As sources of authority, I have principally in mind science textbooks, together with two other kinds of works: popular books modeled on textbooks, and philosophical writings. All three of these categories__until quite recently, apart from engaging in research, there was no other significant source of information about science.__have one thing in common. They discuss an already clarified body of problems, data, theories, and, in the most frequent cases, a set of particular paradigms to which the scientific community of the time in which they were written was committed. Textbooks themselves aim to convey the vocabulary and syntax of the scientific language of their day. Popularizations attempt to describe these same applications in language closer to that of everyday life. And philosophy of science, especially in the English-speaking world, analyzes the logical structure of that completed body of scientific knowledge.
A fuller consideration would necessarily deal with the practical differences among these three genres, but what concerns us most here is their similarity. All three record the stable outcome of past scientific revolutions, and in doing so reveal the basis of the contemporary tradition of normal science. To perform their function, such writings need not provide reliable information about how that basis was first recognized and then accepted by the professional field. At least in the case of textbooks, there are ample reasons why they will be systematically misleading on these matters.
In Section II we saw that the heavy reliance on textbooks or their equivalents is a circumstance invariably accompanying the emergence of the first paradigm in any field of science. In the final section of this essay, we shall discuss how the domination of mature science by such textbooks makes its pattern of development markedly different from that of other fields. For the moment, let us simply take for granted that, to a degree without parallel in other fields, the scientific knowledge of both laymen and specialists is based on textbooks and several other kinds of literature derived from textbooks.
But textbooks, as pedagogical instruments for the perpetuation of normal science, must be rewritten, wholly or in part, whenever the language, problem-structure, or standards of normal science change. In short, textbooks change with every scientific revolution, and once rewritten, they inevitably obscure not only the role of the revolution that produced them but even the very existence of that revolution. Unless one has personally lived through a scientific revolution in one’s own lifetime, whether one is a research scientist or a reader of textbook
literature among the general public, one’s historical sense is limited to the outcome of the most recent revolution in the field.
Thus textbooks begin by truncating the scientist’s sense of the history of his field, and then proceed to provide a substitute for what they have eliminated.
Characteristically, science textbooks treat history only in fragments, in introductory chapters or, more commonly, in scattered references to the great masters of an earlier age. From such references, students and specialists alike come to feel that they are participants in a long historical tradition. But the textbook-derived tradition in which scientists come to feel their role has in fact never existed. Science textbooks (and all too many old-style histories), for reasons both obvious and highly functional, cite only those portions of the work of past scientists that can readily be judged to have contributed to the statement and solution of the textbook’s paradigm problems.
Partly by selection and partly by distortion, scientists of earlier ages are implicitly represented as having conducted research on a fixed set of problems in accordance with just that set of fixed norms which the most recent revolution in scientific theory and method has made appear scientific. It is no surprise that the textbook and the historical tradition it implies have had to be rewritten after every scientific revolution. Nor is it strange that, as they are rewritten, science once again comes to appear largely cumulative.
Of course, scientific groups are not alone in tending to see the past of their field as developing linearly toward the privileged position of the present. The temptation to write history backward is everywhere present, and always has been. But scientists feel the temptation to rewrite history more intensely, partly because the results of scientific research do not clearly display their dependence on the historical context of inquiry, and partly because, except in times of crisis and revolution, the scientist’s position appears very secure. Whether with respect to the present or the past of science, more historical detail, or a stronger commitment to the historical details presented, would grant only an artificial status to human characteristics, errors, and confusions. Why respect what the best and most enduring efforts of science have discarded? The attitude of deprecating historical fact seems deeply and functionally embedded in the ideology of a professional field that accords the highest value to other kinds of factual items: the profession of science. When Whitehead wrote, “A science that hesitates to forget its founders is lost,” he had perceived the ahistorical disposition of the scientific community. Yet he was not entirely right, for science, like other professional activities, needs its heroes and preserves their names. Fortunately, instead of forgetting those heroes, scientists have been able to forget or revise their work.
The result is a persistent tendency to view the history of science as linear or cumulative, a tendency that leads even scientists to look back upon their own work in that way. For example, three incompatible accounts of the development of Dalton’s chemical atomism all make it appear that he had from early on been interested in chemical problems such as combining proportions, which later made him famous. In fact, however, such problems seem to have occurred to him only together with their solution, and even then only when his creative work was nearly complete.1) What Dalton’s entire account left out was a series of questions and work previously confined to physics and meteorology, the result of which was a reorientation into the field—a reorientation that taught chemists to ask new questions of old data and to draw new conclusions from old data.
