Copyright (c) 1999 by the Massachusetts Institute of Technology. All rights reserved.

Perspectives on Science 6.3 (1998) 209-231

 

 

 

Studying the Study of Science Scientifically

David L. Hull

 

Abstract: Testing the claims that scientists make is extremely difficult. Testing the claims that philosophers of science make about science is even more difficult, difficult but not impossible. I discuss three efforts at testing the sorts of claims that philosophers of science make about science: the influence of scientists' age on the alacrity with which they accept new views, the effect of birth order on the sorts of contributions that scientists make, and the role of novel predictions in the acceptance of new scientific views. Without attempting to test philosophical claims, it is difficult to know what they mean.

 

In the summer of 1996, David Bloor (1997, p. 498) was dismayed to hear a philosopher argue before an international audience that, according to Kuhn, "communication between scientists in different, incommensurable paradigms is impossible, but such communication is indeed possible--and here examples were provided--so Kuhn is wrong." In response, Bloor (1997, p. 499) argues that Kuhn was "not equating incommensurability with incommunicability, though he was saying that it can attenuate communication and may demand an effort to overcome." I do not know whether the philosopher in question was as obtuse as Bloor claims, but he or she was doing something right--attempting to test claims about science made by students of science. If Kuhn is right about paradigms, one would expect that scientists working in the same paradigm should be able to communicate with each other more easily than scientists working in different paradigms. This claim can actually be tested, and the results of this test might reflect positively or negatively on Kuhn's famous incommensurability [End Page 209] theses--and without equating incommensurability with incommunicability.

In this paper I urge more students of science, especially philosophers of science, to test their claims about science as often and as rigorously as possible. To perform this task, science must be distinguished from the study of science. History of science poses one set of problems. In certain respects it is like all other historical disciplines, such as cosmology, but human history provides all sorts of other problems in addition. Historians commonly claim that human histories are as much literature as they are science. Sociologists of science are put in the position of having to study themselves. After all, sociology of science is part of sociology, and sociology is itself a science. But of all the disciplines that make up science studies, philosophy of science is furthest removed from science itself. In this paper I concentrate on the hardest case of all--philosophy of science.

Those of us engaged in science studies study science, but what sort of thing is science? In the past, philosophers of science have taken science to be something like a natural kind, about which general knowledge is possible. On this view science is a certain sort of activity distinguishable from other human activities by characteristics such as serious attempts at falsification. At the other extreme, current advocates of pluralism claim that science is merely a "loose and heterogeneous collection of more or less successful investigative practices" (DuprŽ 1990, p. 69). In this paper I treat science neither as a natural kind nor as a loose heterogeneous collection, but as an "historical entity," an entity that is variable both through time and at any one time but which retains sufficient cohesiveness and continuity in the face of all this variability.

After numerous false starts, science dribbled into existence and, once established, continues to the present. To be sure, it cannot be characterized in terms of universally unvarying processes, but it exhibits much more coherence and continuity than DuprŽ would allow. At any one time, science is variable. For example, some cognitive psychologists allow reference to introspection while behavioral psychologists do not. Science also varies through time. While reference to God was acceptable in the early years of science, it is not acceptable now. Furthermore, science has not yet reached its definitive state today. It is very likely to continue changing in the future, possibly in ways that few of us would like. For example, political correctness may supersede empirical truth as a criterion for scientific acceptance. If so, my interest in science will decrease markedly. But, as variable as science has been at any one time as well as through time, it is not a hodgepodge. It exhibits enough coherence and continuity to be considered an historical entity. [End Page 210]

Regardless of the way in which one construes science, it still can be studied scientifically. If science is a kind, possibly a natural kind, then one task of those of us who study science might well be to discern the regularities that characterize it. If science is an historical entity, we can still study its variability at any one time and its modifications through time. Finally, if science is just a variable and heterogeneous collection of investigative practices, then all we can do is to chronicle its vicissitudes. Thus, on any of these interpretations, science can be studied scientifically because scientific understanding encompasses both the specification of lawful regularities and the description of particular sequences of events. However, if we are to study science scientifically, we must meet the same standards as those that govern science itself. As Sulloway and others have remarked, too often those of us who study science tend to employ methodological practices that we would condemn as inadequate in the work of those scientists whom we study.

