition of the first three styles dries up at the end of the seventeenth century. Only the final style is developed for the nineteenth century, with Darwin being the major figure. But I as philosopher am decidedly Whiggish. The history that I want is the history of the present. That is Michel Foucault's phrase, implying that we recognize and distinguish historical objects in order to illumine our own predicaments. Hence I might modify Crombie's list not to revise his history but to view it from here.

(2) Crombie's (a) to (f) is itself a historical progression, each style beginning later than its predecessor in the list, and his presentation of each successive style concludes closer to the present than his descriptions of pre-ceding styles. What strikes me, however, is the ahistorical point that all six styles are alive and quite well right now. I am writing about what styles of scientific reasoning do for us. What is important now may be different from what was important in the early days.

(3) Crombie did not intend to write down an exhaustive list of mutually exclusive styles. He transcribed what he found central and enduring in the formative period of the Western vision. Quite aside from any styles that we might properly want to call scientific, and which evolved largely outside the West, there might also be yet earlier styles of "science" found, say, in re-cords of Babylonian computations, and not to be identified with a mere anticipation of (a). And certainly new styles may have evolved after the "classical" events Crombie recounts, just as new styles of reasoning may emerge in the future. There could also be mergings of two or more styles. I don't mean the truism that we commonly use more than one style in any modern inquiry, but that there may have evolved a style that is essentially composed of two classical styles, not a mixture but a compound, in the chemist's sense of the word—a new intellectual substance.

Now I turn to two examples. As a philosopher of mathematics, I see proof where Crombie sees postulation. His first style emphasizes the Greek search for first principles. It is there that he brings in Greek medicine, with its battle between empirics and dogmatists. We meet Aristotle during Crombie's discussion of (a)—even when the Stagyrite is canonizing what later becomes the taxonomic style (d). That is correct history, putting (a) and its contemporary correlatives first, in their place. Yet there is no doubt that what individuates ancient mathematics for us is that we recognize proof and to a limited extent calculation. Wilbur Knorr speculatively ordered segments of actual and lost texts by the development of proof procedures (Knorr 1975). Mathematics has the astonishing power to establish truths about the world independently of experience. That is the phenomenon that so astounded Socrates in the Meno, and has so vexed every serious epistemologist of mathematical science ever since. I will want my account of the mathematical style to help understand that phenomenon. Hence my emphases will differ from Crombie's.

For another example, the historical distinction between styles (b) and (c) is profoundly important. It has to do with the familiar tensions between today's experimenter and theoretician. The former is heir to the medical empirics, who insisted that we should never go beyond observables in our descriptions of the course of disease and its cure, while it was the dogmatists who introduced what we would now call theoretical entities that play so major a part in hypothetical modeling (c). Crombie speaks of "controlling postulation" in his summary description of (b), but the postulation is at the level of observables and measurable quantities. It is by and large the science of phenomena given or measured in nature that is not much tampered with. Something else began just about the end of the period for which Crombie describes (b) and (c). I call it the laboratory style, characterized by the building of apparatus in order to produce phenomena to which hypothetical modeling may be true or false, but using another layer of modeling, namely models of how the apparatus and instruments themselves work. The relationship between the laboratory style, call it (bc), and styles (b) and (c) is complex. Peter Galison (1998) describes it with the metaphor of a trading zone between the producers of analyzed data and the merchants of theoretical approximations. He took the idea from linguists studying the development of "Creole" or "pidgin" languages in which a new language develops, for purposes of trade and social inter-course, at the interface between two established languages. The trading-zone idea will be useful in the study of styles of reasoning when we begin to describe any inquiry that employs several styles. It is often not the case that a single investigator is at home in more than one style of reasoning. Instead, there is collaboration in which a person expert in style X makes use of a handy robust core of techniques from style Y. This is at its most obvious in "cookbooks" of statistical reasoning prepared for this or that branch of science, psychology, cladistic taxonomy, high energy physics, and so forth. With no understanding of principles, and perhaps using only a mindless statistical package for the computer, an investigator is able to use statistics without understanding its language in any meaningful way whatsoever.

