THOMAS S. KUHN

The Function of Dogma in Scientific Research 1

At some point in his or her career every mem­ that phrase may mean, the individual scientist ber of this Symposium has, I feel sure, been ex­ is very often not. Whether rus work is predom­ posed to the image of the scientist as the inantly theoretical or experimental, he usually uncommitted searcher after truth. He is the ex­ seems to know, before his research project is plorer of nature-the man who rejects preju­ even well under way, all but the most intimate dice at the threshold of his laboratory, who details of the result which that project will collects and examines the hare and objective achieve. If the result is quickly forthcoming, facts, and whose allegiance is to such facts and well and good. If not, he will struggle with his to them alone. These are the characteristics apparatus and with his equations until, if at all which make the testimony of scientists so valu­ possible, they yield results which conform to able when advertising proprietary products in the sort of pattern which he has foreseen from the United States. Even for an international au­ the start. Nor is it only through his own re­ dience, they should require no further elabora­ search that the scientist displays his firm con­ tion. To be scientific is, among other things, to victions about the phenomena which nature be objective and open-minded. can yield and about the ways in which these Probably none of us believes that in practice may be fitted to theory. Often the same con­ the real-life scientist quite succeeds in fulfilling victions show even more clearly III his response this ideal. Personal acquaintance, the novels of to the work produced by others. From Galileo's Sir Charles Snow, or a cursory reading of the reception of Kepler's research to Nageli's re­ history of science provides too much counter­ ception of Mendel's, from Dalton's rejection of evidence. Though the scientific enterprise may Gay Lussac's results to Kelvin's rejection of be open-minded, whatever this application of Maxwell's, unexpected novelties of fact and

Reprinted from Scientific Change, Alistair C. Crombie, Ed. (1963), pp. 347-369. © 1963 by Heinemann Educa­ tional Books, Ltd. Reprinted by pennission of Basic Books, Inc., publishers.