Or again, Newton wrote that Galileo had discovered that the force of continuously acting gravity produces motion proportional to the square of the time. In fact, Galileo’s kinematic theorem takes that form when placed within the foundational elements of Newton’s own dynamical concepts. But Galileo said nothing of the kind. His discussion of falling bodies scarcely mentioned force itself, much less the uniform gravitational force that causes bodies to fall.2) By crediting Galileo with an answer to a question that could not even have been posed under Galileo’s paradigm, Newton’s account conceals the effect of a small but truly revolutionary reformulation, not only in the answers scientists felt they could accept, but also in the questions that were asked about motion. More than new empirical discoveries, it is precisely this kind of change in the formulation of questions and answers that explains far better the transition from Aristotle to Galileo, and from Galileo to Newtonian mechanics. By concealing such changes, the textbook tendency to make the development of science appear linear hides from view the process that occurs at the heart of the most meaningful episodes of scientific development.
The preceding examples,
each in the context of a single revolution, reveal the beginning of the reconstruction of history that is regularly completed by post-revolutionary science textbooks. But its completion involves more than the repetition of the distorted historical interpretations illustrated above. Those misinterpretations obscure revolutions, rendering them invisible. The arrangement of the still visible material in science textbooks implicitly suggests a process__if there is such a process__that denies the function of revolutions. Because a textbook aims to enable students to learn quickly what the contemporary scientific community thinks it knows, it treats the various experiments, concepts, laws, and theories of current normal science individually and as continuously as possible. As a method of teaching, this technique of presentation is beyond reproach. But when combined with the generally ahistorical disposition of scientific writing and with the recurrent systematic misinterpretations mentioned above, an overwhelmingly powerful impression inevitably follows: science has reached its present state through a series of discrete discoveries and inventions that, brought together, constitute the totality of modern technical knowledge. According to what the textbook suggests, scientists, beginning with the origins of scientific activity, have striven toward the particular goals embodied in today’s paradigms. As is often compared to laying bricks in architecture, scientists have added, one by one, another fact, concept, law, or theory to the heap of information provided in the science textbooks of their day.
But that is not the way science has developed
at all. Most of the puzzles in modern normal science did not exist until the most recent scientific revolution had been completed. Only a very few of them can be traced back, in the form in which they now arise, to the historical beginnings of science. Earlier generations investigated problems of their own, with their own instruments and their own standards of solution. It was not simply the problems that changed. Rather, the entire network-structure of fact and theory that the textbook paradigm fits to nature underwent change. For example, is the fact that chemical composition is constant merely a simple empirical fact that could have been discovered by experiment in any world in which chemists worked? Or is it an element__indeed, an unquestionably indispensable element__within a new system of related facts and theories that Dalton fitted to the whole of previous chemical experience while transforming that experience in the process? Or, by the same logic, is it an element__indeed, an unquestionably indispensable element__within a new system produced by a constant force? Or, by the same logic, is the constant acceleration produced by a constant force merely a simple fact that researchers in mechanics had always sought, or is it the answer to a question that was first raised only in the context of Newtonian theory, and that only that theory could have answered from the total information available before the question was posed?
Here, these questions, as they appear in textbooks
are raised about what appear to be fragmentary discoveries of fact. Yet these questions clearly have an implicit significance for what textbooks present as theories as well. Of course those theories “fit the facts,” but they do so only by transforming information that had long been available into facts that had not existed at all under the earlier paradigm. And this means that theories, too, do not evolve piecemeal so as to conform to facts that have always existed. Rather, theories emerge, together with the facts they fit, from a revolutionary reformulation of a prior scientific tradition, within which the knowledge-mediated relationship between scientist and nature was not the same as the new one.
One final example will make clear this account of how textbook descriptions influence our impression of scientific
development. Every elementary chemistry textbook must deal with the concept of the chemical element. Whenever this concept is introduced, its origin is almost always attributed to the seventeenth-century chemist Robert Boyle, and the careful reader will find in his work The Sceptical Chymist a definition of “element” very similar to the one used today. Mention of Boyle’s contribution helps the beginner realize that chemistry did not originate, for example, with sulfa drugs. In addition, it tells him that one of the scientist’s traditional tasks is to devise concepts of this sort. As part of the educational arsenal for training scientists, the citation of Boyle’s contribution has been highly successful. Nevertheless, it once again displays the aspect of a historical error that misleads both students and the general public about the nature of scientific activity.