To return to the laboratory style, I do not mean that it has supplanted Crombie's (b), experimentation, and (c) modeling. On the contrary, there are whole fields of specialization in which either (b) or (c) is in full play on its own. On the one hand, despite all the talk about intervening variables and the like, many of the social sciences operate only at the empirical level of (b). On the other hand, cosmology and cognitive science—none other than the chief modern instances of the Galilean style so admired by Weinberg and Chomsky—remain at the level of (c), hypothetical modeling.

Those sciences answer to observation, but experimental manipulation and intervention is almost never practicable. That is precisely why Weinberg and Chomsky invoke (a certain Koyrean vision of) Galileo to legitimate their own work. Cosmology and cognitive science remain sciences that represent; the laboratory style introduced sciences that intervene.

I judge that the laboratory style began about the time that Boyle made the air pump in order to investigate the spring of the air. It is characteristic of styles that they have popular myths of origin. Crombie's list strikes the right note just because it codifies familiar legend. How could it be other-wise if one is recapitulating European science from within? There was that legendary moment when, as Althusser put it, Thales "discovered the continent of mathematics" (Althusser 1972, 185). Next in the list of continents is "and Galileo discovered the continent of mechanics." Well, Galileo is everybody's favorite hero—not only for Chomsky and Weinberg but also for Husserl (for whom Galileo is simply The Hero of Science) and Spengler. Crombie's talk on styles of scientific thinking that aroused my interest long ago was about—Galileo. At that same conference, Winifred Wisan read a paper titled "Galileo and the Emergence of a New Scientific Style (Wisan 1981). All these authors referred chiefly to some aspect of style (c), so let us not forget that according to Stillman Drake, it was Galileo who, by the purest use of style (b), established the very first experimental and quantitative law of nature. Galileo is the stuff of myth, a point made by Crombie him-self (1987). Althusser continued, "and Marx discovered the continent of history." Good myth, wrong man; I much prefer Michel Foucault's retelling with Bopp, Cuvier, and Ricardo. Cuvier, as many have noticed, is question-able, and we'd add a geologist, but Bopp's philology seems perfect as the start of the historico-genetic style. As for style (e), that too has its legends. "A problem about games of chance proposed to an austere Jansenist by a man of the world was the origin of the calculus of probabilities," or so wrote Poisson (1837, 1). And I take Schaffer and Shapin's book, subtitled Hobbes, Boyle, and the Experimental Life (1986), as setting out the myth of origin for the laboratory style. Their hero, as both Bruno Latour (1990) and I (1991) have observed, is not a person but an instrument, the apparatus, the air pump.

Styles, to continue Althusser's metaphor, open up new territory as they go. I am sure that the Indo-Arabic style of applied mathematics, little interested in postulation but dedicated to finding algorithms, is a distinct style with, of course, non-European origins. I call it the algorismic style, referring to yet another legend. "Al-gorismi" was the early European name for the Arab mathematician who flourished in the early ninth century. (Abu Ja'far Mohammed Ben Musa, native of Khwarazm, or al-Khowarazmi.) His book on algebra (which is also probably the source of our word "algebra") was the text from which Europeans learned the Arabic numerals—and the algorismic style of reasoning.

The algebrizing of geometry, the Arabicizing of the Greek, was an essential piece of territorial expansion. Every such expansion is contested. We can overhear today's battles. For example: are computer-generated concepts and proofs really mathematics? When I was a student, I went around with some topologists who would talk and draw pictures and tell tall stories; today, when I have topological house guests, the first thing they do is set up their Macs in my basement, not calculating but generating ideas to which real-time computation is integral. And I know others who say that my friends have stopped doing mathematics. That's how it is, when a style goes into new territory.

For all these differences in emphases, I do not differ significantly from Crombie, either in my individuation of styles or in how I describe them. Without his three-volume vindication of his canonical list, I would be left with dubious anecdotes and fables. I'm not claiming that I'm on solid non-ideological ground when I resort to a historian for an initial individuation of styles. I claim only a certain independence: his motivation is very different from mine, but the list he presents admirably suits my purposes. It is a good workhorse of a list that holds no surprises. To use yet another obsolete metaphor, it covers the waterfront, and provides a directory to the main piers, in a readily recognizable and fairly satisfactory way. And it could be the wrong waterfront for me. Maybe he was describing a once wondrous but now gutted Liverpool, or at any rate a dignified San Francisco that has taken up leisure pursuits like denim and tourism—harbors that history has passed by. Perhaps I should instead be attending to a bustling container port like Felixstowe or Oakland. Maybe science as we know it began late in the nineteenth century and the philosopher who is not an antiquarian should just forget about the olden times. I don't think so. The proof of my confidence that Crombie's list remains germane is, however, not a matter of principle but of the success of the resultant philosophical analysis.