301 302 PART IV THE HISTORICAL DEVElOPMENT OF SCIENTIFIC KNOWLEDGE theory have characteristically been resisted and off scientific progress, their omnipresence is have often been rejected by many of the most nonetheless symptomatic of characteristics creative members of the professional scientific upon which the continuing vitality of research community. The historian, at least, scarcely depends. Those characteristics I shall collec­ needs Planck to remind him that "A new sci­ tively call the dogmatism of mature science, entific truth is not usually presented in a way and in the pages to come I shall try to make the that convinces its opponents ... ; rather they following points about them. Scientific educa­ gradually die off, and a rising generation is fa­ tion inculcates what the scientific community miliarized with the truth from the start.'" had previously with difficulry gained-a deep Familiar facts like these-and they could commitment to a particular way of viewing the easily be multiplied-do not seem to bespeak world and of practicing science in it. That an enterprise whose practitioners are notably commitment can be, and from time to time is, open-minded. Can they all be reconciled with replaced by another, but it cannot be merely our usual image of productive scientific re­ given up. And, while it continues to character­ search? If such a reconciliation has not seemed ize the community of professional practition­ to present fundamental problems in the past, ers, it proves in two respects fundamental to that is probably because resistance and precon­ productive research. By defining for the indi­ ception have usually been viewed as extrane­ vidual scientist both the problems available for ous to science. They are, we have often been pursuit and the nature of acceptable solutions told, no more than the product of inevitable to them, the commitment is actually constitu­ human limitations; a proper scientific method tive of research. Normally the scientist is a puz­ has no place for them; and that method is pow­ zle-solver like the chess player, and the erful enough so that no mere human idiosyn­ commitment induced by education is what crasy can impede its success for very long. On provides him with the rules of the game being this view, examples of a scientific parti pris are played in his time. In its absence he would not reduced to the status of anecdotes, and it is that be a physicist, chemist, or whatever he has evaluation of their significance that this essay been trained to be. aims to challenge. Verisimilitude, alone, sug­ In addition, commitment has a second and gests that such a challenge is required. Precon­ largely incompatible research role. Its very ception and resistance seem the rule rather than strength and the unanimiry with which the the exception in mature scientific develop­ professional group subscribes to it provides the ment. Furthermore, under normal circum­ individual scientist with an immensely sensitive stances they characterize the very best and most detector of the trouble spots from which sig­ creative research as well as the more routine. nificant innovations of fact and theory are al­ Nor can there be much question where they most inevitably educed. In the sciences most come from. Rather than being characteristics discoveries of unexpected fact and all funda­ of the aberrant individual, they are community mental innovations of theory are responses to a characteristics with deep roots in the proce­ prior breakdown in the rules of the previously dures through which scientists are trained for established game. Therefore, though a quasi­ work in their profession. Strongly held convic­ dogmatic commitment is, on the one hand, a tions that are prior to research often seem to source of resistance and controversy, it is also be a precondition for -success in the sciences. instrumental in making the sciences the most Obviously I am already ahead of my story, consistently revolutionary of all human activi­ but in getting there I have perhaps indicated its ties. One need make neither resistance nor principal theme. Though preconception and dogma a virtue to recognize that no mature sci­ resistance to innovation could very easily choke ence could exist without them. Before exam- THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 303 ining further the nature and effects of scientifIc early years not exclusively. But in the sciences dogma, consider the pattern of education different textbooks display different subject through which it is transmitted from one gen­ matters rather than, as in the humanities and eration of practitioners to the next. Scientists many social sciences, exemplifying different ap­ are not, of course, the only professional com­ proaches to a single problem field. Even books munity that acquires from education a set of that compete for adoption in a single science standards, tools, and techniques which they course differ mainly in level and pedagogic de­ later deploy in their own creative work. Yet tail, not in substance or conceptual structure. even a cursory inspection of scientific peda­ One can scarcely imagine a physicist's or gogy suggests that it is far more likely to induce chemist's saying that he had been forced to professional rigidity than education in other begin the education of his third-year class al­ fields, excepting, perhaps, systematic theology. most from first principles because its previous Admittedly the following epitome is biased to­ exposure to the field had been through books ward the American pattern, which I know that consistently violated his conception of the best. The contrasts at which it aims must, how­ discipline. Remarks of that sort are not by any ever, be visible, if muted, in European_ and means unprecedented in several of the social British education as well. sciences. Apparently scientists agree about what Perhaps the most striking feature of scien­ it is that every student of the field must know. tific education is that, to an extent quite un­ That is why, in the design of a pre-professional known in other creative fields, it is conducted curriculum, they can use textbooks instead of through textbooks, works written especially for eclectic samples of research. students. Until he is ready, or very nearly ready, Nor is the characteristic technique of text­ to begin his own dissertation, the student of book presentation altogether the same in the chemistry, physics, astronomy, geology, or bi­ sciences as elsewhere. Except in the occasional ology is seldom either asked to attempt trial re­ introductions that students seldom read, science search projects or exposed to the immediate texts make little attempt to describe the sorts of products of research done by others-to, that problems that the professional may be asked to is, the professional communications that scien­ solve or to discuss the variety of techniques that tists write for their '-peers. Collections of experience has made available for their solution. "source readings" playa negligible role in sci­ Instead, these books exhibit, fiom the very start, entijic education. Nor is the science student en­ concrete problem-solutions that the profession couraged to read the historical classics of his has come to accept as paradigms, and they then field-works in which he might encounter ask the student, either with a pencil and paper other ways of regarding the questions discussed or in the laboratory,' to solve for himself prob­ in his text, but in which he would also meet lems closely modelled in method and substance problems, concepts and standards of solution upon those through which the text has led him. that his future profession had long since dis­ Only in elementary language instruction or in carded and replaced. 3 Whitehead somewhere training a musical instrumentalist is so large or caught this quite special feature of the sciences essential a use made of "finger exercises." And when he wrote, "A science that hesitates to those are just the fields in which the object of forget its founders is lost." instruction is to produce with maximum rapid­ An ahnost exclusive reliance on textbooks is ity strong "mental sets" or Einstellungen. In the not all that distinguishes scientific education. sciences, I suggest, the effect of these techniques Students in other fields are, after all, also ex­ is much the sarne. Though scientific develop­ posed to such books, though seldom beyond ment is particularly productive of consequen­ the second year of college and even in those tial novelties, scientific education remains a 304 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE relatively dogmatic initiation into a pre-estab­ ognize numeroUS classics, some of them-like lished problem-solving tradition that the stu­ the works of Ptolemy and Copernicus or New­ dent is neither invited nor equipped to ton and Descartes-quite incompatible one evaluate. with the other. But that sarue group, if it has a The pattern of systematic textbook educa­ paradigm at all, can have only one. Unlike the tion just described existed in no place and in community of artists-which can draw simul­ no science (except perhaps elementary mathe­ taneous inspiration from the works of, say, matics) until the early nineteenth century. But Rembrandt and Cezanne and which therefore before that date a number of the more devel­ studies both-the community of astronomers oped sciences clearly displayed the special char­ had no alternative to choosing between the acteristics indicated above, and in a few cases competing models of scientific activity supplied had done so for a very long time. Where there by Copernicus and Ptolemy. Furthermore, were no textbooks there had often been uni­ having made their choice, astronomers could versally received paradigms for the practice of thereafter neglect the work which they had re­ individual sciences. These were scientific jected. Since the sixteenth century there have achievements reported in books that all the been only two full editions of the Almagest, practitioners of a given field knew intimately both produced in the nineteenth century and and admired, achievements upon which they directed exclusively to scholars. In the mature modelled their own research and which pro­ sciences there is no apparent function for the vided them with a measure of their own ac­ equivalent of an art museum or a library of complishment. Aristotle's Phys;ca, Ptolemy's classics. Scientists know when books, and even Almagest, Newton's Principia and Opticks, journals, are out of date. Though they do not Franklin's Electricity, Lavoisier's Chemistry, and then destroy them, they do, as any historian of Lyell's Geology--these works and many others science can testify, transfer them from the ac­ all served for a time implicitly to define the le­ tive departmental library to desuetude in the gitimate problems and methods of a research general university depository. Up-to-date field for succeeding generations of practition­ works have taken their place, and they are all ers. In their day each of these books, together that the further progress of science requires. with others modelled closely upon them, did This characteristic of paradigms is closely re­ for its field much of what textbooks now do lated to another, and one that has a particular for these same fields and for others besides. relevance to my selection of the term. In re­ All of the works named above are, of course, ceiving a paradigm the scientific community classics of science. As such their role may be commits itself, consciously or not, to the view thought to resemble that of the main classics in that the fundamental problems there resolved other creative fields, for exaruple the works of have, in fact, been solved once and for all. That a Shakespeare, a Rembrandt, or an Adam is what Lagrange meant when he said of New­ Smith. But by calling these works, or the ton: "There is but one universe, and it can hap­ achievements which lie behind them, para­ pen to but one man in the world's history to digms rather than classics, I mean to suggest be the interpreter of its laws.'" The example of that there is something else special about them, either Aristotle or Einstein proves Lagrange something which sets them apart both from wrong, but that does not make the fact of his some other classics of science and from all the commitment less consequential to scientific de­ classics of other creative fields. velopment. Believing that what Newton had Part of this "something else" is what I shall done need not be done again, Lagrange was call the exclusiveness of paradigms. At any time not tempted to fundamental reinterpretations the practitioners of a given specialty may rec- of nature. Instead, he could take up where the THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 305 men who shared his Newtonian paradigm had saguliers, Du Fay, N ollet, Watson, and left off, striving both for neater formulations of Franklin. All their numerous concepts of elec­ that paradigm and for an articulation that tricity had something in common-they were would bring it into closer and closer agreement partially derived from experiment and obser­ with observations of nature. That sort of work vation and partially from one or another ver­ is undertaken only by those who feel that the sion of the mechartico-corpuscular philosophy model they have chose is entirely secure. There that guided all scientific research of the day. Yet is nothing quite like it in the arts, and the par­ these common elements gave their work no allels in the social sciences are at best partial. more than a family resemblance. We are forced Paradigms determine a developmental pattern to recognize the existence of several compet­ for the mature sciences that is unlike the one ing schools and sub-schools, each deriving familiar in other fields. strength from its relation to a particular version That difference could be illustrated by com­ (Cartesian or Newtonian) of the corpuscular paring the development of a paradigm-based metaphysics, and each emphasizing the particu­ science with that of, say, philosophy or litera­ lar cluster of electrical phenomena which its ture. But the same effect can be achieved more own theory could do most to explain. Other economically by contrasting the early develop­ observations were dealt with by ad hoc elabora­ mental pattern of almost any science with the tions or remained as outstanding problems for pattern characteristic of the same field in its further research. 5 maturity. I cannot here avoid putting the point One early group of electricians followed too starkly, but what I have in mind is this. Ex­ seventeenth-century practice, and thus took at­ cepting in those fields which, like biochem­ traction and frictional generation as the £unda­ istry, originated in the combination of existing mental electrical phenomena. They tended to specialties, paradigms are a relatively late acqui­ treat repulsion as a secondary effect (in the sev­ sition in the course of scientific development. enteenth century it had been attributed to During its early years a science proceeds with­ some sort of mechanical rebounding) and also out them, or at least without any so unequivo­ to postpone for as long as possible both discus­ cal and so binding as those named illustratively sion and systematic research on Gray's newly above. Physical optics before Newton or the discovered effect, electrical conduction. An­ study of heat before Black and Lavoisier exem­ other closely related group regarded repulsion plifies the pre-paradigm developmental pattern as the fundamental effect, while still another that I shall innnediately examine in the history took attraction and repulsion together to be of electricity. While it continues, until, that is, equally elementary manifestations of electric­ a first paradigm is reached, the development of ity. Each of these groups modified its theory a science resembles that of the arts and of most and research accordingly, but they then had as social sciences more closely than it resembles much difficulty as the first in accounting for the pattern which astronomy, say, had already any but the simplest conduction effects. Those acquired in antiquity and which all the natural effects provided the starting point for still a sciences make familiar today. third group, one which tended to speak of To catch the difference between pre- and electricity as a "fluid" that ran through con­ post-paradigm scientifIc development, consider ductors rather than as an "effiuvium" that em­ a single example. In the early eighteenth cen­ anated from non-conductors. This group, in its tury, as in the seventeenth and earlier, there turn, had difficulty reconciling its theory with were almost as many views about the nature of a number of attractive and repulsive effects. 6 electricity as there were important electrical At various times all these schools made sig­ experimenters, men like Hauksbee, Gray, De- nificant contributions to the body of concepts, 306 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE phenomena, and techniques from which nized by the various earlier schools that within Franklin drew the first paradigm for electrical a generation all electricians had been converted science. Any definition of the scientist that ex­ to some view very like it. Though it did not cludes the members of these schools will ex­ resolve quite all disagreements, Franklin's the­ clude their modern successors as well. Yet ory was electricity's first paradigm, and its exis­ anyone surveying the development of electric­ tence gives a new tone and flavor to the ity before Franklin may well conclude that, electrical researches of the last decades of the though the field's practitioners were scientists, eighteenth century. The end of inter-school the immediate result of their activity was some­ debate ended the constant reiteration of funda­ thing less than science. Because the body of be­ mentals; confidence that they were on the right lief he could take for granted was very small, track encouraged electricians to undertake each electrical experimenter felt forced to more precise, esoteric, and consuming sorts of begin by building his field anew from its foun­ work. Freed from concern with any and all dations. In doing so his choice of supporting electrical phenomena, the newly united group observation and experiment was relatively free, could pursue selected phenomena in far more for the set of standard methods and phenom­ detail, designing much special equipment for ena that every electrician must employ and ex­ the task and employing it more stubbornly plain was extraordinarily small. As a result, and systematically than electricians had ever throughout the first half of the century, electri­ done before. In the hands of a Cavendish, a cal investigations tended to circle back over the Coulomb, or a Volta the collection of electri­ same ground again and again. New effects were cal facts and the articulation of electrical the­ repeatedly discovered, but many of them were ory were, for the first time, highly directed rapidly lost again. Among those lost were many activities. As a result the efficiency and effec­ effects due to what we should now describe as tiveness of electrical research increased im­ inductive charging and also Du Fay's famous mensely, providing evidence for a societal discovery of the two sorts of electrification. version of Francis Bacon's acute methodologi­ Franklin and Kinnersley were surprised when, cal dictum: "Truth emerges more readily from some fifteen years later, the latter discovered error than from confusion." that a charged ball which was repelled by Obviously I exaggerate both the speed and rubbed glass would be attracted by rubbed seal­ the completeness with which the transition to ing-wax or amber.7 In the absence of a well­ a paradigm occurs. But that does not make the articulated and widely received theory (a phenomenon itself less real. The maturation of desideratum which no science possesses from electricity as a science is not coextensive with its very beginning and which few if any of the the entire development of the field. Writers on social sciences have achieved today), the situa­ electricity during the first four decades of the tion could hardly have been otherwise. During eighteenth century possessed far more infor­ the first half of the eighteenth century there mation about electrical phenomena than had was no way for electricians to distinguish con­ their sixteenth- and seventeenth-century pre­ sistently between electrical and non-electrical decessors. During the half-century after 1745 effects, between laboratory accidents and eS­ very few new sorts of electrical phenomena sential novelties, or between striking demon­ were added to their lists. Nevertheless, in im­ stration and experiments which revealed the portant respects the electrical writings of the essential nature of electricity. last two decades of the century seemed further This is the state of affairs which Franklin removed from those of Gray, Du Fay, and even changed. B His theory explained so many­ Franklin than are the writings of these early