The preceding practice of normal science had provided more than sufficient reason to think that such problems had been, or were very nearly, solved, and this helps explain why the significance of failure is so grave when failure occurs. Failure in dealing with a new type of problem is often disappointing, but it is never surprising. Problems or puzzles, as a rule, do not yield to the first assault. Finally, these examples possess another feature that helps make striking the case concerning the role of crisis. The solution to each of them had, at least in part, been anticipated during periods when the corresponding science was not in crisis. And in circumstances in which no crisis was felt, those anticipations had been ignored.
The one complete anticipation is also the most famous case: the Copernican heliocentric system had already been proposed by Aristarchus in the third century B.C. It is often argued that if Greek science had been less deductive and less dominated by dogma, heliocentric astronomy might have begun to develop eighteen centuries earlier than it actually did.15) But that is to disregard the historical context entirely.
At the time Aristarchus made his proposal, the overwhelmingly more rational geocentric system had no deficiency that a heliocentric system might possibly have satisfied. The overall development of Ptolemaic astronomy—both its triumph and its downfall—took place several centuries after Aristarchus’s claim. Moreover, there was no evident reason to take Aristarchus seriously. Even the more elaborate theory of Copernicus was neither simpler nor more accurate than the Ptolemaic system. The observational tests available at the time, as we shall see more clearly from now on, provided no basis for choosing between the two theories. Under these circumstances, one of the factors that led astronomers to Copernican heliocentrism—and that could not have led them to Aristarchus’s system—was the sense of crisis that had been the first cause of the innovation in the first place. Ptolemaic geocentrism had failed to solve the problems of astronomy. When the time was ripe, a competing theory was given its chance. The two other examples mentioned above do not permit such complete anticipations. But certainly one reason why theories of combustion by absorption from the atmosphere—theories advanced in the seventeenth century by Rey, Hooke, and Mayow—failed to attract sufficient interest was that they had no contact with controversies already recognized in the actual practice of normal science.16) And the long neglect of relativistic criticisms of Newtonianism by scientists of the eighteenth and nineteenth centuries must have been due, in much the same way, to an immaturity of the relevant sense of problem.
Philosophers of science have steadily demonstrated that more than one theory can always be established on the basis of any given collection. The history of science shows that, especially in the early stages before a new paradigm, this is work scientists rarely undertake; and so long as the tools provided by one paradigm are proved capable of solving the problems defined by that paradigm, science operates at its highest speed and penetrates most deeply by confidently applying those tools. The reason is clear. As in productive activity, science-retooling is a kind of luxury reserved for occasions that demand it. The significance of crises lies in their indication that the occasion has arrived for changing tools.
"Notes"
1) A.R.Hall, The Scientific Revolution, 1500__1800(London, 1954), p.16.
2) Marshall Clagett, The Science of Mechanics in the Middle Ages(Madison, Wis., 1959), Parts II-III. A. Koyre presents various medieval elements in Galileo’s thought in his book Etudes Galieennes (Paris, 1939), especially in Volume I.
3) On Newton, see T.S Kuhn, "Newtons`s Optical Papers", Issac Newton`s Papers and Letters in Natural Philosophy, ed.I.B. Cohen (Cambridge, Mass., 1958), pp.27__45. On the prelude to the wave theory, see E.T.Whittaker, A History of the theories of Aether and Elextricity, I(2d ed.;London, 1951), 94__109; W.Whewell, History of the Inductive Sciences(rev.ed.;London, 1847), II, 396__466.
4) On thermodynamics, see Silvanus P. Thompson, Life of William Thomson Baron Kelvin of Largs(London, 1910), I, 266__81. On quantum theory, see Fritz Reiche, Theory, trans.H.S.Hatfield and H.L. Brose(London, 1922), chaps.i__ii.
5) J.L.E.Dreyer, A History of Axtornomy from Thales to Kelper(2d ed.; New York, 1953), chaps.xi-xii.
6) T.S.Kuhn, The Copernican Revolution(Cambridge, Mass., 1957), pp.135__43.
7) J.R.Partington, A Short History of Chemistry(2d ed.;London, 1951), pp.48__51, 73__85, 90__120.
8) Though their principal concern lies somewhat later, much relevant material is contained in the following: J.R.Partington and Douglas McKie, "Historical Studies on the Phlogiston Theory", Annals of Science, II(1937), 361__404; III(1938), 1__58, 337__71; IV(1937), 337__71.
9) H.Guerlac, Lvoisier-the Crucial Year (Ithaca, N.Y., 1961). This entire book contains material on the emergence and first recognition of the crisis. For a section that clearly describes the situation concerning Lavoisier, see p.35.