Our differences lie not in the identity of styles or their description, but in the use to which we put the idea. Crombie's advance notice of his book began: "When we speak today of natural science we mean a specific vision created within Western culture, at once of knowledge and of the object of that knowledge, a vision at once of natural science and of nature" (Crombie 1988, 1). He said on the next page that,

The whole historical experience of scientific thinking is an invitation to treat the history of science, both in its development in the West and in its complex diffusion through other cultures, as a kind of comparative historical anthropology of thought. The scientific movement offers an invitation to examine the identity of natural science within an intellectual culture, to distinguish that from the identities of other intellectual and practical activities in the arts, scholarship, philosophy, law, government, commerce and so on, and to relate them all in a taxonomy of styles. It is an invitation to analyse the various elements that make up an intellectual style in the study and treatment of nature: conceptions of nature and of science, methods of scientific inquiry and demonstration diversified ac-cording to subject matter, evaluations of scientific goals with consequent motivations, and intellectual and moral commitments and expectations generating attitudes to innovation and change.

This is history in the grand manner, an invitation to a comparative historical anthropology of thought. Regardless of interest, philosophical or historical, many of us may be glad that at a time of so many wonderfully dense and detailed but nevertheless fragmented studies of the sciences, we are offered such a long-term project. This is especially so for philosophers, to whom the most fascinating current historiography of the sciences is work of the "social studies of knowledge" schools of philosophically motivated history: the strong program, network theory, the doctrine of the construction of scientific facts by negotiation. Increasingly fine-grained analyses of incidents, sometimes made tape-recorder in hand, have directed the history of science towards the fleeting. On the other hand, many of my philosophical colleagues take it to the quasi-timeless end, as when Hilary Putnam writes of "the ideal end of inquiry." Crombie's styles may also seem to be edging off towards the excessively long run. But his intentions are plain, to conduct a historical investigation of that specific vision mostly created around the Mediterranean basin and then in more northerly parts of Europe, "a comprehensive historical inquiry into the sciences and arts mediating man's experience of nature as perceiver and knower and agent [that must] include questions at different levels, in part given by nature, in part made by man." Crombie was well aware of the need to establish the historical continuity of styles across periods of latency, of the need to understand "the intellectual and social commitments, dispositions and habits, and of the material conditions, that might make scientific activity and its practical applications intellectually or socially or materially easy for one society but difficult or impossible for another." He wanted to compare those now familiar items, "the numbers, social position, education, occupation, institutions, private and public habits, motives, opportunities, persuasions and means of communications of individuals," and so on: "military context," "rhetorical techniques of persuasion." The grand view need not neglect the fashionable topics of the moment, nor on the other hand ignore philosophical chestnuts like the existence of theoretical entities. That conundrum is described in the mandarin manner, as you will have noticed from my other quotations: "distinguishing the argument giving rational control of subject-matter from an implication of the existence of entities appearing in the language used" (that is, the polarizing electron gun works, but do electrons exist?).

I have hardly begun to enumerate Crombie's historiographic aims. How can a philosopher make use of so expansive an idea of a style of scientific thinking or reasoning in the European tradition? First, I notice the way in which styles become autonomous. Every style comes into being by little microsocial interactions and negotiations. It is a contingent matter, to be described by historians, that some people with disposable time and avail-able servants should value finding something out. Yet each style has be-come independent of its own history. We can forget the history or enshrine it in myth. Each style has become what we think of as a rather timeless canon of objectivity, a standard or model of what it is to be reasonable about this or that type of subject matter. We do not check to see whether mathematical proof or laboratory investigation or statistical studies are the right way to reason: they have become (after fierce struggles) what it is to reason rightly, to be reasonable in this or that domain.

