"Notes"
1) For the classic discussion of the discovery of oxygen, see A.N. Meldrum, The Eighteenth-Century Revolution in Science-the first Phase (Calcutta, 1930), chap. 5. For an indispensable recent survey, including an account of the priority controversy, see Maurice Daumas, Lavoiser, th oricien et exp rimentateur (Paris, 1995), chaps. 2__3. For a fuller account and a bibliography of related works, see T.S. Kuhn, "The Historical Structure of scientific Discovery", Science, CXXXVI (June 1, 1962), pp. 760__64.
2) For a different assessment of Scheele’s role, however, see Uno Bocklund, "A Lost Letter from Scheele to Lavoisier", Lychnos, 1957__58, pp. 39__62.
3) J.B. Conant, The Overthrow of the Phlogiston Theory:The Chemical Revolution of 1775__89 ("Harvard Case Histories in Experimental science", Case 2: Cambridge, Mass, 1950), p. 23. This very useful pamphlet includes many of the relevant evidentiary papers.
4) H. Metzger, La Philosophie de la mati re chez Lavoisier (Paris, 1935); Daumas, op.cit., chap. vii
5) The most authoritative account of the sources of Lavoisier’s doubts is found in the following book: Henry Guerlac, Lavoisier-the Crucial Year:The Back-ground and Origin of His First Experiments Combustion in 1772 (Ithaca, N.Y., 1961).
6) L.W. Taylor, Physis, the pioneer Science (Boston, 1941), pp. 790__94; T.W. Chalmers, Historic Reserches (London, 1949), pp. 218__19.
7) E.T. Whittaker, A History of the Theories of Aether and Electricity, I (2d ed.; London, 1951), 358, n. 1. Sir George Thomson informed me of another researcher who had come similarly close. Astonished by photographic plates that had been inexplicably fogged, Sir William Crookes too had been on the road to the discovery.
8) Silvanus P. Thompson, The Life of Sir William Thomson Baron Kelvin of Largs (London, 1910), II, 1125.
9) Conant, op.cit., pp. 18__20
10) K.K. Darrow, "Nuclear Fission", Bell System Technical journal, XIX (1940), 267__89. Krypton, one of the two principal fission products, appears not to have been identified by chemical methods until the reaction was sufficiently understood. The other product, barium, was almost identified chemically only in the latter part of this research, because nuclear chemists had to add that element to the radioactive solution in order to precipitate the heavy element they were seeking. The failure to separate the added barium from the radioactive product led, almost five years after the reaction had been repeatedly studied, to the following report: "As chemists we should, on the basis of these research results... change all the names in the preceding [reaction] diagram and write Ba, La, Ce in place of Ra, Ac, Th. But as 'nuclear chemists,' being closely connected with physics, we cannot allow ourselves to make such a leap, which contradicts all previous experience in nuclear physics. Perhaps a strange series of coincidences is making our results unbelievable." [Otto Hahn and Fritz Strassman, "Uber den Nachweis und das Verhalten der bei der Bestrahlung des Urans Mittels Neutronen entstehended Erdalkalimetalle", Die Natrurwissenschaften, XXXII (1939), 15].
11) The various stages in the emergence of the Leyden jar are described in I.B. Cohen, Franklin and Newton:An Inquiry into Speculative Newtonian Experimental Science and Franklin`s Work in Electricity as an Example Thereof (Philadelphia, 1956), pp. 50__52.
12) J.S. Bruner and Leo Postman, "On the Perception of Incongruilty:A Paradigm", Journal of Personality, XXIII (1949), 206__23.
13) Ibid., p. 218. My colleague Postman tells me that, even though he knew in advance exactly what was to be done and how, it was exceedingly strange to see the anomalous cards.
VII. Crisis, and the Emergence of Scientific Theories
Crisis and the Emergence of Scientific Theories
The discoveries examined in Section VI were all either causes of, or contributing factors in, paradigm change. Moreover, the changes implicitly contained within those discoveries were all destructive as well as constructive. After a discovery had been assimilated, scientists were able to account for a broader range of natural phenomena, or to give more precise accounts of some already known phenomena. But such gains were achieved by partially abandoning existing standard concepts or methods, or by simultaneously replacing such components of the earlier paradigm with others. I have argued that this type of transition is associated with all discoveries achieved through normal science. The one exception is those unsurprising discoveries that had all been predicted except in their details. But discovery is not the only source of such destructive__constructive paradigm changes. In this section we shall consider the changes that arise from the invention of new theories, which are similar to these yet ordinarily far more extensive.
As has already been discussed above, in science facts and theories, discovery and invention, are not categorically and permanently distinct; thus we may expect the contents of this section and the preceding one to overlap. (The absurd idea that Priestley first discovered oxygen and Lavoisier then invented it has a certain charm of its own. Oxygen has already been treated as a subject of discovery.