though not all-of the electrical effects recog- eighteenth-century electricians from those of THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 307 their predecessors a hundred years before. tion that descended from it had ultimately to Some time between 1740 and 1780 electri­ be replaced by an incompatible one derived cians, as a group, gained what astronomers had from the work of Copernicus and Kepler. Nor achieved in antiquity, students of motion in the was Newton's Opticks less a paradigm for eigh­ Middle Ages, of physical optics in the late sev­ teenth-century students of light because it was enteenth century, and of historical geology in later replaced by the ether-wave theory of the early nineteenth. They had, that is, Young and Fresnel, a paradigm which in its achieved a paradigm, possession of which en­ turn gave way to the electromagnetic displace­ abled them to take the foundation of their field ment theory that descends from Maxwell. Un­ for granted and to push on to more concrete doubtedly the research work that any given and recondite problems.' Except with the ad­ paradigm permits results in lasting contribu­ vantage of hindsight, it is hard to find another tions to the body of scientific knowledge and criterion that so clearly proclaims a field of sci­ technique, but paradigms themselves are very ence. often swept aside and replaced by others that These remarks should begin to clarifY what are quite incompatible with them. We can have I take a paradigm to be. It is, in the first place, no recourse to notions like the "truth" or "va­ a fundamental scientific achievement and one lidity" of paradigms in our attempt to under­ which includes both a theory and some exem­ stand the special efficacy of the research which plary applications to the results of experiment their reception permits. and observation. More important,- it is an On the contrary, the historian can often rec­ open-ended achievement, one which leaves all ognize that in declaring an older paradigm out sorts of research still to be done. And, fmally, it of date or in rejecting the approach of some is an accepted achievement in the sense that it one of the pre-paradigm schools a scientific is received by a group whose members no community has rejected the embryo of an im­ longer try to rival it or to create alternates for portant scientific perception to which it would it. Instead, they attempt to extend and exploit later be forced to return. But it is very far it in a variety of ways to which I shall shortly from clear that the profession delayed scientIfic turn. That discussion of the work that para­ development by doing so. Would quantum digms leave to be done will make both their mechanics have been born sooner if nine­ role and the reasons for their special efficacy teenth-century scientists had been more will­ clearer still. But first there is one rather differ­ ing to admit that Newton's corpuscular view ent point to be made about them. Though the oflight might still have something significant reception of a paradigm seems historically pre­ to teach them about nature? I think not, al­ requisite to the most effective sorts of scientific though in the arts, the humanities, and many research, the patadigms which enhance re­ social sciences that less doctrinaire view is very search effectiveness need not be and usually are often adopted toward classic achievements of not permanent. On the contrary, the develop­ the past. Or would astronomy and dynamics mental pattern of mature science is usually have advanced more rapidly if scientists had from paradigm to paradigm. It differs from the recognized that Ptolemy and Copernicus had pattern characteristic of the early or pre-para­ chosen equally legitimate means to describe digm period not by the total elimination of de­ the earth's position? That view was, in fact, bate over fundamentals, but by the drastic suggested during the seventeenth century. But restriction of such debate to occasional periods in the interim it was firmly rejected together of paradigm change. with Ptolemaic astronomy, emerging again Ptolemy's Almagest was not, for example, only in the very late nineteenth century when, any less a paradigm because the research tradi- for the first time, it had concrete relevance to 308 PART IV THE HISTORICAL DEVElOPMENT OF SCIENTIFIC KNOWLEDGE unsolved problems generated by the continu­ glee ted the sort of work that even the greatest ing practice of non-relativistic physics. One scientists necessarily do most of the time. could argue, as indeed by implication I shall, My point will become clearer if I now ask that close eighteenth- and nineteenth-century what it is that the existence of a paradigm attention either to the work of Ptolemy or to leaves for the scientific community to do. The the relativistic views of Descartes, Huygens, answer-as obvious as the related existence of and Leibniz would have delayed rather than ac­ resistance to innovation and as often brushed celerated the revolution in physics with which under the carpet-is that scientists, given a par­ the twentieth century began. Advance from adigm, strive with all their might and skill to paradigm to paradigm rather than through the bring it into closer and closer agreement with continuing competition between recognized nature. Much of their effort, particularly in the classics may be a functional as well as a factual early stages of a paradigm's development, is di­ characteristic of mature scientific development. rected to articulating the paradigm, rendering Much that has been said so far is intended it more precise in areas where the original for­ to indicate that-except during occasional ex­ mulation has inevitably been vague. For exam­ traordinary periods to be discussed in the last ple, knowing that electricity was a fluid whose section of this paper-the practitioners of a individual particles act upon one another at a mature scientific specialty are deeply commit­ distance, electricians after Franklin could at­ ted to some one paradigm-based way of regard­ tempt to determine the quantitative law of ing and investigating nature. Their paradigm force between particles of electricity. Others tells them about the sorts of entities with could seek the mutual interdependence of which the universe is populated and about the spark length, electroscope deflection, quantity way the members of that population behave; in of electricity, and conductor-configuration. addition, it informs them of the questions that These were the sorts of problems upon which may legitimately be asked about nature and of Coulomb, Cavendish, and Volta worked in the the techniques that can properly be used in the last decades of the eighteenth century, and they search for answers to them. In fact, a paradigm have many parallels in the development of tells scientists so much that the questions it every other mature science. Contemporary at­ leaves for research seldom have great intrinsic tempts to determine the quantum mechanical interest to those outside the profession. forces governing the interactions of nucleons Though educated men as a group may be fas­ fall precisely in this same category, paradigm­ cinated to hear about the spectrum of funda­ articulation. mental particles or about the processes of That sort of problem is not the only chal­ molecular replication, their interest is usually lenge which a paradigm sets for the commu­ quickly exhausted by an account of the beliefs nity that embraces it. There are always many that already underlie research on these prob­ areas in which a paradigm is assumed to work lems. The outcome of the individual research but to which it has not, in fact, yet been proj ect is indifferent to them, and their interest applied. Matching the paradigm to nature is unlikely to awaken again until, as with parity in these areas often engages much of the nonconservation, research unexpectedly leads best scientific talent in any generation. The to paradigm-change and to a consequent alter­ eighteenth-century attempts to develop a ation in the beliefS which gnide research. That, Newtonian theory of vibrating strings provide no doubt, is why both historians and popular­ one significant example, and the current work izers have devoted so much of their attention on a quantum mechanical theory of solids pro­ to the revolutionary episodes which result in vides another. In addition, there is always much change of paradigm and have so largely ne- fascinating work to be done in improving the THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 309 match between a paradigm and nature in an or interpretable results were obtained. The pre­ area where at least limited agreement has al­ requisite for success proved to be a paradigm ready been demonstrated. Theoretical work on that reduced electrical action to a gravity-like proble= like these is illustrated by eighteenth­ action between point particles at a distance. century research on the perturbations that Mer Franklin electricians thought of electrical cause planets to deviate from their Keplerian action in those terms; both Coulomb and orbits as well as by the elaborate twentieth-cen­ Cavendish designed their apparatus accord­ tury theory of the spectra of complex atoms ingly. Finally, in both these cases and in all the and molecules. And accompanying all these others as well a commitment to the paradigm problems and still others besides is a recurring was needed simply to provide adequate moti­ series of instrumental hurdles. Special appara­ vation. Who would design and build elaborate tus had to be invented and built to permit special-purpose apparatus, or who would spend Coulomb's determination of the electrical months trying to solve a particular differential force law. New sorts of telescopes were re­ equation, without a quite firm guarantee that quired for the observations that, when com­ his e/fort, if successful, would yield the antici­ pleted, demanded an improved Newtonian pated fruit? perturbation theory. The design and construc­ This reference to the anticipated outcome tion of more flexible and more powerful accel­ of a research project points to the second strik­ erators is a continuing desideratum in the ing characteristic of what I am now calling attempt to articulate more powerfUl theories of normal, or paradigm-based, research. The sci­ nuclear forces. These are the sorts of work on entist engaged in it does not at all fit the preva­ which almost all scientists spend almost all of lent image of the scientist as explorer or as their time.10 inventor of brand new theories which permit Probably this epitome of normal scientific striking and unexpected predictions. On the research requires no elaboration in this place, contrary, in all the problems discussed above but there are two points that must now be everything but the detail of the outcome was made about it. First, all of the problems men­ known in advance. No scientist who accepted tioned above were paradigm-dependent, often Franklin's paradigm could doubt that there was in several ways. Some-for example, the de­ a law of attraction between small particles of rivation of perturbation terms in Newtonian electricity, and they could reasonably suppose planetary theory-could not even have been that it would take a simple algebraic form. stated in the absence of an appropriate para­ Some of them had even guessed that it would digm. With the transition from Newtonian to prove to be an inverse square law. Nor did relativity theory a few of them became differ­ Newtonian astronomers and physicists doubt ent problems and not all of these have yet been that Newton's laws of motion and of gravita­ solved. Other proble=-for example, the at­ tion could ultimately be made to yield the 0 b­ tempt to determine a law of electric forces­ served motions of the moon and planets even could be and were at least vaguely stated before though, for over a century, the complexity of the emergence of the paradigm with which the requisite mathematics prevented good they were ultimately solved. But in that older agreements being uniformly obtained. In all form they proved intractable. The men who these problems, as in most others that scientists described electrical attractions and repulsions undertake, the challenge is not to uncover the in terms of effluvia attempted to measure the unknown but to obtain the known. Their fas­ resulting forces by placing a charged disc at a cination lies riot in what success may be ex­ measured distance beneath one pan of a bal­ pected to disclose but in the difficulty of ance. Under those circumstances no consistent obtaining success at all. Rather than resembling 310 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE exploration, normal research seems like the ef­ 1J4m. Given that paradigm and the requisite fort to assemble a Chinese cube whose finished confidence in it, the scientist largely ceases to outline is known from the start. be an explorer at all, or at least to be an ex­ Those are the characteristics of normal re­ plorer of the unknown. Instead, he struggles to search that I had in mind when, at the start of articulate and concretize the known, designing this essay, I described the man engaged in it as much special-purpose apparatus and many spe­ a puzzle-solver, like the chess player. The para­ cial-purpose adaptations of theory for that task. digm he has acquired through prior training From those puzzles of design and adaptation he provides him with the rules of the game, de­ gets his pleasure. Unless he is extraordinarily scribes the pieces with which it must be played, lucky, it is upon his success with them that his and indicates the nature of the required out­ reputation will depend. Inevitably the enter­ come. His task is to manipulate those pieces prise which engages him is characterized, at within the rules in such a way that the required anyone time, by drastically restricted vision. outcome is produced. Ifhe fails, as most scien­ But within the region upon which vision is fo­ tists do in at least their first attacks upon any cused the continuing attempt to match para­ given problem, that failure speaks only to his digms to nature results in a knowledge and lack of skill. It cannot call into question the understanding of esoteric detail that could not rules that his paradigm has supplied, for with­ have been achieved in any other way. From out those rules there would have been no puz­ Copernicus and the problem of precession to zle with which to wrestle in the first place. No Einstein and the photo-electric effect, the wonder, then, that the problems (or puzzles) progress of science has again and again de­ which the practitioner of a mature science nor­ pended upon just such esoterica. One great mally undertakes presuppose a deep commit­ virtue of commitment to paradigms is that it ment to a paradigm. And how fortunate it is frees scientists to engage themselves with tiny that that commitment is not lightly given up. puzzles. Experience shows that, in almost all cases, the Nevertheless, this image of scientific re­ reiterated efforts, either of the individual or of search as puzzle-solving or paradigm-matching the professional group, do at last succeed in must be, at the very least, thoroughly incom­ producing within the paradigm a solution to plete. Though the scientist may not be an ex­ even the most stubborn problems. That is one plorer, scientists do again and again discover of the ways in which science advances. Under new and unexpected sorts of phenomena. Or those circumstances can we be surprised that again, though the scientist does not normally scientists resist paradigm-change? What they strive to invent new sorts of basic theories, such are defending is, after all, neither more nor less theories have repeatedly emerged from the than the basis of their professional way oflife. continuing practice of research. But neither of By now one principal advantage of what I these types of innovation would arise if the en­ began by calling scientific dogmatism should terprise I have been calling normal science be apparent. As a glance at any Baconian nat­ were always successful. In fact, the man en­ ural history)X"survey of the pre-paradigm de­ gaged in puzzle-solving very often resists sub­ velopment of any science will show, nature is stantive novelty, and he does so for good vastly too complex to be explored even ap­ reason. To him it is a change in the "rules of the proximately at random. Something must tell game and any change of rules is intrinsically the scientist where to look and what to look subversive. That subversive element is, of for, and that something, though it may not last course, most apparent in major theoretical in­ beyond his generation, is the paradigm with novations like those associated with the names which his education as a scientist has supplied of Copernicus, Lavoisier, or Einstein. But the THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 311 discovery of an unanticipated phenomenon can scientific theory. Mer a first paradigm has been have the same destructive effects, although usu­ achieved, a breakdown in the rules of the pre­ ally on a smaller group and for a far shorter established game is the usual prelude to signifi­ time. Once he had performed his first follow­ cant scientific innovation. up experiments, Roentgen's glowing screen Examine the case of discoveries first. Many demonstrated that previously standard cathode of them, like Coulomb's law or a new element ray equipment was behaving in ways for which to fill an empty spot in the periodic table, pre­ no one had made allowance. There was an sent no problem. They were not "new sorts of unanticipated variable to be controlled; earlier phenomena" but discoveries anticipated researches, already on their way to becoming through a paradigm and achieved by expert paradigms, would require re-evaluation; old puzzle-solvers: That sort of discovery is a nat­ puzzles would have to be solved again under a ural product of what I have been calling nor­ somewhat different set of rules. Even so readily mal science. But not all discoveries are of that assimilable a discovery as that of X rays can vi­ sort: Many could not have been anticipated by olate a paradigm that has previously guided re­ any extrapolation from the known; in a sense search. It follows that, if the normal they had to be made "by accident." On the puzzle-solving activity were altogether success­ other hand the accident through which they ful, the development of science could lead to emerged could not ordinarily have occurred to no fundamental innovations at all. a man just looking around. [n the mature sci­ But of course normal science is not always ences discovery demands much special equip­ successful, and in recognizing that fact we en­ ment, both conceptual and instrumental, and counter what I take to be the second great ad­ that special equipment has invariably been de­ vantage of paradigm-based research. Unlike veloped and deployed for the pursuit of the many of the early electricians, the practitioner puzzles of normal research. Discovery results of a mature science knows with considerable when that equipment fails to function as it precision what sort of result he should gain should. Furthermore, since some sort of at least from his research. As a consequence he is in a temporary failure occurs during almost every particularly favorable position to recognize research project, discovery results only when when a research problem has gone astray. Per­ the failure is particularly stubborn or striking haps, like Galvani or Roentgen, he encounters and only when it seems to raise qu'estions about an effect that he knows ought not to occur. Or accepted beliefs and procedures. Established perhaps, like Copernicus, Planck, or Einstein, paradigms are thus often doubly prerequisite to he concludes that the reiterated failures of his discoveries. Without them the project that goes predecessors in matching a paradigm to nature astray would not have been undertaken. And is presumptive evidence of the need to change even when the project has gone astray, as most the rules under which a match is to be sought. do for a while, the paradigm can help to deter­ Or perhaps, like Franklin or Lavoisier, he de­ mine whether the failure is worth pursuing. cides after repeated attempts that no existing The usual and proper response to a failure in theory can be articulated to account for some puzzle-solving is to blame one's talents or one's newly discovered effect. In all of these ways and tools and to turn next to another problem. If in others besides the practice of normal puz­ he is not to waste time, the scientist must be zle-solving science can and inevitably does lead able to discriminate essential anomaly from to the isolation and recognition of anomaly. mere failure. That recognition proves, I think, prerequisite That pattern-discovery through an anom­ for almost all discoveries of new sorts of phe­ aly that calls established techniques and beliefs nomena and for all fundamental innovations in in doubt-has been repeated again and again 312 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE I:

in the course of scientific development. New­ something is fundamentally wrong with the ton discovered the composition of white light theory upon which their work is based, scien­ " , when he was unable to reconcile measured dis­ tists will attempt more fundamental articula­ persion with that predicted by Snell's recently tions of theory than those which were discovered law of refraction. it The electric bat­ admissible before. (Characteristically, at times tery was discovered when existing detectors of of crisis, one encounters numerous different static charges failed to behave as Franklin's par­ versions of the paradigm theory. IS) Simultane­ adigm said they shouldn The planet Neptune ously they will often begin more nearly ran­ was discovered through an effort to account for dom experimentation within the area of recognized anomalies in the orbit ofUranus.13 difficulty, hoping to discover some effect that The element chlorine and the compound car­ will suggest a way to set the situation right. bon monoxide emerged during attempts to Only under circumstances like these, I suggest, reconcile Lavoisier's new chemistry with labo­ is a fundamental innovation in scientific theory ratory ohservations. 14 The so-called noble gases both invented and accepted . . !. were the products of a long series of investiga­ The state of Ptolemaic astronomy was, for tions initiated by a small but persistent anom­ example, a recognized scandal before Coperni­ 1,,'11 "ill '; i'l aly in the measured density of atmospheric cus proposed a basic change in astronomical ;,' i!11 1S , ,II nitrogen. The electron was posited to explain theory, and the preface in which Copernicus some anomalous properties of electrical con­ described his reasons for innovation provides a 1 19 .11:1'1I !, I," duction through gases, and its spin was sug­ classic description of the crisis state. Galileo's "e I ! I, gested to account for other sorts of anomalies contributions to the study of motion took their , " observed in atomic spectra. 16 Both the neutron point of departure from recognized difficulties and the neutrino provide other examples, and with medieval theory, and Newton reconciled ·1" ,i 'I", the list could lfe extended almost indefmitely.!7 Galileo's mechanics with Copernicanism.2o !"I In the mature/sciences unexpected novelties are Lavoisier's new chemistry was a product of the ·!IIII discovered pnncipally after somethillg has gone anomalies created jointly by the proliferation " I wrong. of new gases and the first systematic studies of "ii' I If, however, anomaly is significant in prepar­ weight relations.'! The wave theory of light "," II ing the way for new discoveries, it plays a still was developed amid growing concern about !·I,I larger role in the invention of new theories. anomalies in the relation of diffraction and po­ I" Contrary to a prevalent, though by no means larization effects to Newton's corpuscular the­ universal, belief, new theories are not invented ory.22 Thermodynamics, which later came to to account for observations that have not pre­ seem a superstructure for existing sciences, was viously been ordered by theory at all. Rather, established only at the price of rejecting the at almost all times in the development of any previously paradigmatic caloric theory. 23 advanced science, all the facts whose relevance Quantum mechanics was born from a variety