I assert neither that people have decided what shall count as objectivity, nor that we have discovered what does the trick. I am concerned with the way in which objectivity comes into being, and shall shortly state how to address the question of what keeps certain standards of objectivity in place. Why do I not say that we have simply discovered how to be objective, how to get at the truth in a long haul? This is because there are neither sentences that are candidates for truth, nor independently identified objects to be correct about, prior to the development of a style of reasoning. Every style of reasoning introduces a great many novelties including new types of:

One will also notice, on occasion, new types of classification and new types of explanations. We should not envisage first a style and then the novelties. That is one of the many merits of the word "style." We did not first have fauvism, and then Matisse and Derain painting fauve pictures in 1905. The style comes into being with the instances, although (as the example of the fauves makes plain) the recognition of something as new, even the naming of it, may solidify the style after it has begun. What the word "style" does not make plain is why fauvism fades almost as soon as named, while a few styles of reasoning become autonomous of their origins and their originators. That is a pressing philosophical issue in the study of styles of reasoning.

Each style, I say, introduces a number of novel types of entities, as just listed. Take objects. Every style of reasoning is associated with an ontological debate about a new type of object. Do the abstract objects of mathematics exist? That is the problem of Platonism in mathematics. Do the unobservable theoretical entities of the laboratory style really exist? That is the problem of scientific realism in the philosophy of the natural sciences. Do the taxa exist in nature, or are they, as Buffon urged, mere artifacts of the human mind? Are there objects, such as languages, to be understood in terms of their historical derivation, or are they just a way of organizing a mess of complexity on top of the only reality, a postulated innate universal grammar? Are coefficients of correlation or the rates of unemployment real features of populations, or are they products of institutional arrangements of classification and measurement?

Each style of reasoning has its own existence debate, as illustrated, be-cause the style introduces a new type of object, individuated by means of the style, and not previously noticeable among the things that exist. In-deed, the realism-antirealism debates so familiar in recent philosophy will now be understood in a new and encyclopedic fashion, as a by-product of styles of reasoning. That is not true of questions of global idealism, mind versus matter debates, which are not engendered by this or that style of reasoning.

Objects are only one kind of novelty. One may run down my list of novelties checking that each style introduces these novelties. That, I argue, is an essential and definitive feature of a style of reasoning, accounting for the relatively small number of styles on Crombie's list. Hence we are in a position to propose a necessary condition for being a style of reasoning: each style should introduce novelties of most or all of the listed types, and should do so in an open-textured, ongoing, and creative way. Mathematicians do not just introduce a few sorts of abstract objects, numbers, and shapes, and then stop; the type "abstract object" is open-ended once we be-gin reasoning in a certain way. Note that on this criterion, logic, be it deductive, inductive, or abductive, does not count as a style of reasoning. This is as it should be. Crombie did not list the branches of logic, and no wonder. People everywhere make inductions, draw inferences to the best explanation, make deductions; those are not peculiarly scientific styles of thinking, nor are they Mediterranean in origin.

I use my list of novelties as a criterion, as a necessary condition for being a style of reasoning. I've mentioned the ontological debates arising from one sort of novelty, the new types of objects; now I shall say a little more about new types of sentences. Each new style, and each territorial extension, brings with it new sentences, things that were quite literally never said before. That is hardly unusual. That is what lively people have been doing since the beginning of the human race. What's different about styles is that they introduce new ways of being a candidate for truth or for falsehood. As Comte put it—and there is a lot of Comtianism in my philosophy—they introduce new kinds of "positivity," ways to have a positive truth value, to be up for grabs as true or false. Any reader who fears too much early positivism should know also that I took the word "positivity" in the first in-stance from Michel Foucault, whose influence on my idea of styles of reasoning is more profound than that of Comte or Crombie. I should repeat for philosophers what was said in chapter 11, that his idea of positivity falls far short of what Michael Dummett calls bivalence, of being definitely true, or definitely false. Bivalence commonly requires far more to be in place than a style of reasoning. It may demand a Foucauldian episteme, some of Jardine's questions, or even, as Gavroglu argues, an entirely personal research style localized in a single laboratory. And even after all that, as Dummett has well taught, even when similarities in the surface gram-mar and in possible ways of inquiry may make us think that sentences we investigate using them are beyond question bivalent, closer scrutiny abetted by a stern theory about meaningfulness may make us skeptical.