10) Max Jammer, Concepts of Space:The History of Theories of Space in Physics (Cambridge, Mass., 1954), pp114__24.
11) Joseph Larmor, Aether and Matter...INcluding a Discussion of the Influence of the Earth`s Motion on Optical Phenomena(Cambridge, 1900), pp.6__20, 320__22.
12) R.T. Glazebrook, James Clerk Maxwell and Modern Phyusics (London, 1896), chap.ix.
On Maxwell’s final attitude, see his A Treatise on Electricity and Magnetism (3d ed.;Oxford, 1892), p. 470.
13) On the role of astronomy in the development of mechanics, see Kuhn, op.cit., chap.vii.
14) Whittaker, oop.cit., I, 386__410;II(London, 1953), 27__40.
15) On Aristarchus’s work, see T.L.Heath, Arixtarchus of Samos:The Ancient Copernicus(Oxford, 1913), PartII. For an extreme statement of the traditional position that dismissed Aristarchus’s achievement, see Arthur Koestler, The Sleepwalkers: A Hostory of Man`s Changing Vision of the Universe(London, 1959), p.50.
16) Partington, op.cit., pp.78__85.
VIII. The Response to Crisis
The Response to Crisis
Then, assuming that crisis is an essential precondition for the emergence of a new theory, let us next ask how scientists respond to the existence of crisis. Part of the answer, as obvious as it is important, can be found first by noting what scientists never do when confronted with a serious and widespread anomaly. Scientists may begin to lose faith and then begin to consider other alternatives, but they do not thereby discard the paradigm that has led them into crisis. In other words, even though the meaning is valid in the philosophical sense of science, they do not regard anomalies as counterinstances. In part, this generalization is simply a statement drawn from historical fact on the basis of what has gone before. These facts suggest what will be more fully revealed by what we shall later examine concerning the abandonment of paradigms. Once a scientific theory has secured the status of a paradigm, that theory becomes invalid only when another candidate theory appears that can take its place. No process revealed by the historical study of scientific development has ever resembled the methodological framework of falsification through direct comparison with nature. This does not mean that scientists do not discard theories, nor does it mean that experience and experiment are not essential in the process by which scientists discard theories. But it does mean—and this will ultimately become the crucial point—that the act of judgment leading a scientist to reject an existing accepted theory is always based on more than a comparison of that theory with the world. The decision to reject one paradigm is always at the same time the decision to accept another, and the judgment that leads to that decision includes both a comparison of the paradigm with nature and a comparison of paradigms with one another.
In addition, there is a second reason for doubting that scientists abandon a paradigm because they encounter anomalies or counterinstances. In developing this discussion, my argument will itself exemplify the other major themes of this essay. The reasons for doubt mentioned above are purely factual. That is, they are themselves counterinstances to the epistemology that is generally believed. If my view is correct, they can at most encourage the formation of a crisis—or, more precisely, deepen a crisis that is already ripe. By themselves, they neither can nor will refute such a philosophical theory. For when confronted with anomalies, the advocates of that paradigm will, as we have already seen, behave as scientists do. They will devise various clarifications and modify their theory ad hoc in order to remove the apparent contradictions.
Many of the modifications and qualifications involved here have in fact already appeared in the literature. Therefore, if these epistemological counterinstances are to serve as more than minor irritants, it will be because they help bring about the emergence of a new and different analysis of science, one within which they are no longer troublesome. Moreover, if the typical pattern we shall later observe in scientific revolutions applies here, these anomalies will no longer appear merely as facts. From the standpoint of a new theory of scientific knowledge, they will rather seem like tautological statements of a situation that otherwise could not even have been imagined.
For example, Newton’s second law of motion, although centuries of arduous factual and theoretical research had to be traversed before it could be obtained, has often been regarded by believers in Newtonian theory as functioning like a purely logical statement that no amount of observation could ever refute.1) In Section X, we shall see that the chemical law of fixed proportion, which before Dalton had been an experimental result of exceedingly ambiguous generality, became after Dalton’s work an essential element in the definition of a compound, one that no experimental investigation could disturb.
Something similar will also happen to the generalization that scientists, when faced with anomalies or refutations, do not reject their paradigms. Scientists will be unable to reject a paradigm and still remain scientists.
Though history is unlikely to have recorded their names, there were certainly some who abandoned science because they could not accept a crisis. Like artists, creative scientists often have to be able to live in a disordered world—in another book I have called that necessity “the essential tension” inherent in scientific research.2) But abandoning science and choosing another profession is, I think, the only form of paradigm rejection that refuting facts themselves can induce. Once the first paradigm through which nature is to be interpreted has been found, there can never be research in the absence of any paradigm at all. At the same time, to discard one paradigm without replacing it with a new one is to abandon science itself. Such an act affects not the paradigm, but the person himself. Inevitably, he will appear to his colleagues as “the carpenter who blames his tools.”