" is admitted seem either to fit existing theory of difficulties surrounding black-body radia­ well or to be in the process of conforming. tion, specific heat, and the photo-electric ef­ Making them conform better provides many fect.24 Again the list could be extended, but the of the standard problems of normal science. point should already be clear. New theories And almost always committed scientists suc­ arise from work conducted under old ones, and ceed in solving them. But they do not always they do so only when something is observed succeed, and, when they fail repeatedly and in to have gone wrong. Their prelude is widely increasing numbers, then their sectC?r of the recognized anomaly, and that recognition can scientific community encounters what I am come only to a group that knows very well elsewhere calling "crisis." Recognizing that what it would mean to have things go right. THE FUNCTION OF DOGMA IN SCIENTIFIC RESEARCH 313

Because limitations of space and time force Sdentific Revolutions, published during 1962 by the me to stop at this point, my case for dogma­ University of Chicago Press. Some of them were tism must remain schematic. I shall not here also partially developed in an earlier essay, "The Essential T enslOn: Tradition and Innovation in even attempt to deal with the fine structure SCIentific Research," which appeared in Calvin that scientific development exhibits at all times. W. Taylor (ed.), The Third (1959) Univey,ity of But there is another more positive qualification Utah Research Conference on the Identification oj of my thesis, and it requires one closing com­ Creative Sdentific Talent (Salt Lake City 1959). ment. Though successful research demands a On this whole subject see also I. B. Cohen, "Orthodoxy and Scientific Progress," Proceedings of deep commitment to the status quo, innova­ the American Philosophical Sodety, XCVI (1952) pp. tion remains at the heart of the enterprise. Sci­ 505-12, and Bernard Barber, "Resistance by entists are trained to operate as puzzle-solvers Scientists to Scientific Discovery," Science, from established rules, but they are also taught CXXXIV (1961) pp. 59&--602. I am indebted to to regard themselves as explorer and inventors Mr. Barber for an advance copy of that helpful paper. Above all, those concerned with the im­ who know no rules except those dictated by portance of quasi-dogmatic conunitments as a nature itself. The result is an acquired tension, requisite for productive scientific research should partly within the individual and partly within see the works of Michael Polanyi, particularly his the community, between professional skills on Personal Knowledge (Chicago, 1958) and The Logic the one hand and professional ideology on the ofLiberty (London, 1951). The chscussion which follows this paper will indicate that Mr. Polanyi other. Almost certainly that tension and the and I differ somewhat about what scientists are ability to sustain it are important to science's conunitted to, but that should not disguise the success. Insofar as I have dealt exclusively with very great extent of our agreement about the the dependence of research upon tradition, my issues discussed explicitly below. discussion is inevitably one-sided. On this 2. Wissenschciftliche Selbstbiographie (LeipZIg, 1948) 22, my translation. whole subject there is a great deal more to be 3. The individual sciences display some variation in said, these respects. Students in the newer and also in But to be one-sided is not necessarily to be the less theoretical sciences-e.g., parts of biology, wrong, and it may be an essential preliminary geology, and medical science-are more ltkely to to a more penetrating examination of the req­ encounter both contemporary and hIstOriCal source materials than those in, say, astronomy, uisites for successful scientific life. Almost no mathematics, or physics. one, perhaps no one at all, needs to be told that 4. Quoted in tills form by S. F. Mason, Main Cur­ the vitality of science depends on the continu­ rents of Scientijic Thought (New York, 1956) 254. ation of occasional tradition-shattering innova­ The onginal, which is identical inspirit but not in tions. But the apparently contrary dependence words, seems to derie from Delambre's contem­ porary eloge, Memoires de ... I'Institut ... , annee of research upon a deep commitment to estab­ 1812, 2nd part (Paris, 1816) p. xlvl. lished tools and beliefS receives the very mini­ 5. Much documentation for this account of electrical mum of attention. I urge that it be given more. development can be retrieved from Duane Roller Until that is done, some of the most striking and Duane H. D. Roller, The Development of the characteristics of scientific education and de­ Concept of Electric Charge: Electricity fonn the Greeks to Coulomb (Harvard Case HIstories in Expenmen­ velopment will remain extraordinarily difficult tal SCIence, VIII, Cambridge, Mass., 1954) and to understand. from I. B. Cohen, Franklin and Newton: An Inquiry into Speculative Newtonian Experimental Science and Franklin's Work in Electricity as an Example Thereof Notes (philadelprua, 1956). For analytic detail I am, however, very much indebted to a still unpub­ 1. The ideas developed in this paper have been lished paper by my student, John L. Heilbron, abstracted, in a drastically condensed form, from who has also assisted m the preparation of the the fmt third of my monograph, The Structure of three notes that follow. THOMAS KUHN

The Nature and Necessity of Scientific Revolutions

... What are scientific revolutions, and what is they have in part created. In much the same their function in scientific development? .. way. scientific revolutions are inaugurated by a [S]cientific revolutions are here taken to be growing sense, again often restricted to a nar­ those non-cumulative developmental episodes row subdivision of the scientific community, in which an older paradigm is replaced in that an existing paradigm has ceased to func­ whole or in part by an incompatible new one. tion adequately in the exploration of an aspect There is more to be said, however, and an es­ of nature to which that paradigm itself had pre­ sential part of it can be introduced by asking viously led the way. In both political and sci­ one further question. Why should a change of entific development the sense of malfunction paradigm be called a revolution? In the face of that can lead to crisis is prerequisite to revolu­ the vast and essential differences between po­ tion. Furthermore, though it admittedly strains litical and scientific development, what paral­ the metaphor, that parallelism holds not only lelism can justify the metaphor that finds for the major paradigm changes, like those at­ revolutions in both? tributable to Copernicus and Lavoisier, but also One aspect of the parallelism must already for the far smaller ones associated with the as­ be apparent. Political revolutions are inaugu­ similation of a new sort of phenomenon, like rated by a growing sense, often restricted to a oxygen or X rays. Scientific revolutions .. segment of the political community, that exist­ need seem revolutionary only to those whose ing institutions have ceased adequately to meet paradigms are affected by them. To outsiders the problems posed by an environment that they may, like the Balkan revolutions of the

Repnnted from Chapter IX of The Structure of Scientific Revolutions by Thomas Kuhn, 2nd edition (1970), pp. 92-110. © 1962, 1970 by the University of Chicago Press. Reprinted by permission of the author and University of Chicago Press.

316 THE NATURE AND NECESSITY OF SCIENTIFIC REVOLUTIONS 317 early twentieth century, seem normal parts of vital role in the evolution of political institu­ the developmental process. Astronomers, for tions, that role depends upon their being par­ example, could accept X rays as a mere addi­ tially extrapolitical or extrainstitutional events. tion to knowledge, for their paradigms were The remainder of this essay aims to demon­ unaffected by the existence of the new radia­ strate that the historical study of paradigm tion. But for men like Kelvin, Crookes, and change reveals very similar characteristics in the Roentgen, whose research dealt with radiation evolution of the sciences. Like the choice be­ theory or with cathode ray tubes, the emer­ tween competing political institutions, that be­ gence of X rays necessarily violated one para­ tween competing paradigms proves to be a digm as it created another. That is why these choice between incompatible modes of com­ rays could be discovered only through some­ munity life. Because it has that character, the thing's £lrst going wrong with normal research. choice is not and cannot be determined merely This genetic aspect of the parallel between by the evaluative procedures characteristic of political and scientific development should no normal science, for these depend in part upon longer be open to doubt. The parallel has, a particular paradigm, and that paradigm is at however, a second and more profound aspect issue. When paradigms enter, as they must, into upon which the significance of the first de­ a debate about paradigm choice, their role is pends. Political revolutions aim to change po­ necessarily circular. Each group uses its own litical institutions in ways that those institutions paradigm to argue in that paradigm's defense. themselves prohibit. Their success therefore The resulting circularity does not, of necessitates the partial relinquishment of one course, make the arguments wrong or even in­ set of institutions in favor of another, and in effectual. The man who premises a paradigm the interim, society-is not fully governed by in­ when arguing in its defense can nonetheless stitutions at all. Initially it is crisis alone that at­ provide a clear exhibit of what scientific prac­ tenuates the role of political institutions as we tice will be like for those who adopt the new have already seen it attenuate the role of para­ view of nature. That exhibit can be immensely digms. In increasing numbers individuals be­ persuasive, often compellingly so. Yet, what­ come increasingly estranged from political life ever its force, the status of the circular argu­ and behave more and more eccentrically ment is only that of persuasion. It cannot be within it. Then, as the crisis deepens, many of made logically or even probabilistically com­ these individuals commit themselves to some pelling for those who refuse to step into the concrete proposal for the reconstruction of so­ circle. The premises and values shared by the ciety in a new institutional framework. At that two parties to a debate over paradigIIl5 are not point the society is divided into competing sufficiently extensive for that. As in political camps or parties, one seeking to defend the old revolutions, so in paradigm choice-there is institutional constellation, the others seeking no standard higher than the assent of the rele­ to institute some new one. And, once that po­ vant community. To discover how scientific larization has occurred, political recouTsefails. Be­ revolutions are effected, we shall therefore cause they differ about the institutional matrix have to examine not only the impact of nature within which political change is to be achieved and oflogic, but also the techniques of persua­ and evaluated, because they acknowledge no sive argumentation effective within the quite supra-institutional fuunework for the acljudica­ special groups that constitute the community tion of revolutionary difference, the parties to of scientists. a revolutionary conflict must finally resort to To discover why this issue of paradigm the techniques of mass persuasion, often in­ choice can never be'unequivocally settled by cluding force. Though revolutions have had a logic and experiment alone, we must shortly 318 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE

examine the nature of the differences that sep­ ment would be genuinely cumulative. New arate the proponents of a traditional paradigm sorts of phenomena would simply disclose from their revolutionary successors .... We order in an aspect of nature where none had have, however, already noted numerous exam­ been seen before. In the evolution of science ples of such differences, and no one will doubt new knowledge would replace ignorance that history can supply many others. What is rather than replace knowledge of another and more likely to be doubted than their exis­ incompatible sort. tence-and what must therefore be considered Of course, science (or some other enter­ first-is that such examples provide essential prise, perhaps less effective) might have devel­ information about the nature of science. oped in that fully cumulative manner. Many Granting that paradigm rejection has been a people have believed that it did so, and most historic fact, does it illuminate more than still seem to suppose that cumulation is at least human credulity and confusion? Are there in­ the ideal that historical development would trinsic reasons why the assimilation of either a display if only it had not so often been distorted new sort of phenomenon or a new scientific by human idiosyncrasy. There are important theory must demand the rejection of an older reasons for that belief. ... Nevertheless, de­ paradigm? spite the immense plausibility of that ideal First notice that if there are such reasons, image, there is increasing reason to wonder they do not derive from the logical structure of whether it can possibly be an image of science. scientific knowledge. In principle, a new phe­ Mter the pre-paradigm period the assimilation nomenon might emerge without reflecting de­ of all new theories and of almost all new sorts structively upon any part of past scientific of phenomena has in fact demanded the de­ practice. Though discovering life on the moon struction of a prior paradigm and a consequent would today be destructive of existing para­ conflict between competing schools of scien­ digms (these tell us things about the moon that tific thought. Cumulative acquisition of unan­ seem incompatible with life's existence there), ticipated novelties proves to be an almost discovering life in some less well-known part non-existent exception to the rule of scientific of the galaxy would not. By the same token, a development. The man who takes historic fact new theory does not have to conflict with any seriously must suspect that science does not of its predecessors. It might deal exclusively tend toward the ideal that our image of its cu­ with phenomena not previously known, as the mulativeness has suggested. Perhaps it is an­ quantum theory deals (but, significantly, not other sort of enterprise. exclusively) with subatomic phenomena un­ If, however, resistant facts can carry us that known before the twentieth century. Or again, far, then a second look at the ground we have the new theory might be simply a higher level already covered may suggest that cumulative theory than those known before, one that acquisition of novelty is not only rare in fact linked together a whole group oflower level but improbable in principle. Normal research, theories without substantially changing any. which is cumulative, owes its success to the Today, the theory of energy conservation pro­ ability of scientists regularly to select problems vides just such links between dynamics, chem­ that can be solved with conceptual and instru­ istry, electricity, optics, thermal theory, and so mental techniques close to those already in ex­ on. Still other compatible relationships be­ istence. (That is why an excessive concern with tween old and new theories can be conceived. useful problems, regardless of their relation to Any and all of them might be exemplified by existing knowledge and technique, can so eas­ the historical process through which science ily inhibit scientific development.) The man has developed. If they were, scientific develop- who is striving to solve a problem defined by THE NATURE AND NECESSITY OF SCIENTIFIC REVOLUTIONS 319 existing knowledge and technique is not, cess or. That difference could not occur if the however, just looking around. He knows what two were logically compatible. In the process he wants to achieve, and he designs his instru­ of being assimilated, the second must displace ments and directs his thoughts accordingly. the first. Even a theory like energy conserva­ Unanticipated novelty, the new discovery, can tion, which today seems a logical superstruc­ emerge only to the extent that his anticipa­ ture that relates to nature only through tions about nature and his instruments prove independently established theories, did not de­ wrong. Often the importance of the resulting velop historically without paradigm destruc­ discovery will itselfbe proportional to the ex­ tion. Instead, it emerged from a crisis in which tent and stubbornness of the anomaly that an essential ingredient was the incompatibility foreshadowed it. Obviously, then, there must between Newtonian dynamics and some re­ be a conflict between the paradigm that dis­ cently formulated consequences of the caloric closes anomaly and the one that later renders theory of heat. Only after the caloric theory the anomaly lawlike .... had been rejected could energy conservation The same argument applies even more become part of science. 1 And only after it had clearly to the invention of new theories. There been part of science for some tIme could it are, in principle, only three types of phenom­ come to seem a theory of a logically higher ena about which a new theory might be devel­ type, one not in conflict with its predecessors. oped. The first consists of phenomena already It is hard to see how new theories could arise well explained by existing paradigms, and these without these destructive changes in beliefs seldom provide either motive or point of de­ about nature. Though logical inclusiveness re­ parture for theory construction. When they do mains a permissible view of the relation be­ ... the theories that result are seldom accepted, tween successive scientific theories, it is a because nature provides no ground for discrim­ historical implausibility. ination. A second class of phenomena consists A century ago it would, I think, have been of those whose nature is indicated by existing possible to let the case for the necessity of rev­ paradigms but whose details can be understood olutions rest at this point. But today, unfortu­ only through further theory articulation. These nately, that cannot be done because the view are the phenomena to which scientists direct of the subject developed above cannot be their research much of the time, but that re­ maintained if the most prevalent contemporary search aims at the articulation of existing para­ interpretation of the nature and function of sci­ digms rather than at the invention of new ones. entific theory is accepted. That interpretation, Only when these attempts at articulation fail closely associated with early logical positivism do scientists encounter the third type of phe­ and not categorically rejected by its successors, nomena, the recognized anomalies whose would restrict the range and meaning of an ac­ characteristic feature is their stubborn refusal cepted theory so that it would not possibly to be assimilated to existing paradigms. This conflict with any later theory that made pre­ type alone gives rise to new theories. Para­ dictions about some of the same natural phe­ digms provide all phenomena except anomalies nomena. The best-known and the strongest with a theory-determined place in the scien­ case for this restricted conception of a scien­ tist's field of vision. tific theory emerges in discussions of the rela­ But if new theories are called forth to re­ tion between contemporary Einsteinian solve anomalies in the relation of an existing dynamics and the older dynamical equations theory to nature, then the successful new the­ that descend from Newton's Principia. From the ory must somewhere permit predictions that viewpoint of thIS essay these two theories are different from those derived from its prede- are fundamentally incompatible in the sense 320 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE illustrated by the relation of Copernican to icant group of competent scientists immune to Ptolemaic astronomy: Einstein's theory can be attack. The much-maligned phlogiston theory, accepted only with the recognition that N ew­ for example, gave order to a large number of ton's was wrong. Today this remains a minority physical and chemical phenomena. It explained view. 2 We must therefore examine the most why bodies burned-they were rich in phlo­ prevalent objectioru to it. giston-and why metals had so many more The gist of these objections can be devel­ properties in common than did their ores. The oped as follows. Relativistic dynamics cannot metals were all compounded from different el­ have shown Newtonian dynamics to be wrong. ementary earths combined with phlogiston, for Newtonian dynamics is still used with great and the latter, common to all metals, produced success by most engineers and, in selected ap­ common properties. In addition, the phlogis­ plications, by many physicists. Furthermore, the ton theory accounted for a number of reactions propriety of this use of the older theory can be in which acids were formed by the combustion proved from the very theory that has, in other of substances like carbon and sulphur. Also, it applications, replaced it. Einstein's theory can explained the decrease of volume when com­ he used to show that predictions from Newton's bustion occurs in a confined volume of air­ equations will be as good as our measuring in­ the phlogiston released by combustion "spoils" struments in all applications that satisfy a small the elasticity of the air that absorbed it, just as number of restrictive conditions. For example, fire "spoils" the elasticity of a steel spring.' If if Newtonian theory is to provide a good ap­ these were the only phenomena that the phlo­ proximate solution, the relative velocities of the giston theorists had claimed for their theory, bodies considered must be small compared with that theory could never have been challenged. the velocity of light. Subject to this condition A similar argument will suffice for any theory and a few others, Newtonian theory seems to that has ever been successfully applied to any be derivable from Einsteinian, of which it is range of phenomena at all. therefore a special case. But to save theories in this way, their range But, the objection continues, no theory can of application must be restricted to those phe­ possibly conflict with one of its special cases. If nomena and to that precision of observation Einsteinian science seems to make Newtonian with which the experimental evidence in hand dynamics wrong, that is only because some already deals' Carried just a step further (and N ewtonians were so incautious as to claim that the step can scarcely be avoided once the first Newtonian theory yielded entirely precise re­ is taken), such a limitation prohibits the scien­ sults or that it was valid at very high relative ve­ tist from claiming to speak "scientifically" locities. Since they could not have had any about any phenomenon not already observed. evidence for such claims, they betrayed the Even in its present form the restriction forbids standards of science when they made them. In the scientist to rely upon a theory in his own so far as Newtonian theory was ever a truly sci­ research whenever that research enters an area entific theory supported by valid evidence, it or seeks a degree of precision for which past still is. Only extravagant claims for the the­ practice with the theory offers no precedent. ory-claims that were never properly parts of These prohibitions are logically unexception­ science-can have been shown by Einstein to able. But the result of accepting them would be wrong. Purged of these merely human ex­ be the end of the research through which sci­ travagances, Newtonian theory has never been ence may develop further. challenged and cannot be. By now that point too is virtually a tautol­ Some variant of this argument is quite suffi­ ogy. Without commitment to a paradigm there cient to make any theory ever used by a signif- could be no normal science. Furthermore, that THE NATURE AND NECESSITY OF SCIENTIFIC REVOLUTIONS 321 commitment must extend to areas and to de­ rived from Einsteinian, subject to a few limit­ grees of precision for which there is no full ing conditions. precedent. If it did not. the paradigm could Yet the derivation is spurious, at least to this provide no puzzles that had not already been point. Though the N:s are a special case of the solved. Besides, it is not only normal science laws of relativistic mechanics, they are not that depends upon commitment to a paradigm. Newton's Laws. Or at least they are not unless If existing theory binds the scientist only with those laws are reinterpreted in a way that respect to existing applications, then there can would have been impossible until after Ein­ be no surprises, anomalies, or crises. But these stein's work. The variables and parameters that are just the signposts that point the way to ex­ in the Einsteinian E/s represented spatial posi­ traordinary science. If positivistic restrictions tion, time, mass, etc., still occur in the N/s; and on the range of a theory's legitimate applica­ they there still represent Einsteinian space, bility are taken literally, the mechanism that time, and mass. But the physical referents of tells the scientific community what problems these Einsteinian concepts are by no means may lead to fundamental change must cease to identical with those of the Newtonian con­ function. And when that occurs, the commu­ cepts that bear the same name. (Newtonian nity will inevitably return to something much mass is conserved; Einsteinian is convertible like its pre-paradigm state, a condition in with energy. Only at low relative velocities may which all members practice science but in the two be measured in the same way, and even which their gross product scarcely resembles then they must not be conceived to be the science at all. Is it really any wonder that the sarne.) Unless we change the definitions of the price of significant scientific advance is a com­ variables in the N/s, the statements we have de­ mitment that runs the risk of being wrong? rived are not Newtonian. If we do change More important, there is a revealing logical them, we cannot properly be said to have de­ lacuna in the positivist's argument, one that rived Newton's Laws, at least not in any sense will reintroduce us immediately to the nature of "derive" now generally recognized. Our ar­ of revolutionary change. Can Newtonian dy­ gument has, of course, explained why New­ namics really be derived from relativistic dynam­ ton's Laws ever seemed to work. In doing so it ics? What would such a derivation look like? has justified, say, an automobile driver in acting