In recent years, science studies has been characterized by a welter of conflicting schools, philosophies, and methodologies. According to many students of science, much of what we commonly take to be important in science consists in little more than public relations ploys. Scientists do not discover facts; they construct, fabricate, and invent them. To the extent that these expressions were devised to emphasize the active role that scientists, both individually and in groups, play in the generation of knowledge claims, they are unproblematic; but I personally can understand why many readers, including scientists, take these claims to imply much more, as if nothing that can be legitimately termed "nature" or "the world" in any way constrains what scientists come to believe. If any of the advocates of these philosophies actually hold any of these more extreme positions, they will surely reject my call for students of science to test their beliefs about science as one more instance of self-delusion. If all the time that scientists expend testing their views is just so much show, then surely any attempt by those of us who study science to test our views about science scientifically is even more of an illusion.

The relation between science and philosophy of science is usually characterized in terms of "levels." Science is related to philosophy of science in the same way that ethics is related to metaethics. Philosophy of science is one level "higher" than science. The relation between these levels is problematic. For example, logical empiricist philosophers of science insist that empirical data play a necessary role in science. However, they need not on this account be committed to the view that empirical data play any role whatsoever in philosophy of science. The Covering-Law Model of Scientific Explanation might be a totally adequate analysis of "scientific explanation" [End Page 211] even if no scientist ever explained any phenomenon by deriving it from a law of nature. The central role of evidence in science entails nothing about the importance of evidence in our study of science. However, this difference between science and the study of science does, to say the very least, call for an explanation. In this paper I see to what extent the sort of testing that goes on in science can profitably be extended to the study of science. Traditionally, historians and sociologists of science have acknowledged the role of data in their work. Philosophers should do the same. Not only should we use the data gathered by historians and sociologists of science, but we should also generate a bit of this data ourselves. In doing so, we will deepen our understanding of both science and philosophy of science.

One message that both history and sociology of science have to teach us is that testing is difficult. These difficulties have several sources. One difficulty arises from the intricate interconnections between our basic theories and anything that might count as empirical data. It is simply not true that we know what our theories mean independently of any attempt to test them. We may think that we understand a particular theory, but this illusion can be quickly dispelled by attempting to test it. No amount of conceptual analysis can serve as a substitute for empirical investigation--even with respect to meaning. Empirical investigation requires the operationalizing of the concepts we are attempting to apply. Philosophers have contributed to our understanding of science by showing that theoretical terms cannot be operationally defined in a literal sense of "definition," but this impossibility proof is not enough. We still need some discussion of how to operationalize our concepts. What difficulties are we likely to meet? Can they be overcome?

Testing the sorts of claims that scientists make is difficult enough. Testing the sorts of meta-level claims made by those of us who study science is even more difficult. Even so, we must try. In this paper I investigate three attempts to study science scientifically: my own study of the role of age in determining how quickly scientists accept new ideas (Hull, Tessner, and Diamond 1978), Sulloway's (1996) attempt to discover the influence of birth order on the sorts of contributions that scientists are likely to make to science, and Donovan, Laudan, and Laudan's (1988) investigation of the role of novel predictions in science. By discussing these three examples, I intend to show how general claims about science can be tested and to illustrate how the sorts of difficulties that arise in such attempts can be overcome. They are overcome in the same ways as they are in science in general. By emphasizing the role of testing via data, I do not mean to imply that this is all there is to science. Obviously, science is much more than empirical testing. However, such testing is an important aspect of science. Also, in emphasizing the need for philosophers of science [End Page 212] to attempt to test their views about science, I do not mean to imply that this is all that philosophers of science should do. Occasional testing does not entail nothing but testing.