The kinds of sentences that acquire positivity through a style of reasoning are not well described by a correspondence theory of truth. I have no instant objection to a correspondence theory for lots of humdrum sentences, what we might call pre-style or unreasoned sentences, including the maligned category of observation sentences. But I reject any uniform all-purpose semantics. The instant objection to correspondence theories, for sentences that have positivity only in the context of a style of reasoning, is that there is no way of individuating the fact to which they correspond, except in terms of the way in which one can investigate its truth, namely by using the appropriate style. As J. L. Austin showed, that objection does not so instantly apply, for example, to "observation sentences" in subject-predicate or subject-relation-object form. I reject the first dogma of traditional anglophone philosophy of language, that a uniform "theory of truth" or of "meaning" should apply across the board to an entire "language." That is a fundamental lesson to draw from Wittgenstein's talk of different "language-games." Among shop-soiled theories of truth and meaning, the one that best fits sentences of a kind introduced by a style of reasoning is a verification theory.

The truth of a sentence (of a kind introduced by a style of reasoning) is what we find out by reasoning using that style. Styles become standards of objectivity because they get at the truth. But a sentence of that kind is a candidate for truth or falsehood only in the context of the style. Thus styles are in a certain sense "self-authenticating" Sentences of the relevant kinds are candidates for truth or for falsehood only when a style of reasoning makes them so. This statement induces an unsettling feeling of circularity.

The statement is closely connected with the claim that styles of reasoning introduce novelties, including new kinds of sentences. There simply do not exist true-or-false sentences of a given kind for us to discover the truth of, outside of the context of the appropriate style of reasoning, The doctrine of self-authenticating styles is distinct from "constructionist" ac-counts of scientific discovery. For in those accounts individual facts of a typically familiar kind become constructed-as-facts in the course of re-search and negotiation. There was no fact "there" to discover until constructed. According to my doctrine, if a sentence is a candidate for truth or

falsehood, then by using the appropriate style of reasoning we may find out whether it is true or false. There is more to say here, connected with the difference between a sentence having positivity and bivalence, but I may have said enough to show how my doctrine falls a long way short of constructionism.

The apparent circularity in the self-authenticating styles is to be welcomed. It helps explain why, although styles may evolve or be abandoned, they are curiously immune to anything akin to refutation. There is no higher standard to which they directly answer. The remarkable thing about styles is that they are stable, enduring, accumulating over the long haul. Moreover, in a shorter time frame, the knowledge that we acquire using them is moderately stable. It is our knowledges that are subject to revolution, to mutation, and to several kinds of oblivion; it is the content of what we find out, not how we find out, that is refuted. Here lies the source of a certain kind of stability. Some years ago, when I published a brief paper about the stability of the laboratory sciences, I could refer only to some observations by S. S. Schweber and to some work on "finality in science" done by a group in Frankfurt (Hacking 1988b). Now the topic of stability is positively trendy, and at the time of writing this had occupied the correspondence pages of the Times Literary Supplement for several weeks (Durant 1991, cf Hacking 1999a, 84-92).

I believe that understanding the self-authenticating character of styles of reasoning is a step towards grasping the quasi-stability of science. I doubt that Crombie agreed with this. If so, our difference would not be between a historical judgment and a philosophical one, but rather a philosophical difference between two students, one a historian and one a philosopher. Other historians, of a more constructionist bent, will hold that my doctrine of self-authentication does not go far enough; in any event, the issues are philosophical, not historical.

In respect of stability I do wholly endorse one much used lemma from the strong program in the sociology of knowledge. The truth of a proposition in no way explains our discovery of it, or its acceptance by a scientific community, or its staying in place as a standard item of knowledge. Nor does being a fact, nor reality, nor the way the world is. My reasons for saying so are not Edinburgh ones; they are more reminiscent of very traditional philosophy. I would transfer to truth (and to reality) what Kant said about existence, that it is not a predicate, adding nothing to the subject. I may believe that there was a solar eclipse this summer because there was one in the place I was then staying; the eclipse is part of the explanation of my belief (a view which might be resisted in Edinburgh), along with my experience, my memory, my general knowledge, the folderol in the news-papers, etc. But the fact that there was an eclipse, or the truth of the proposition that there was an eclipse, is not part of the explanation, or at any rate not over and above the eclipse itself. This is no occasion to develop that theme, except to say that anyone who endorses the Edinburgh conclusion, that truth is not explanatory, should want an understanding of the stability of what we find out, and not settle for "because that's the way that the world is." I shall now sketch how the theory of styles of reasoning may pro-vide such an understanding.