According to Boyle—and he was certainly right—his “definition” of element was merely an elaboration of a traditional chemical concept. Boyle had offered this “definition” only in order to argue that there were no such things as chemical elements. Historically, the textbook interpretation of Boyle’s contribution is quite mistaken.3) Of course, no more trivial than many other misinterpretations of data, but it was a mistake. What is by no means trivial, however, is the impression of science produced when errors of this kind are compounded and, at the next stage, take their place within the technical structure of the textbook.
Like “time,” “energy,” “force,” or “particle,” the concept of element is the sort of textbook component that is neither invented nor discovered at all. Boyle’s definition in particular can be traced back at least as far as Aristotle before him, and after him through Lavoisier to modern textbooks. But that does not mean that the modern concept of element has existed since antiquity. Linguistic definitions such as Boyle’s contain little scientific content when considered in themselves. Such definitions are not perfectly logical specifications of meaning—if such things exist—but are closer to pedagogical aids. The scientific concepts to which they refer acquire their full meaning only when they are connected, within a textbook or other systematic work, with other scientific concepts, procedures, and paradigm applications.
A concept such as that of element can hardly be invented apart from its background context. Moreover, once the context is given, such concepts are already in hand and require little invention. Both Boyle and Lavoisier changed the chemical meaning of “element” in important ways. Yet they neither invented the concept anew nor changed the verbal formula used as its definition. Einstein, too, as we saw earlier, did not need to invent or explicitly redefine “space” and “time” in order to give them new meanings within the context of his research subject.
What, then, was Boyle’s historical role in the work that included that famous “definition”? By changing the relations among “element,” chemical operations, and chemical theory, he transformed the concept by means very different from those before him, and in the process he revolutionized both chemistry and the chemist’s world. Other scientific revolutions as well were necessarily required in order to give the concept its modern form and function. Yet Boyle provides a typical example both of the process accompanying each of these stages and of the changes that appear in that process when existing knowledge is embodied in textbooks. More powerfully than almost any other aspect of science, this form of education has determined our impression of the nature of science and of the role of discovery in scientific progress.
“Notes”
1) L.K. Nash, “The Origins of Dalton`s Chemical Atomic Theory,” Isis XLVII(1956), 101-16.
2) For Newton’s view, see: Florian Cajori (ed.), Sir Isaac Newton`s Mathematical Principles of Natural Philosophy and His System of the World (Berkeley, Calif., 1946), p.21. This passage should be compared with Galileo’s own discussion in the following reference: Dialogues concerining Two New Sciences, trans. H. Crew and A. de Salvio (Evanston, (III., 1946), pp. 154-76.
3) T. S. Kuhn, “Robert Boyle and Structural Chemistry in the Seventeenth Century,” Isis, XLIII(1952), 26-29.
4) Marie Boas, Robert Boyle and Seventeenth-Century Chemistry (Cambridge, 1958) deals in several places with Boyle’s positive contribution to the emergence of the concept of the chemical element.
XII. The Resolution of Revolutions
The Resolution of Revolutions
The textbooks we have just discussed are produced only after a scientific revolution has occurred.
Textbooks are the foundation of a new tradition of normal science. In raising questions about the structure of textbooks, we have clearly skipped one step. What is the process by which a new paradigm candidate comes to replace its predecessor? Whether as discovery or as theory, a new interpretation of nature first appears in the mind of one individual or a small number of individuals. It is they who first learn to see science and the world in a different way, and their ability to bring about the transition is matured by two circumstances not shared by most other members of their profession. Their attention has always been intensely focused on the problems that create the crisis. Moreover, they are usually young scholars so new to the crisis-ridden field that, compared with most of their contemporaries, they have been somewhat less strongly bound by the worldview and rules determined by the old paradigm. What can they do, and what must they do, in order to convert the entire profession or the relevant subgroup to their way of seeing science and the world? What makes that group abandon one tradition of normal research and choose another?