Imagine a set of statements, E 1, Ez, ... , En' as though he lived in a Newtonian universe. which together embody the laws of relativity An argument of the sarne type is used to justify theory. These statements contain variables and teaching earth-centered astronomy to survey­ parameters representing spatial position, time, ors. But the argument has still not done what rest mass, etc. From them, together with the it purported to do. It has not, that is, shown apparatus of logic and mathematics, is de­ Newton's Laws to be a limiting case of Ein­ ducible a whole set of further statements, in­ stein's. For in the passage to the limit it is not cluding some that can be checked by only the forms of the laws that have changed. observation. To prove the adequacy of New­ Simultaneously we have had to alter the funda­ toman dynamics as a special case, we must add mental structural elements of which the uni­ to the E:s additional statements like (v/c)' < 1, verse to which they apply is composed. restricting the range of the parameters and This need to change the meaning of estab­ variables. This enlarged set of statements is then lished and familiar concep ts is central to the manipulated to yield a new set, N" N 2 , . revolutionary impact of Einstein's theory. N m, which is identical in form with Newton's Though subtler than the changes from geocen­ laws of motion, the law of gravity, and so on. trism to heliocentrism, from phlogiston to Apparently Newtonian dynamics has been de- oxygen, or from corpuscles to waves, the 322 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE resulting conceptual transformation is no less back upon the science that produced them. decisively destructive of a previously established They are the source of the methods, problem­ paradigm. We may even come to see it as a pro­ field, and standards of solution accepted by any totype for revolutionary reorientations in the mature scientific community at any given time. sciences. Just because it did not involve the in­ As a result, the reception of a new paradigm troduction of additional objects or concepts, often necessitates a redefinition of the corre­ the transition from Newtonian to Einsteinian sponding science. Some old problems may be mechanics illustrates with particular clarity the relegated to another science or declared entirely scientific revolution as a displacement of the "unscientific." Others that were previously conceptual network through which scientists non-existent or trivial may, with a new para­ view the world. digm, become the very archetypes of significant These remarks should suffice to show what scientific achievement. And as the problems might, in another philosophical climate, have change, so, often, does the standard that distin­ been taken for granted. At least for scientists, guishes a real scientific solution from a mere most of the apparent differences between a dis­ metaphysical speculation, word game, or math­ carded scientific theory and its successor are ematical play. The normal-scientific tradition real. Though an out-of-date theory can always that emerges from a scientific revolution is not be viewed as a special case of its up-to-date only incompatible but often actually incom­ successor, it must be transformed for the pur­ mensurable with that which has gone before. pose. And the transformation is one that can The impact of Newton's work upon the be undertaken only with the advantages of normal seventeenth-century tradition of scien­ hindsight, the explicit guidance of the more re­ tific practice provides a striking example of cent theory. Furthermore, even if that transfor­ these subtler effects of paradigm shift. Before mation were a legitimate device to employ in Newton was born the "new science" of the interpreting the older theory, the result of its century had at last succeeded in rejecting Aris­ application would be a theory so restricted that totelian and scholastic explanations expressed it could only restate what was already known. in terms of the essences of material bodies. To Because ofits economy, that restatement would say that a stone fell because its "nature" drove have utility, but it could not suffice for the it toward the center of the universe had been guidance of research. made to look a mere tautological wordplay, Let us, therefore, now take it for granted that something it had not previously been. Hence­ the differences between successive paradigms forth the entire flux of sensory appearances, in­ are both necessary and irreconcilable. Can we cluding color, taste, and even weight, was to be then say more explicidy what sorts of differ­ explained in terms of the size, shape, position, ences these are? The most apparent type has al­ and motion of the elementary corpuscles of ready been illustrated repeatedly. Successive base matter. The attribution of other qualities paradigms tell us different things about the pop­ to the elementary atoms was a resort to the oc­ ulation of the universe and about that popula­ cult and therefore out of bounds for science. tion's behavior. They differ, that is, about such Moliere caught the new spirit precisely when questions as the existence of subatomic parti­ he ridiculed the doctor who explained opium's cles, the materiality of light, and the conserva­ efficacy as a soporific by attributing to it a tion of heat or of energy. These are the dormitive potency. During the last half of the substantive differences between successive para­ seventeenth century many scientists preferred digms, and they require no further illustration. to say that the round shape of the opium parti­ But paradigms differ in more than substance, for cles enabled them to soothe the nerves about they are directed not only to nature but also which they moved.' THE NATURE AND NECESSITY OF SCIENTIFIC REVOLUTIONS 323