We shall soon encounter oxygen again as an invention.) In accepting the emergence of new theories, we shall inevitably broaden even our understanding of discovery. But overlap does not mean identity. The type of discoveries dealt with in the preceding section did not, at least not by themselves, produce paradigm transitions comparable to the Copernican, Newtonian, chemical, and Einsteinian revolutions. Nor, for the reason that they were more specialized, did they produce the somewhat smaller changes in paradigm caused by the wave theory of light, the mechanical theory of heat, or Maxwell’s electromagnetic theory. How can theories such as these arise from normal science, an activity less directed toward their pursuit than even toward the pursuit of discoveries?
If the recognition of anomaly plays a part in the emergence of new kinds of phenomena, it should not be surprising that a similar yet more profound recognition is a prerequisite to all acceptable changes of theory. I believe the historical evidence on this point is beyond dispute. The situation of Ptolemaic astronomy was a scandal before Copernicus’s declaration.1) Galileo’s contributions to the study of motion were closely connected with the difficulties revealed in the scholastic critique of Aristotle’s theory.2) As for Newton’s new theory of light and color, it originated in the discovery that none of the existing pre-paradigm theories could account for the length of the spectrum; and the wave theory that replaced Newton’s theory was announced amid a growing concern with anomalous phenomena in relating the effects of diffraction and polarization to Newton’s theory.3) Thermodynamics was born from the collision of two existing nineteenth-century theories of physical science, and quantum mechanics from various difficulties surrounding black-body radiation, specific heats, and the photo-electric effect.4) Moreover, in every case except that of Newton’s theory, the recognition of anomaly was so prolonged and penetrated so deeply that it is appropriate to describe the affected fields as having been in a state of mounting crisis. Because it requires a large-scale destruction of paradigms and major changes in the problems and techniques of normal science, the emergence of new theories is generally preceded by a period of pronounced professional insecurity. As anyone might predict, such insecurity arises from the persistent failure of the puzzles of normal science to be solved as they should. And the failure of existing rules becomes the prelude to the search for new ones.
Let us first examine the birth of Copernican astronomy, a particularly famous case of paradigm change. When the Ptolemaic geocentric system, its predecessor, was first developed from the 2nd century BC through the 2nd century AD, it was astonishingly successful in predicting the changing positions of the stars and planets. No other ancient system fit so well. With regard to the stars, Ptolemaic astronomy is still widely used today as an engineering approximation, and the Ptolemaic system's predictions for the planets matched Copernicus's just as well. However, as a scientific theory, fitting astonishingly well by no means signifies complete success. For two things—the positions of the planets and the precession of the equinoxes—the predictions based on the Ptolemaic system did not agree well with the best observations obtained at the time. The attempt to reduce these slight discrepancies further emerged as the main task of the normal astronomical research conducted by Ptolemy's successors, and this state of affairs was analogous to the way attempts to reconcile celestial observation with Newtonian theory provided regular research topics for Newton's eighteenth-century successors. For some time, astronomers had sufficient reason to regard such attempts as no less successful than past attempts within the Ptolemaic system. Whenever a particular contradiction appeared, astronomers could smoothly eliminate the point of contradiction by making some special adjustment in the Ptolemaic system of compounded circles. But as time passed, they could perceive that the complexity of astronomy was increasing far more rapidly than its accuracy, and that it was commonplace for a discrepancy corrected in one place to appear elsewhere.5) The astronomical tradition suffered constant interference from without, and because the exchange of views among astronomers was limited in the absence of printing, these difficulties were recognized only very slowly. Yet finally they were realized. Around the thirteenth century, Alfonso X declared that had God consulted him when creating the universe, God would have received excellent advice. Entering the sixteenth century, Copernicus's colleague Domenico da Novara claimed that a clumsy and inaccurate system such as the one Ptolemaic theory had developed could never be a science of nature. And Copernicus himself wrote in the preface to De Revolutionibus that the astronomical tradition he inherited had ultimately created nothing but a monster. By the early sixteenth century, gradually more and more of Europe's foremost astronomers were realizing that astronomy's paradigm was failing to do its job when applied to its own traditional problems. Such recognition was the prerequisite required for Copernicus to reject the Ptolemaic paradigm and begin seeking a new one. His famous preface remains to this day one of the classical descriptions of a crisis situation.6) Of course, the breakdown of normal technical puzzle-solving activity was not the sole element of the astronomical crisis Copernicus encountered. Extending the discussion, social pressure for calendar reform—that is, the pressure factor that drove the puzzle of precession into a particularly urgent problem—was also at work.