 

Planck's Principle

In his autobiography, Max Planck (1949, p. 33) remarked that a "new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die and a new generation grows up that is familiar with it." Instance after instance can be given of scientists, who, like Planck, were involved in a basic conceptual revolution in science, claiming that older scientists are less willing to adopt new views than their younger colleagues.1 This claim is important because it casts doubt on the role of reason, argument, and evidence in bringing about change in science. Good evidence should be just as good for young scientists as for their older colleagues. If older scientists are more resistant to new ideas than are younger scientists, it might be due to the biological or psychological effects of aging, to differences in the stage of a scientist's career, to professional commitments to certain views, or to greater knowledge of the effects of this change, and so on (for discussion of the role of age in science, see McCann 1978; Messeri 1988; Rappa and Debackere 1993; and Levin, Stephan, and Walker 1995).

The investigation of Planck's principle raises two questions: what does it mean? and is it true? Until we begin to decide what Planck means, we cannot begin to decide whether or not Planck was right. What counts as a "new scientific truth?" How "new" must it be? To answer these questions, [End Page 213] I turn to two examples of scientific revolutions: the Darwinian revolution and the Mendelian revolution. Inevitably, when scientists publish what they take to be new scientific ideas, historians will find all sorts of precursors. For example, Matthew (1831) anticipated Darwin with respect to natural selection, and numerous authors, including Lamarck and Robert Chambers, anticipated him with respect to evolution. Why then term the revolution that took off in 1859 the Darwinian revolution? Why not the Matthewian revolution? The answer is that Matthew did not produce a revolution of any kind. His allusions to what came to be known later as "natural selection" went totally unnoticed at the time. Authors such as Lamarck and Chambers had some impact with respect to the transmutation of species (e.g., on Wallace), but neither succeeded in producing anything like a "revolution." Darwin did.

The Mendelian revolution poses a different set of problems. Was Mendel the author of the Mendelian revolution or was he simply a precursor like Matthew, Lamarck, and Chambers? Although Mendel's 1865 paper did not go totally unnoticed, it certainly did not initiate anything like a revolution in the study of heredity. The Mendelian revolution did not occur until the turn of the century when others came up with views similar (though far from identical) to those of Mendel. These authors claimed independent discovery. If so, then Mendel's paper played no role in the "rediscovery" of Mendelian genetics. Then why call this revolution "Mendelian?" One answer is that Mendel's paper includes a careful and clear exposition of the theory that eventually came to bear his name. It was not just a vague sketch. Another reason is that it served to stave off a priority dispute. The two junior rediscoverers were not about to have this theory named after them. They were able to sabotage any potential effort by the powerful de Vries to get his name attached to this emerging field by emphasizing the role of Mendel. Better Mendelian genetics than de Vriessian genetics. The sense of novelty that is relevant to Planck's Principle concerns reception, not first discovery or even first publication. Until scientists notice a new idea, they cannot accept or reject it.

What counts as a scientific truth, and must it be true? Theologians, art critics, wine connoisseurs, and hosts of others may or may not be resistant to change, but Planck's principle applies only to scientists. Who are we to count as scientists? In evaluating Planck's principle, are we to count scientists as they were conceived of at the time or as we conceive of them today? After all, who is or is not considered a scientist has changed during the course of science. For example, Descartes, and later Newton, were engaged in roughly the same array of activities. They were engaged in what we today would call philosophy, mathematics, and science, not to mention theology, alchemy, and other questionable practices, but the preceding [End Page 214] distinctions were not made at the time the way that we make them today. Nowadays, college students are likely to come across Descartes in an introductory philosophy course, while they hear about Newton as a scientist. Why is this so? The reason is obvious: Descartes generated what we would now call a philosophical research program (only specialists pay any attention to Descartes's science), while Newton generated a successful scientific research program (again, only specialists are much interested in Newton's philosophy). Hence, today, Descartes is usually characterized as a philosopher and Newton as a scientist. But such presentist ascriptions are sorely misleading.