The idea of self-authentication is only a step, a fingerpost, towards an understanding of the quasi-stability of some of our knowledge. We shall not progress further by thinking about method in general, let alone "science" in general. Each style of reasoning has its own characteristic self-stabilizing techniques. An account of each technique requires detailed analysis, specific to the style, and it is aided by vivid historical illustration. Each is a long story. I have published three papers about the statistical, the mathematical, and the laboratory styles (Hacking 1991a, 1992a, 1995b). There is little overlap between these essays, because the techniques and the histories involved differ substantially from case to case.

Almost the only thing that stabilizing techniques have in common is that they enable a self-authenticating style to persist, to endure. Talk of techniques that I describe are quite well known, but, I claim, inadequately understood. For example, Duhem's famous thesis about how to save theories by adjusting auxiliary hypotheses is (by one measure) exactly 1/14th of the stabilizing techniques that I distinguish in the laboratory sciences. I owe much more to recent work by Andrew Pickering, to whom I would at-tribute another 3/14th (Pickering 1989). Overall we are concerned with a mutual adjustment of ideas (which include theories of different types), materiel (which we revise as much as theories) and marks (including data and data analysis). All three are what Pickering calls plastic resources that we jointly mold into semi-rigid structures. I should emphasize that, al-though I use Duhem, this account does not go in the direction of the underdetermination of theory by data (Quine's generalization of Duhem's remarks). On the contrary, we come to understand why theories are so de-terminate, almost inescapable. Likewise my account of the stability of the mathematical style owes much to two unhappy bedfellows, Lakatos and

Wittgenstein. It introduces an idea of "analytification"—of how some synthetic if a priori propositions are made analytic: thus the logical positivist doctrine of the a priori is historicized. But we no more arrive at the radical conventionalism or constructionism sometimes read into Remarks on the Foundations of Mathematics than we arrive at the underdetermination of theory by data (Hacking 2000).

A happy by-product of this analysis is that not only has each style its own self-stabilizing techniques, but also some are more effective than others. The taxonomic and the historic-genetic styles have produced nothing like the stability of the laboratory or the mathematical style, and I claim to be able to show why. On the other hand, although Mark Twain, Disraeli, or whoever could, in the earlier days of the statistical style, utter the splendid canard about lies, damn lies, and statistics, the statistical style is so stable that it has grown its own word that gives a hint about its most persistent techniques: "robust." In the case of statistics there is an almost too evident version of self-authentication (the use of probabilities to assess the probabilities). But that is only part of the story, for I emphasize the material, institutional requirements for the stability of statistical reasoning. Indeed, if my accounts deserve to be pegged by any one familiar philosophical "-ism," then it is materialism. That is most notably true of my account of the laboratory style, despite my incorporating the idealist "Quine-Duhem thesis" as an adjunct.

Techniques of self-stability return us to the question of how to individuate styles. We began with an ostensive definition, Crombie's list. Then we moved to a criterion, a necessary condition: a style must introduce certain novelties, new kinds of objects, laws, and so on. But now we get closer to the heart of the matter. Each style persists, in its peculiar and individual way, because it has harnessed its own techniques of self-stabilization. That is what constitutes something as a style of reasoning.

This three-stage definition of styles must be treated with caution. Consider for example the question of whether some styles of reasoning have simply died out, after a robust life in recorded history. In chapter 11 I suggested one extinct style, the Renaissance reasoning by similitudes so well represented by Paracelsus. My characterization of styles above began with a historian's classification of styles, not intended to be exhaustive: of course that allows that some styles are no longer with us. I don't know what Crombie thought about Renaissance medicine, but I know of nothing in his published writing to exclude it as participating in an additional style that has now been abandoned. So the possibility of "dead" styles is unproblematic for the first, ostensive, definition of styles. Barry Allen (1996) has suggested witchcraft as another.