Furthermore, to provide a more complete explanation, one would have to consider the medieval critique of Aristotelianism, the flourishing of Renaissance Neoplatonism, and other significant historical elements. Yet technical breakdown must still remain at the core of the crisis. In a mature science—and astronomy had already been such a science since antiquity—external factors such as those mentioned above are important in determining the timing of the breakdown, the ease with which the breakdown can be perceived, and the area in which the breakdown first occurs by attracting special interest. Despite their importance, topics of this kind are not addressed in this essay.
If these points are clear in the case of the Copernican revolution, let us now turn to a second, relatively different example: the crisis that preceded the birth of Lavoisier's oxygen theory of combustion. In the 1770s, a combination of factors produced a crisis in chemistry, and historians of science have not reached consensus on the nature or relative importance of those factors. Yet two of them are generally acknowledged to be of first-rate importance: the flourishing of pneumatic chemistry and doubts regarding weight relationships. The history of pneumatic chemistry begins with the development of the air pump in the seventeenth century and its deployment in chemical experimentation. During the eighteenth century, as the use of air pumps and various pneumatic devices became widespread, chemists gradually came to realize that air was undoubtedly an active component in chemical reactions. Yet with a few exceptions—so ambiguous that they might not even be exceptions at all—chemists maintained the belief that air was the only kind of gas.
By 1756, when Joseph Black showed that fixed air (carbon dioxide) is always distinct from ordinary air, the two gas samples were considered to differ only in their impurities.7)
After Black's research, the study of gases advanced most strikingly at the hands of Cavendish, Priestley, and Scheele, all of whom developed various new techniques to distinguish gas samples one by one. These scholars, from Black to Scheele, all adhered to the phlogiston theory and frequently applied it in their experimental apparatus and interpretation of results. Scheele was in fact the first to obtain oxygen through a series of elaborate experiments designed to remove phlogiston by means of heat. Nevertheless, because the results obtained in those experiments were samples and properties of gases of exceedingly troublesome variety, the phlogiston theory gradually proved to have little chance of being solved by laboratory work.
Although none of these chemists proposed that the phlogiston theory should be replaced, they could not apply it consistently. By the early 1770s, when Lavoisier began his experiments on air, there were about as many modifications of the phlogiston theory as there were pneumatic chemists.8) That so many modifications of a theory should arise is one symptom very commonly displayed during a state of crisis. Copernicus, too, had complained of it in his preface.
But the increasing ambiguity and decreasing utility of the phlogiston theory for pneumatic chemistry was not the sole source of the crisis that confronted Lavoisier. He also took a profound interest in the problem of explaining the increase in weight that appears when most bodies are burned or calcined, and this too was a problem with a long history. At least some Islamic chemists had known that certain metals increase in weight when heated. In the seventeenth century, some researchers concluded from this very fact that calcined metals absorb some component from the atmosphere. However, such seventeenth-century conclusions seemed unnecessary to most chemists. If chemical reactions can inevitably alter the volume, color, and structure of components, why should weight alone not change in a chemical reaction? Weight was not always treated as a measure of the quantity of matter. Moreover, the weight increase accompanying combustion was dealt with as a separate phenomenon. Natural objects (e.g., wood), as the phlogiston theory later naturally had to explain, mostly decreased in weight when burned.
Yet over the course of the eighteenth century, this early response to the problem of weight-gain gradually became difficult to sustain. Partly because the balance came into widespread use as standard chemical equipment, and partly because the development of pneumatic chemistry made it possible and desirable to preserve the gaseous products of reactions, chemists increasingly discovered more cases in which weight-gain occurred during combustion. At the same time, the gradual assimilation of Newton's theory of gravitation led chemists to assert that weight-gain undoubtedly signified an increase in the quantity of matter. Yet such conclusions did not lead to the abandonment of the phlogiston theory, because that theory could be adjusted in various ways. Perhaps phlogiston possessed negative weight, or perhaps when phlogiston departed from a body, fire particles or something else entered the burning body. Besides these, various other explanations appeared.
The problem of weight-gain did not lead to the overthrow of the phlogiston theory, but it did gradually increase the number of special researches in which the problem figured prominently. One of these was a paper "On Phlogiston Considered as a Substance Possessing Weight and Analyzed by the Change in Weight It Causes in Bodies to Which It Binds," presented at the French Academy in early 1772—the very year in which Lavoisier sent his famous sealed note to the president of the Academy. Before that note was written, one problem that had attracted extraordinary attention from chemists for many years was highlighted as a conspicuous unsolved puzzle.9) The phlogiston theory was being diversely modified and transformed to satisfy that problem. Like the problems in pneumatic chemistry, the problems of weight-gain were making the identity of the phlogiston theory increasingly difficult to understand. Although it had been believed and respected as a tool that fit well until then, the paradigm of eighteenth-century chemistry was gradually and progressively losing its unrivaled status. Consequently, the research it directed came to resemble increasingly the research conducted under the rivalry of various schools in the pre-paradigm era—a phenomenon that is another typical spectacle of crisis.