I see no easy way out of the dilemma posed by the evolution of language. All of us are saddled with the requirement that we write for present-day readers and must use present-day language to do so, even if we know that present-day language is sure to mislead. The best we can do is to warn our readers of the most obvious and important instances of possible misunderstandings. After all, publishing a book on sixteenth century Italian science in sixteenth century Italian would be a waste of time. Those who could make the most sense of such a book have long been dead. Even the most anti-presentist historians now see the need for what they term "legitimate anachronisms" (Lightman 1997, p. 10). However, we must constantly remind ourselves that calling Descartes a philosopher and Newton a scientist in the absence of such explanations can lead to serious misunderstandings.

With respect to Planck's Principle and the reception of Darwin's theory of evolution, we cast our net broadly. We included anyone who was treated as a scientist between 1859 and 1869 by their contemporaries, even if we had our doubts. Happily, we may term these workers "scientists" because William Whewell first coined this term in 1833 and published it in 1834, only to reject it as inappropriate (Whewell 1834). It caught on anyway but not until a half century later. Why we are permitted to use a term only after it was introduced--even if rejected by its author and by the relevant audience for half a century--remains, for me, one more historiographic mystery. In any case, we were well aware that we were not sampling the scientific community at large. Those scientists who were most likely to pass judgment on evolutionary theory were those most closely connected to it conceptually, e.g., zoologists, botanists, geologists, those physicists working on the age of the earth, etc.

In addition, well-known scientists are likely to leave more extensive records than their lessor-known brethren. This second limitation might seem on the surface not to be of great significance, except for the work of Desmond (1989) and Kim (1994) with respect to the reception of Darwinian evolution and Mendelian genetics, respectively. They discovered that [End Page 215] second- and third-tier scientists reacted quite differently from their more prominent colleagues. Although Darwin valued the opinion of Charles Lyell over that of an obscure pigeon fancier, Planck clearly intended to be talking about scientists, not just the scientific elite (for discussions of the elitist nature of science, see Meltzer 1956; Price 1963; Cole and Cole 1973; Yoels 1973; Reskin 1977; Ladd and Lipset 1977; Garvey 1979; and Hull 1988).

Are older scientists resistant only to new truths? Planck's principle is usually raised only with respect to new scientific ideas that we now take to be true or at least significant improvements over past beliefs. How about all the new ideas that are rejected, and, in retrospect, we think rightly so? Rarely does anyone complain about the resistance of scientists to phrenology, the Piltdown man, polywater, cold fusion, and a host of other ideas that we now take to be mistaken. Instead, the usual question is why serious scientists were taken in. Are older scientists more able than younger scientists to tell truth from falsity? Or are older scientists supposedly more resistant to all new ideas, independent of the eventual fates of those ideas? On my reading, Planck's principle concerns the latter. Once again, older scientists as they were conceived of at the time are supposedly more resistant than younger scientists to all new ideas, regardless of how we judge, in retrospect, the truth-values of these ideas.

Ideas and theories are very difficult to individuate. How much of a new theory must other scientists accept in order for us to count them as accepting that theory? Darwinism, for example, is a very general, heterogeneous theory. Different elements have had very different fates. If one of Darwin's contemporaries had to agree with Darwin on all his basic tenets in order to be counted as accepting Darwin's theory, then very few scientists could be considered Darwinians. (Numerous evolutionary biologists can be found making claims about what the "essence" of Darwinism is, but such pronouncements are simply efforts to throw Darwin's mantle over their own shoulders.) In order to test Planck's principle for something like Darwinism, one element of the general theory must be singled out, for example, the basic claim that species evolve. One might disagree with Darwin about the role of sexual selection in evolution and still count as accepting his theory, but it would be difficult to consider anyone a Darwinian who refused to acknowledge that species evolve. Of course, Darwin's belief that species evolve was the least original part of his theory (for further discussion, see Hull 1985 and Numbers 1998).

Must older scientist literally die? Retirement from scientific life is good enough. The relevant generation time is professional, not biological. After all, roughly 75 percent of professional scientists had accepted the evolution of species by 1869 without any upsurge in the death rates of scientists [End Page 216] (Hull, Tessner, and Diamond 1978). In different areas of science, periods of productivity vary tremendously. A scientist who is thirty years old might be considered young in one field, over the hill in another (see Merton and Zuckermann 1973; McCann 1978; Messeri 1988; and Rappa and Debackere 1993 for further discussion).