This point can be restated at least as effectively. There is no such thing as research conducted in the absence of counterinstances. What distinguishes normal science from science in crisis? It is certainly not that normal science does not encounter refutations. Rather, what we earlier called the puzzles that constitute normal science exist precisely because no paradigm that serves as the framework for scientific research has ever completely solved all of its problems. The very few that seemed to solve their problems—for example, geometrical optics—soon ceased altogether to produce research problems and instead became tools in engineering. Except for those that depend entirely on instruments, any problem that normal science regards as a puzzle can, from another point of view, be seen as a refutation and thus as a source of crisis.
Copernicus called refutations what most of Ptolemy’s other successors had seen as puzzles in the agreement between observation and theory; Lavoisier saw as a refutation what Priestley had considered a puzzle successfully solved through the clarification of phlogiston theory. And Einstein regarded as refutations what Lorentz, Fitzgerald, and others had seen as puzzles in the clarification of Newtonian and Maxwellian theory. Moreover, even the existence of a crisis does not by itself transform a puzzle into a refutation. There is no sharp line of division there. Rather, by the proliferation of proposed modifications to a paradigm, a crisis loosens the rules of normal puzzle-solving in such a way that it ultimately permits the emergence of a new paradigm. I think there are only two possible ways to view the matter: either no scientific theory ever encounters refutations, or all such scientific theories are always confronted by them.
How could the situation appear otherwise? This question inevitably leads to a historical and critical elucidation of philosophy, and such topics are excluded here. Yet at least two reasons may be noted for why science seems to provide an example of the generalization that truth and falsity are uniquely and clearly determined by a confrontation between fact and statement. Normal science constantly strives, and must strive, to bring theory and fact into closer agreement, and such activity can easily be seen as a test or investigation of confirmation or refutation.
But its purpose is to solve puzzles for which the validity of the paradigm must be acknowledged precisely because those puzzles exist. Failure to obtain a solution is the scientist’s fault, not a defect in the scientific theory. A superior formulation of this point is the maxim, “A man who blames his tools is a poor carpenter.” Moreover, the way scientific pedagogy intertwines discussion of a theory with remarks on its exemplary applications has helped reinforce a confirmation-theory largely drawn from other materials. Given even the slightest reason to do so, the reader of a science textbook readily accepts those applications as evidence for the theory—that is, as reasons for believing it.
But students of science accept theories not because of evidence, but because of the authority of their teachers and textbooks.
What other means or competence could students have? The applications given in textbooks are included there not as evidence, but because learning them is part of learning the paradigm at the foundation of current practice. If the applications were presented as evidence, then the fact that textbooks restrict or fail to discuss alternative interpretations of problems for which scientists have failed to obtain paradigm solutions would make their authors seem extremely biased. There is no reason at all to make such an accusation.
Let us now return to the original question: what reaction does a scientist show when he recognizes an anomaly in the agreement between theory and nature? What we have just discussed suggests that even a very large difference compared with what has been experienced in various applications of a theory does not necessarily provoke a serious reaction. There is always bound to be some degree of discrepancy. Even the most stubborn differences usually end up being accommodated within the practice of normal research. Very often scientists are willing to wait, especially when there are many problems to be dealt with in other areas of the field. As we have already noted, for example, for sixty years after Newton’s original calculation, the predicted motion of the point at which the moon comes closest to the earth, the perigee, was left at only half the observed value. When Europe’s most outstanding mathematical physicists, no matter how hard they worked, could not resolve that obvious error, proposals naturally emerged to modify Newton’s inverse square law. But no one took these proposals very seriously, and in fact this patience with a major anomaly proved to have been correct. In 1750, Clairaut was able to demonstrate that only the mathematics of the application had been wrong and that Newtonian theory still held.3) Even in cases where a small mistake seems unlikely—perhaps because the mathematics involved is simpler or more familiar, and of a kind that works well elsewhere—a persistent and recognized anomaly does not necessarily bring about a crisis. No one seriously doubted Newtonian theory merely because it had long been recognized that predictions from that theory disagreed with two observations: the speed of sound and the motion of Mercury. The discrepancy regarding the speed of sound was eventually resolved unexpectedly by experiments on heat that had been conducted for entirely different purposes. The inconsistency concerning Mercury’s motion disappeared with the emergence of general relativity after a crisis in whose formation that theory had played no role.4) Neither appeared fundamental enough to produce the instability of a crisis. They could be acknowledged as counterinstances and still be set aside for later work.