Now, as a third and final example, let us consider the late nineteenth-century crisis in physics that opened the way to the birth of relativity theory. One root of this crisis dates back to the late seventeenth century, when numerous natural philosophers, among whom Leibniz was the most severe, criticized Newton for insisting on revising and maintaining the classical concept of absolute space.10) They could prove, very closely if not completely, that absolute position and absolute motion served no function whatever in Newton's system. And they succeeded in suggesting the considerable aesthetic charm that a perfectly relative concept of space and motion would later develop. Yet their criticism was purely logical. Like the early Copernicans who criticized Aristotle's proof of the immobility of the earth, these natural philosophers never imagined that a shift to a relativistic system would yield new results in observation. They never related their views, at any point, to any problem that had arisen in applying Newtonian theory to nature. As a result, their views disappeared along with them within a few decades of the early eighteenth century, only to be revived after several decades of the late nineteenth century when they became related in an entirely new way to the actual practice of physics.
The technical problems with which the relativistic philosophy of space would ultimately become involved did not provoke any crisis until the 1890s, but in fact they had begun to enter normal science after 1815 with the acceptance of the wave theory of light. If light was a wave motion propagated through a mechanical ether governed by Newton’s laws, then both celestial observations and terrestrial experiments made it possible to detect a flow through the ether. In the case of astronomical observations, only observations of aberration could promise sufficient accuracy to provide relevant information, and accordingly the detection of ether drift through measurements of aberration emerged as a recognized problem in normal scientific research. And a very special apparatus was devised in order to solve it. But since that apparatus failed to detect any observable drift, the problem came to be transferred from experimenters and observers to theorists. During the middle decades of the nineteenth century, a number of scholars, including Fresnel and Stokes, proposed various articulated modifications of ether theory designed to explain the failure to observe such a drift. Each of these articulations assumed that moving bodies carried some amount of the ether along with them. And each explanation gave a fairly plausible account of the failure to detect drift not only in celestial observations but also in terrestrial experiments, including the famous experiment of Michelson and Morley.11) Except for what arose among the various modified theories, there was as yet no contradiction. In the absence of related experimental techniques, such contradictions never appeared sharply.
The situation changed during the last two decades of the nineteenth century with the gradual acceptance of Maxwell’s electromagnetic theory. Maxwell himself was a Newtonian who believed that light and electromagnetism generally arose from the nonuniform displacement of particles in a mechanical ether. Maxwell’s earliest views on the theory of electricity and magnetism made direct use of the hypothetical properties he assigned to this medium. Those views were dropped from his final version. Yet he still believed that his electromagnetic theory was compatible with some articulation of Newton’s mechanical view.12) The task of developing an appropriate articulation was a challenge to him and to his successors. But, as has always been the case in the development of science, it proved enormously difficult to obtain the articulation actually required. Despite its author’s optimism, just as the proposal of Copernican astronomical theory had brought about a mounting crisis for existing theories of motion, Maxwell’s theory too, despite its Newtonian origins, ultimately created a crisis for the paradigm from which it had derived.13) Moreover, the focus at which the crisis intensified most severely was concentrated on the very problems we have just been considering: problems of motion with respect to the ether.
In Maxwell’s discussion of the electromagnetic behavior of moving bodies, there was no mention of ether drag, and it proved very difficult to introduce such drag into his theory. Thus the earlier series of observations intended to detect drift through the ether all became anomalies.
Therefore, after 1890, for several years long and persistent attempts were made, both experimental and theoretical, to detect motion relative to the ether and to introduce ether drag into Maxwell’s theory. Some analysts regarded their results as ambiguous, and experiments to detect drift were uniformly unsuccessful. In the case of theorists attempting to connect Maxwell’s theory with ether drag, a number of encouraging beginnings were made—especially in the work of Lorentz and FitzGerald—but they too still produced a proliferation of competing theories that were found to be other, subsidiary phenomena.14) It was against this historical current that Einstein’s special theory of relativity appeared in 1905.
These three examples are almost entirely typical. In each case, a new theory appeared only after a conspicuous failure in the activity of normal problem-solving. Moreover, except in the case of Copernicus, where extra-scientific factors played a particularly large role, the collapse and the aspects of the theories that signaled it occurred a decade or two before the announcement of the new theory. The new theory appears to be a direct response to crisis. Also note, though this may not be quite so typical, that the problems in which the breakdown occurred were all of a form that had been recognized for many years.