The preceding discussion might sound as if it were designed to discourage anyone from attempting to test hypotheses about science, but it is not. In spite of the need to "operationalize" hypotheses about science in order to test them, testing can still be done. Scientists do it repeatedly in their various areas of science. Students of science can do the same. After all, two decades ago, Tessner, Diamond, and I were able to reach at least some tentative conclusions about the applicability of Planck's Principle to the Darwinian revolution, and the sophistication of the literature has only increased since then. For example, we discovered that when comparisons between the mean age in 1859 of scientists who accepted evolution between 1859 and 1869 is compared to those who still held out after 1869, the results are statistically significant (39.6 to 48.1). However, if one looks just at scientists who came to accept evolution between 1859 and 1869, age accounts for less than ten percent of the variance. Of course, in studies such as these, 10 percent is substantial. It is also true, as pointed out by Sulloway, that small variance does not imply small effect size.

However, the correlation between age and acceptance of the evolution of species after 1859 is not as dramatic as participants and later commentators alike have claimed. It is statistically significant but not so overwhelming that participants should have picked it up through casual observation. My suspicion is that revolutionaries such as Planck and Darwin were not reacting to the resistance of older scientists as such but to the resistance of the leaders in their respective fields--the scientists whose ideas were under attack. The relevant issue is professional age and standing in the community, not biological age. Elite scientists reacted negatively to the views of Planck and Darwin, not because they were old but because they were the authors of the views that were under attack. For the same reason, prestigious professional journals tended to ignore Darwin's theory (Burkhardt 1974).

The message of the preceding example is that even a claim as apparently straightforward as Planck's principle requires extensive reformulation and operationalization before it can be tested. As a result, any attempts to test hypotheses about science are likely to result in their being modified extensively. In attempting to test Planck's principle, I was forced to address questions that would have never occurred to me independently of my attempt to test it. I understand the principle much more deeply and thoroughly after testing it than before. I now have good reason to think [End Page 217] that, on certain interpretations, Planck's principle is true; on other interpretations, it is false. The important consideration involves differences in interpretations. Although we all thought that we understood Planck's principle, we were seriously mistaken. Vague ideas floated before our eyes, and that is all. (For more recent discussions of Plank's principle, see Rappa and Debackere 1993; as well as Levin, Stephan, and Walker 1995; and Diamond 1997.)

 

Birth Order and Science

The issue that I investigated in the preceding section is as narrow and tractable as any issue is likely to get in the study of science. Others are sure to be much more complicated. Almost three decades ago, Sulloway began to investigate the effects of birth order on science. In the course of his investigations, he has expanded his study to include over forty variables, twenty-eight scientific controversies, and hundreds of scientists (Sulloway 1996). Starting with evolutionary theory, Sulloway discovered a marked tendency for laterborns to accept Darwin's theory, but as he expanded his investigation to include other scientific controversies, a more complicated correlation emerged. New theories with a strongly conservative cast attract firstborns, while those that are ideologically radical attract laterborns.

At first glance, deciding birth order would seem to be as straightforward a task as determining a scientist's age, but Sulloway soon discovered that it is much more complicated than one might expect. For example, the sex of one's siblings, the spacing between siblings, one's precise birth rank, and overall number of siblings all turned out to matter. In male-dominated societies, having an older brother is quite different from having an older sister. In addition, Sulloway discovered that a firstborn whose closest sibling is more than six years younger functions as if he were an only child, whereas twelve years have to elapse before a laterborn functions as an only child. Although Louis Agassiz was his parents' fifth child, he functioned as a firstborn because the other four died in infancy. Parents also remarry, combine and separate families, and much, much more. Sulloway had to make decisions on these and other issues, including some of the same issues that I had to address with respect to Planck's principle. Who is to count as a scientist or as accepting Darwin's theory? On the second score, he opted for more stringent requirements than I did. In addition to accepting the evolution of species, a Darwinian had to acknowledge natural selection as an important factor in evolution (Sulloway 1996, p. 29).