If an anomaly is to provoke a crisis, it must usually be more than a mere irregularity. Somewhere in the paradigm-nature fit, difficulties are always lurking. Most of them are soon corrected, often by processes that could not have been anticipated in advance; they are merely matters of time. A scientist who stopped to investigate every anomaly that drew his attention would rarely accomplish any real work. We must therefore ask what makes an anomaly worth intensive investigation, and this question does not seem to have a completely general answer. The cases already examined above were characteristic, but rarely prescriptive. Sometimes an anomaly will call into question an explicit and fundamental generalization of the paradigm, as the problem of ether-drag did for those who accepted Maxwell’s theory.5) Or, as in the Copernican Revolution, even an anomaly with no obvious fundamental importance may trigger a crisis if the applications it obstructs are of special practical significance—here, for calendar making and astrology. Or, as in eighteenth-century chemistry, the development of normal science may transform an anomaly that had previously been merely troublesome into a source of crisis. The problem of weight relations took on an entirely different character after the emergence of pneumatic-chemical techniques. There are surely many other circumstances that make an anomaly especially urgent, and usually several of these factors are combined. As we have already seen, for example, one source of the crisis that confronted Copernicus was the length of time during which astronomers had labored in vain to reduce the inconsistencies remaining in Ptolemy’s geocentric system.
For these reasons, or for others like them, when an anomalous phenomenon comes to seem more than just another puzzle of normal science, the transition to crisis and to extraordinary science has begun. The anomaly itself now comes to be more generally accepted as such by the profession. More and more attention is devoted to it by many of the most eminent scholars in the field. If, as is generally not the case, it still remains unresolved, many of those scholars may come to take its solution as the foremost subject of their field of research. For them, the field will no longer look quite the same as it did before. That altered appearance is partly the result of a new point of fixation in scientific inquiry. An even more important source of change is the varied character of the numerous partial solutions made possible by concentrating attention on the problem. The earliest attacks on a stubbornly unresolved problem will follow the rules of the paradigm very closely. But as the problem continues to resist solution, attacks upon it will gradually involve minor, or not so minor, articulations of the paradigm; each of these will differ from the others, some will be partially successful, but none will be satisfactory enough to be accepted as a paradigm by the group. Through these divergent articulations, which “more and more often come to be described as ad hoc adjustments,” the rules of normal science become increasingly blurred. Though the paradigm still exists, it becomes apparent that very few of those actually engaged in research are in full agreement about it. Even the standard solutions of already solved problems come to be called into question.
In serious cases, such a situation is sometimes recognized by the scientists concerned.
Copernicus complained that the astronomers of his time had “no consistency at all in these [astronomical] investigations... so that they cannot even explain or observe the fixed length of the period of revolution.” He went on to say, “It is just as if a painter were to compose a portrait by arbitrarily joining together hands, feet, head, and other parts from various models; though each part might be excellently drawn, they would not be connected to one another as a single body, and since the parts would in no way harmonize with one another, the result would be closer to a monster than to a man.” Einstein, confined to the relatively modest language of his time, wrote only, “It was as if the ground had been pulled out from under one, with no firm foundation left on which anything could be built.”6) And Wolfgang Pauli, a few months before Heisenberg’s paper on matrix mechanics pointed the way to the new quantum theory, wrote to a friend: “At present physics is again in a state of terrible confusion. In any case, it is very difficult for me, and I wish I had become a comic actor or something of the sort, and had never heard of physics.” Compared with what Pauli said less than five months later, this testimony is truly striking: “Heisenberg’s form of mechanics has again given me hope and joy in life. Certainly it does not provide a solution to the riddle, but I am convinced that we can once more move forward.”7)
Such a clear recognition of breakdown is extremely rare, but the consequences of crisis do not depend entirely on a conscious awareness of it. What, then, can we say those consequences are? Only two of them seem to be universal. Every crisis begins with the blurring of a paradigm and, accordingly, with the loosening of the rules of normal science. In this context, research during a period of crisis comes to resemble very closely the research of the pre-paradigm period, except that in crisis research the focus of disagreement is narrower and more clearly defined. And every crisis comes to an end in one of three ways. In some cases, despite the despair of those who regarded the crisis as the end of the existing paradigm, normal science ultimately proves capable of dealing with the problem that caused the crisis. In other cases, the problem stubbornly resists even markedly radical new approaches. Then scientists may conclude that no answer will emerge in the present state of their field. Thus the problem is labeled and set aside for a future generation with more advanced tools. Or, finally, in the case that most concerns us here, a crisis may end with the appearance of a new candidate for paradigm, and with the ensuing struggle over its acceptance. This third mode of bringing a crisis to an end will be dealt with in detail later, but in order to conclude these remarks on the evolution and structure of the crisis state, we must anticipate a little of what will be discussed in the following section.