Next I gave a philosopher's necessary condition for being a style, in terms of introducing a battery of new kinds of objects. That too allows that some styles die out. It is at least arguable that the reasoning of a Paracelsus satisfies this criterion. Finally I have suggested—and here, only suggested—a more analytic definition, in terms of a style being constituted by self-stabilizing techniques. There arise two questions: first, whether hermetic medicine of those times did have such techniques, and then, if the answer is yes, why this style of reasoning has been so brusquely displaced. I believe that the answer to the first question is a qualified affirmative—but see my observation above about some techniques being more effective than others. The second question leads into dense history; recall complaints addressed to Michel Foucault that he never explained why epistemes die out, in particular why his Renaissance episteme of resemblance expired. I do not believe that one can give purely internal explanations of why we abandon certain practices, but have no confidence in external explanations either. It does not discredit the philosopher's use of styles of reasoning that it leads directly to such historical chestnuts; the contrary, I should imagine.

The historian will want to distinguish several types of events. There is the extinction of a style, perhaps exemplified by reasoning in similitudes. There is the insertion of a new style that may then be integrated with an-other, as has happened with algorismic reasoning, combined with geometrical and postulational thought. There is the challenge offered by a new style, the laboratory style, to an old one, the postulational style, and the en-suing triumph of the new. I am inclined to go with the contingency theorists among historians on all these points. It is altogether contingent that there have been such replacements, and the concept of style is of no aid in explaining what happened. Style is a more metaphysical concept, important for understanding truth-or-falsehood once a style has become autonomous.

More pressing to the philosopher than dead, merging, or emerging styles are questions posed in the present by a number of special-interest groups. What are the other styles of reasoning? Historical reasoning? Legal reasoning? Mystical reasoning? Magical reasoning? John Forrester (1996) makes a case for the "case," in psychoanalysis, as a style of reasoning. Arnold Davidson (1996) has a more general argument for the whole of psychiatry.

We may even, in a modest way, get a grip on Richard Rorty's question,

"Is Science a Natural Kind?" (Rorty 1988) without being reduced to his view that there is only the largely undifferentiated conversation of man-kind. We can do so without embracing the opposed idea of Bernard Williams, that science leads us to something worth calling an absolute conception of reality (Williams 1984, 214 and 1985, 138-139). The very mention of styles, in the plural, corrects the direction of the debate: we shall stop talking of science in the singular and return to that healthy nineteenth-century practice of William Whewell and most others: we shall speak of the history and philosophy of the sciences—in the plural. And we shall not speak of the scientific method as if it were some impenetrable lump, but instead address the different styles (Hacking 1996). On the other hand, once we have a clearer understanding of what, from case to case, keeps each style stable in its own way, we shall not think that there are just end-less varieties of Rortian "conversation" My doctrine of self-authentication, which sounds like part of the current mood for sceptically undermining the sciences, turns out to be a conservative strategy explaining what is peculiar about science, distinguishing it to some extent from humanistic and ethical inquiry.

A proposed account of self-stabilizing techniques begins by observing that a style becomes autonomous of the local microsocial incidents that brought it into being. Then there is the detailed account of how each style does stabilize itself. That is not the end of the matter. It is a contingent fact about us and our world that the techniques work at all—that we can analytify the sentences of mathematics or create phenomena in the laboratory to which our models are true. The persistence of a style demands some brute conditions about people and their place in nature. These conditions are not topics of the sciences, to be investigated by one or more styles, but conditions for the possibility of styles. An account of them has to be brief and banal, because there is not much to say. What we have to supply are, to quote Wittgenstein, "really remarks on the natural history of man: not curiosities, however, but rather observations on facts which no one has doubted and which have only gone unremarked because they are always there before our eyes" (Wittgenstein 1981, 47). Wittgenstein and others also called this (philosophical) anthropology (cf Bloor 1983). The resonance is with Kant's Anthropologie rather than the ethnography or ethnology commonly studied in departments of anthropology or sociology. Crombie's "comparative historical anthropology of thought" is by and large historical ethnology, a comparative study of one profoundly influential aspect of Western culture. Wittgenstein's philosophical anthropology is about the "natural history of man," or, as I prefer to put it, about human beings and their place in nature. It concerns facts about all people, facts that make it possible for any community to deploy the self-stabilizing techniques of styles of reasoning. It is in philosophical anthropology that we slough off the Eurocentrism with which this study began.