More problematic still is rating the controversies on an ideological scale running from conservative to radical. Sulloway opted to have experts make these decisions intuitively based on their own expertise. Although lots of in-principle arguments can be raised against the use of expert opinions in [End Page 218] making such a rating, the results are more important than the arguments. In Sulloway's study, the average interrater reliability turned out to be quite high. At the very least, these authorities shared common biases. As impressionistic as expert opinion may be, it has proved to be reliable in certain contexts. For example, practicing taxonomists are very good at estimating taxonomic relationships--except in very special circumstances--and those systematists who have attempted to quantify taxonomic judgments have gradually uncovered what these special circumstances are (see, e.g., Fernholm, Brener, and Jšrnvall 1989).

The most important bias that present-day observers are likely to introduce stems from our presentist perspective. Although Sulloway acknowledges such difficulties, I think they pose one of the most serious problems for his undertaking. He treats certain themata as if they were independent of time and place. Firstborns tend to prefer continuity, order, causality, hierarchy, and essentialism, while laterborns are inclined toward discontinuity, chaos, acausality, equality, and population thinking. However, the assertive content of these themata seem to have changed through time. For example, postmodernist radicals take "essentialism" to be the primary source of the evils that have afflicted societies throughout history. Hence, they would be counted as confirming Sulloway's hypothesis. But as postmodernists use "essentialism," it does not contrast with population thinking but with realism, and realism is not on Sulloway's list. Idealism has raised its head periodically in the course of science. In Darwin's day it was a conservative view, but today it is quite radical (Webster and Goodwin 1996). How are we to score scientists from one age to the next?

I do not raise these problems in order to discount or reject Sulloway's studies. To the contrary, these issues arise only because Sulloway chose to take birth order seriously. Nor do I think that these problems are insuperable. The reason that certain students of science are likely to resist studies such as those conducted by Sulloway is that they seem to imply that certain factors, which we traditionally think do not or should not have a significant effect on science, actually do. However, the traditional view of science is not in the least threatened if increased financial support for science leads to its acceleration, or if firstborns tend to be overrepresented in the rank of eminent scientists. Truth is truth regardless of the birth order of those who discovered it. However, the correlations that Sulloway found between birth order and preference for certain sorts of ideas do pose problems. If firstborns tend to have disproportionate effects on science and prefer certain ideas, then the cognitive content of science is likely to reflect these preferences. If science is to reflect the world in which we live more strongly than the character of scientists themselves, it must incorporate one or more mechanisms capable of neutralizing the effects that Sulloway [End Page 219] has discovered (see Ruse 1996 for an exhaustive detailing of the effects that a belief in progress has had on science, especially evolutionary theory).

 

The Role of Novel Predictions in Science

Over a decade ago, the people at Virginia Polytechnic Institute and State University decided that it was about time to subject some of the key claims made by theorists of scientific change to the same kind of "empirical scrutiny that has been so characteristic of science itself" (Donovan, Laudan, and Laudan 1988, p. 3). In the past, students of science have sporadically attempted not just to illustrate general principles about science but also to test them by reference to episodes in science. The trouble is that the results of these studies are not comparable. They embody different assumptions, attempt to test different principles, and proceed in very different ways. In this respect, science studies is not all that different from other areas of science. Certain naive investigators (e.g., Collins 1985) were shocked to discover that scientists do not replicate each other's work all that often or all that precisely. They tend to concentrate on results that conflict with their own views, and when they do run these experiments, they are rarely narrow replications. Scientists modify certain aspects of the experimental setup for a variety of reasons. However, in many cases, these differences are so slight that the studies can be compared. One problem with the empirical studies that have been done in the study of science is that they are so different that they cannot be meaningfully compared.