To understand the urgency of such questions, it is necessary to remember that these are the only means of reinterpretation that the historian of science can provide for the philosopher’s inquiry into the testing, verification, or falsification of established scientific theories. So long as he is engaged in normal science, the researcher is a solver of puzzles, not a tester of paradigms. While seeking the solution to a particular problem, the scientist may try numerous alternative approaches, avoiding those that do not produce the desired result, but in doing so he is not testing the paradigm. Rather, the scientist is much like a chess player who sets before himself a stated problem and an actual or imagined chessboard, then moves the pieces about in search of a solution. These trial exercises, whether by the chess player or by the scientist, are tests in themselves, not tests of the game and its rules. They are possible only insofar as the paradigm itself is taken for granted. Thus paradigm-testing occurs only after repeated failure to solve a notable puzzle has brought about a crisis. And even then it occurs only after the sense of crisis has produced an alternative candidate for paradigm.
In science, the testing situation does not consist, as it does in puzzles, simply in the comparison of a single paradigm with nature. Rather, verification occurs as part of a contest between two rival paradigms for the allegiance of the scientific community.
On closer examination, this formulation reveals an unexpected and perhaps significant resemblance to the two most common contemporary philosophical theories of verification. Very few philosophers of science still seek an absolute criterion for the verification of scientific theories. Noting that no theory can be subjected to all relevant tests, they ask not whether a theory has been verified, but rather about the probability of that theory in light of the evidence that actually exists. And in order to answer such a question, an influential school is led to compare the capacities of different theories to explain the evidence at hand.
This insistence on comparison among theories also characterizes the historical situation in which a new theory is accepted. Perhaps it surely points to one direction in which future discussions of verification will move.
However, in their most usual forms, probabilistic verification theories all rely on the pure or neutral observation-terms discussed in Section X. One theory of probability requires that a given scientific theory be compared with all other theories that might be thought to accord with the same heap of observational data. Another theory requires that we imaginatively construct all the tests that a given scientific theory is supposed to pass.1) Clearly, some such constructions, whether absolute or relative, are necessarily required for the calculation of particular probabilities, but some such constructions, whether absolute or relative, are difficult to know. As has already been emphasized, if there can be no linguistically or conceptually scientific or empirically neutral system, then the proposed framework of alternative tests and theories must come from one or another paradigm-grounded tradition. Limited in this way, that structure cannot have access to all possible experiences or all possible theories. Consequently, probabilistic theories conceal the situation of verification no less than they clarify it. As they emphasize, although such a situation depends on contrasts among theories and among widely known evidence, the theory and observations in question are always closely bound up with what already exists. Verification is like natural selection. It selects the fittest among the actual alternatives in a particular historical situation. Where still other alternatives remain, and where the data are of another kind, whether that selection was truly the best one that could have been made is not a useful question.
For there are no tools with which to seek an answer to it, or to questions like it.
A wholly different approach to this general network of problems was developed by Karl R. Popper, who denied outright the existence of any process of verification.2) Instead, he emphasizes falsification, that is, the importance of testing; because its result is negative, falsification drives the established theory toward inevitable rejection. Clearly, the role assigned to falsification is very similar to the role this essay has assigned to anomalous experience: experience that, by inducing crisis, prepares the way for a new theory. Nevertheless, the experience of anomaly is not to be identified with the experience of falsification. In fact, I doubt whether the latter experience of falsification exists at all. As has been repeatedly emphasized above, no theory can solve all the puzzles it faces at any given time. The solutions already obtained are also often imperfect. On the other hand, it is the indeterminacy and imperfection of the existing data-theory fit that define most of the puzzles characteristic of normal science at any given time. If any failure of such fit were always grounds for rejecting a theory, then all theories would have to be rejected at all times. If, on the other hand, a single serious failure justified the abandonment of a theory, the Popperians would need some criterion of “improbability” or “degree of falsification.” In advancing such a theory, they would almost certainly confront the same network of difficulties that has troubled the supporters of various probabilistic theories of verification.
Many of these difficulties can be avoided by recognizing that both camps, with their widespread and opposed views of the fundamental logic of scientific inquiry, have attempted to bind two broadly distinct processes into one.
Popper’s anomalous experience is important to science in that it provokes the emergence of rival candidates to the existing paradigm. But falsification, though it certainly occurs, does not occur together with, or simply because of, the emergence of an anomalous or falsifying instance. Rather, it is a subsequent and separate process, one that might better be called verification, because it consists in the victory of a new paradigm over the old one.