We should be wary of giving grand names to our modest projects. In 1990, two works were subtitled "Towards an Anthropology of Science." They were written a few hundred meters apart, at the École Polytechnique and the École des Mines in Paris. They were both published in England (Atran 1990, Latour 1990). They are both written from perspectives at right angles to Wittgenstein's. My own abuse of what Wittgenstein meant by anthropology has more in common with Atran than with Latour. This is because Atran is concerned with—among many other things in his extraordinarily versatile book—what made possible the taxonomic style (d). He also has a Chomskian vision of an underlying, innate, universal structure for what he calls folk-taxonomy. In contrast, Latour's projected anthropology of science is profoundly anti-innatist, anti-universalist. These two authors do, however, have one important thing in common, as distant as possible from Wittgenstein's Anthropologie or natural history of man-kind. Atran does real ethnography, studying classification systems used by Mayan peoples in the jungles of Guatemala. Latour too was trained as an ethnographer, and his study of the synthesis and identification of a tripeptide was conceived of as an ethnography of the workplace, the laboratory (Latour and Woolgar 1979). That work now serves as a role-model—or as horrid cautionary tale—for a generation of trainee anthropologists whose field site is a laboratory. (I myself think it is a plausible example of a less plausible general thesis about constructionism; Hacking 1988d.)

What should we call my inquiry, less ambitious than philosophical anthropology, namely the detailed study of the stabilizing techniques used by a given style? If we are to use the suffix "-ology," then a fitting name for the study of self-stabilizing techniques would be philosophical technology. This label does not carry its meaning on its face, for I am not talking about what we usually mean by "technology," namely the development, application, and exploitation of the arts, crafts, and sciences. What I mean by philosophical technology is the philosophical study of certain techniques, just as philosophical anthropology is the study of certain aspects of man, epidemiology of epidemic diseases.

We have finally reached the fundamental difference between the historian's and the philosopher's use of the idea of a scientific style of thinking or reasoning, a difference that has nothing to do with disagreements about history or divergence in philosophies. Crombie led us to a comparative historical anthropology (moved, he also told us, by the experiences of teaching in Japan, and of crossing parts of Asia and its oceans when visiting his native Australia). I invite what I call philosophical technology: a study of the ways in which the styles of reasoning provide stable knowledge and be-come not the uncoverers of objective truth but rather the standards of objectivity. And when asked how those techniques could be possible at all, I fall back on a few and very obvious remarks about people, of the sort to which Wittgenstein has already directed us. Less all-encompassing histories will provide the social conditions within which a style emerged and those in which it flourished; less ambitious essays in philosophical technology will describe, in a more fine-grained way, the ways in which a style took on new stabilizing techniques as it pursued its seeming destiny in new territories. Comparative historical anthropology is a fundamentally different enterprise from either philosophical anthropology or philosophical technology.

I began by saying that the philosopher requires the historian. If Crombie's three volumes did not present a coherent ordering and analysis of European scientific practice and vision, then my talk of self-authenticating styles and of philosophical technology would be suspect. That is why I called the relation between the history and the philosophy of the sciences asymmetric. The philosopher who conceives of the sciences as a human production and even invention requires the historian to show that analytic concepts have application. After learning from the historian's analysis, I turn to a different agenda, which, you will have noticed, summons all the old gang: truth, reality, existence. But also, as is always the case in philosophy, we are directed to a complementary range of entirely new topics, such as philosophical technology.

For all the manifest differences of endeavor between the historian and the philosopher, they have this in common: we share a curiosity about our Western "scientific" vision of objectivity. That is as central a philosophical concern as could be: the core question of Kant's first critique. Crombie's volumes will, I hope, be read in part as an account of how conceptions of objective knowledge have come into being, while the philosopher can de-scribe the techniques which become autonomous of their historical origins, and which enable styles of reasoning to persist at all. Yet I would not push this division of labor too far. As I said in chapter 1, objectivity, in its several guises, is a hot topic for active historians of science such as Lorraine Daston, Peter Galison, Theodore Porter, and many others. Even when objectivity is not explicitly in view, however much the historian may abjure philosophical issues, every sound history is imbued with philosophical concepts about human knowledge, nature, and our conception of it. And aside from central shared concerns, there is a more general predicament that the historian and the philosopher experience. Crombie was powerfully aware of the reflexive elements of his volumes. He knew that he who de-scribes a certain vision of ourselves and our ecology has that vision him-self. More constraining, although more difficult to come to coherent terms with, philosopher and historian alike are part of the community of living things that has been transformed by bearers of that vision in their interactions with nature as they saw it.