In an earlier period explanations in terms had been. Therefore, while the standards of of occult qualities had been an integral part of corpuscularism remained in effect, the search productive scientific work. Nevertheless, the for a mechanical explanation of gravity was one seventeenth century's new commitment to of the most challenging problems for those mechanico-corpuscular explanation proved who accepted the Principia as paradigm. New­ immensely fruitful for a number of sciences, ton devoted much attention to it and so did ridding them of problems that had defied gen­ many of his eighteenth-century successors. erally accepted solution and suggesting others The only apparent option was to reject New­ to replace them. In dynamics, for example, ton's theory for its failure to explain gravity, Newton's three laws of motion are less a prod­ and that alternative, too, was widely adopted. uct of novel experiments than of the attempt Yet neither of these views ultimately tri­ to reinterpret well-known observations in umphed. Unable either to practice science terms of the motions and interactions of pri­ without the Principia or to make that work mary neutral corpuscles. Consider just one conform to the corpuscular standards of the concrete illustration. Since neutral corpuscles seventeenth century, scientists gradually ac­ could act on each other only by contact, the cepted the view that gravity was indeed innate. mechanico-corpuscular view of nature di­ By the mid-eighteenth century that interpreta­ rected scientific attention to a brand-new sub­ tion had been almost universally accepted, and ject of study, the alteration of particulate the result was a genuine reversion (which is not motions by collisions. Descartes announced the same as a retrogression) to a scholastic stan­ the problem and provided its first putative so­ dard. Innate attractions and repulsions joined lution. Huyghens, Wren, and Wallis carried it size, shape, position, and motion as physically still further, partly by experimenting with col­ irreducible primary properties of matter.7 liding pendulum bobs, but mostly by applying The resulting change in the standards and previously well-known characteristics of mo­ problem-field of physical science was once tion to the new problem. And Newton em­ again consequential. By the 1740s, for exam­ bedded their results in his laws of motion. The ple, electricians could speak of the attractive equal "action" and "reaction" of the third law "virtue" of the electrical fluid without thereby are the changes in quantity of motion experi­ inviting the ridicule that had greeted Moliere's enced by the two parties to a collision. The doctor a century before. As they did so, elec­ same change of motion supplies the definition trical phenomena increasingly displayed an of dynamical force implicit in the second law. order different from the one they had shown In this case, as in many others during the sev­ when viewed as the effects of a mechanical ef­ enteenth century, the corpuscular paradigm fluvium that could act only by contact. In par­ bred both a new problem and a large part of ticular, when electrical action-at-a-distance that problem's solution.6 became a subject for study in its own tight, the Yet, though much of Newton's work was phenomenon we now call charging by induc­ directed to problems and embodied standards tion could be recognized as one of its effects. derived from the mechanico-corpuscular world Previously, when seen at all, it had been attrib­ view, the effect of the paradigm that resulted uted to the direct action of electrical "atmos­ from his work was a further and partially de­ pheres" or to the leakages inevitable in any structive change in the problems and standards electrical laboratory. The new view of induc­ legitimate for science. Gravity, interpreted as tive effects was, in turn, the key to Franklin's an innate attraction between every pair of par­ analysis of the Leyden jar and thus to the emer­ ticles of matter, was an occult quality in the gence of a new and Newtonian paradigm for same sense as the scholastics' "tendency to fall" electricity. Nor were dynamics and electricity 324 PART IV THE HISTORICAL DEVELOPMENT OF SCIENTIFIC KNOWLEDGE

the only scientific fields affected by the legit­ Designing a mechanical medium to support imization of the search for forces innate to mat­ such waves was a standard problem for many of ter. The large body of eighteenth-century his ablest contemporaries. His own theory, literature on chemical affinities and replace­ however, the electromagnetic theory of light, ment series also derives from this suprame­ gave no account at all of a medium able to sup­ chanical aspect of N ewtonianism. Chemists port light waves, and it clearly made such an who believed in these differential attractions account harder to provide than it had seemed between the various chemical species set up before. Initially, Maxwell's theory was widely previously unimagined experiments and rejected for those reasons. But, like Newton's searched for new sorts of reactions. Without theory, Maxwell's proved difficult to dispense the data and the chemical concepts developed with, and as it achieved the status of a para­ in that process, the later work of Lavoisier and, digm, the community's attitude toward it more particularly, of Dalton would be incom­ changed. In the early decades of the twentieth prehensible.' Changes in the standards govern­ century Maxwell's insistence upon the exis­ ing permissible problems, concepts, and tence of a mechanical ether looked more and explanations can transform a science .... more like lip service, which it emphatically had Other examples of these nonsubstantive dif­ not been, and the attempts to design such an ferences between successive paradigms can be ethereal medium were abandoned. Scientists retrieved from the history of any science in al­ no longer thought it unscientific to speak of an most any period of its development. For the electrical "displacement" without specifying moment let us be content with just two other what was being displaced. The result, again, and far briefer illustrations. Before the chemical ,was a new set of problems and standards, one revolution, one of the acknowledged tasks of which, in the event, had much to do with the chemistry was to account for the qualities of emergence of relativity theory. to chemical substances and for the changes these These characteristic shifts in the scientific qualities underwent during chemical reactions. conununity's conception of its legitimate prob­ With the aid of a small number of elementary lems and standards would have less significance "principles"-of which phlogiston was one­ to this essay's thesis if one could suppose that the chemist was to explain why some substances they always occurred from some methodologi­ are acidic, others metalline, combustible, and so cally lower to some higher type. In that case forth. Some success in this direction had been their effects, too, would seem cumulative. No achieved. We have already noted that phlogis­ wonder that some historians have argued that ton explained why the metals were so much the history of science records a continuing in­ alike, and we could have developed a similar ar­ crease in the maturity and refinement of man's gument for the acids. Lavoisier's reform, how­ conception of the nature of science. 11 Yet the ever, ultimately did away with chemical case for cumulative development of science's "principles," and thus ended by depriving problems and standards is even harder to make chemistry of some actual and much potential than the case of cumulation of theories. The explanatory power. To compensate for this loss, attempt to explain gravity, though fruitfully a change in standards was required. During abandoned by most eighteenth-century scien­ much of the nineteenth-century failure to ex­ tists, was not directed to an intrinsically illegit­ plain the qualities of compounds was no indict­ imate problem; the objections to innate forces ment of a chemical theory. 9 were neither inherently unscientific nor meta­ Or again, Clerk Maxwell shared with other physical in some pejorative sense. There are no nineteenth-century proponents of the wave external standards to permit a judgment of that theory of light the conviction that light waves sort. What occurred was neither a decline nor must be propagated through a material ether. a raising of standards, but simply a change de- THE NATURE AND NECESSITY OF SCIENTIFIC REVOLUTIONS 325 manded by the adoption of a new paradigm. standards together, usually in an inextricable Furthermore, that change has since been re­ mixture. Therefore, when paradigms change, versed and could be again. In the twentieth there are usually significant shifts in the criteria century Einstein succeeded in explaining grav­ determining the legitimacy both of problems itational attractions, and that explanation has and of proposed solutions. returned science to a set of canons and prob­ That observation returns us to the point lems that are, in this particular respect, more from which this section began, for it provides like those of Newton's predecessors than of his our first explicit indication of why the choice successors. Or again, the development of quan­ between competing paradigms regularly raises tum mechanics has reversed the methodologi­ questions that cannot be resolved by the crite­ cal prohibition that originated in the chemical ria of normal science. To the extent, as signifi­ revolution. Chemists now attempt, and with cant as it is incomplete, that two scientific great success, to explain the color, state of ag­ schools disagree about what is a problem and gregation, and other qualities of the substances what a solution, they will inevitably talk used and produced in their laboratories. A sim­ through each other when debating the relatlve ilar reversal may even be underway in electro­ merits of their respective paradigms. In the par­ magnetic theory. Space, in contemporary tially circular arguments that regularly result, physics, is not the inert and homogenous sub­ each paradigm will be shown to satisfY more or stratum employed in both Newton's and less the criteria that it dictates for itself and to Maxwell's theories; some of its new properties fall short of a few of those dictated by its op­ are not unlike those once attributed to the ponent. There are other reasons, too, for the ether; we may someday come to know what an incompleteness of logical contact that consis­ electric displacement is. tently characterizes paradigm debates. For ex­ By shifting emphasis from the cognitive to ample, since no paradigm ever solves all the the normative functions of paradigms, the pre­ problems it defines and since no -rnro paradigms ceding examples enlarge our understanding of leave all the same problems unsolved, paradigm the ways in which paradigms give form to the debates always involve the question: Which scientific life. Previously, we had principally ex­ problems is it more significant to have solved? amined the paradigm's role as a vehicle for sci­ Like the issue of competing standards, that entific theory. In that role it functions by telling question of values can be answered only in the scientist about the entities that nature does terms of criteria that lie outside of normal sci­ and does not contain and about the ways in ence altogether, and it is that recourse to ex­ which those entities behave. That information ternal criteria that most obviously makes provides a map whose details are elucidated by paradigm debates revolutionary .... mature scientific research. And since nature is too complex and varied to be explored at ran­ dom, that map is as essential as observation and Notes experiment to science's continuing develop­ 1. Silvanus P. Thompson, Life d William Thomson ment. Through the theories they embody, par­ Baron Kelvin d Largs (London, 1910), I, pp.266-81. adigms prove to be constitutive of the research 2. See, for example, the remarks by P. P. Wiener m activity. They are also, however, constitutive of Philosophy of Sa,nce, XXV (1958), pp. 298. science in other respects, and that is now the 3. James B. Conant, Overthrow of the Phlogiston Theory point. In particular, our most recent examples (Cambridge, 1950), pp. 13-16; and]. R. Parting­ show that paradigms provide scientists not only ton, A Short History of Chemistry (2d ed.; London, 1951), pp. 85-88. The full~st and most sympa­ with a map but also with some of the directions thetIC account of the phlogiston theory's achieve­ essential for map-making. In learning a para­ ments lS by H. Metzger, Newton, Stahl, Boerhaave digm the scientist acquires theory, methods, and et al doctrine chimique (paris, 1930), Part II.