The folks at Virginia Tech decided to remedy this situation. They gave over thirty historians, philosophers, and sociologists of science the task of evaluating a variety of claims commonly made about science by members of the post-positivist historical school in the philosophy of science. They ended up publishing the results of only sixteen of these case studies (Donovan, Laudan, and Laudan 1988). The organizers of this science studies research program were well aware that their activity was necessarily reflexive. For example, one of the theses of the historical school is that no thesis can be expressed in an entirely theory-free vocabulary. Popperians use one vocabulary, Kuhnians another, Laudanians yet another, and so on. If the claims to be tested are set out in Popperian terms, then the Popperians are likely to have an edge. Hence, if these claims withstand a test, it may simply be a function of the bias built into the formulation of the problem situation in the first place. Of course, one of the best ways to refute the principles of a particular school is to set them out in their own terms and then show that they are mistaken. If you can refute the views of a particular school using the vocabulary of that school, then you certainly have presented a strong refutation.

The people at Virginia Tech took another tack--setting out and classifying [End Page 220] the principles to be tested as neutrally as possible. The hope was that any significant biases built into their terminology and evaluative procedures would be discovered in the course of the study. Of all the principles that they proposed to test, I will discuss only one--the role of novel predictions in science. Controversies in the English-speaking community over the role of novel predictions in science go back from Popper (1962), Hempel (1966), and Lakatos (1970), to at least Mill (1843) and Whewell (1849). The problem is not just that these philosophers meant different things by "novel predictions" but that they were not all that good about telling their readers what they actually meant by this phrase. One important distinction is between the phenomenon actually being observed (i.e., "known") and whether it follows from a particular theory. Four possibilities present themselves:

 

(i) Some phenomena are well known (they have been observed) and follow from one or more accepted theories. They are explained.

(ii) Some phenomena are not known (they have never been observed) but follow from a particular theory. These are epistemologically novel predictions.

(iii) Some phenomena are known to occur but have yet to be explained by current theories. They are curiosities.

(iv) Finally, some phenomena have not been experienced by anyone, and no theory implies that they should exist. They are unknown.

 

The situations relevant to our discussion are (ii) and (iii). In (ii) a phenomenon has yet to be noticed by the relevant scientists. Hence, it cannot contribute to the construction of a theory. However, once the theory has been sufficiently well constructed, it might well imply that such a phenomenon should occur. These scientists go check and viol‡, they confirm this epistemologically novel prediction. In (iii) a phenomenon has been observed, but it has yet to be derived from any theory. Perhaps no theory currently exists from which it can be derived, or it could be the case that a theory exists, but no one has yet to produce the derivation. It is derivable but not yet derived. After the fact, a scientist shows how this phenomenon can be derived from this new theory. The phenomenon is not novel, but the derivation is. This situation is termed "use novelty".

Why should the prediction of epistemologically novel phenomena be all that special? The usual answer turns on the avoidance of ad hoc hypotheses. If scientists are already aware of a particular phenomenon, they can build it into their theory from the start. Hence, no one is much impressed when they extract from a theory the very phenomenon that they built into it in the first place. However, if the scientists constructing a theory are not aware of this particular phenomenon, they cannot incorporate it into their [End Page 221] theory in advance. Hence, the derivation of this unknown phenomenon and its later observation can be very persuasive.

This line of reasoning can be found in Hempel's discussion of novel predictions. According to Hempel (1966, p. 37), "it is highly desirable for a scientific hypothesis to be confirmed by 'new' evidence--by facts that were not known or not taken into account when the hypothesis was formed." But immediately thereafter, Hempel (1966, p. 38) brings himself up short:

from a logical point of view, the strength of the support that a hypothesis receives from a given body of data should depend only on what the hypothesis asserts and what the data are: the question of whether the hypothesis or the data were presented first, being a purely historical matter, should not count as affecting the confirmation of the hypothesis.

In the first quotation, Hempel distinguishes between (ii) epistemological novelty (facts that were not known) and (iii) use novelty (facts that were not taken into account). According to the "positivist" school that Hempel represents, scientists might take the prediction of phenomena that have yet to be observed as providing greater support to a hypothesis than the explanation of phenomena that have already been observed, but they are mistaken to do so. Similar observations follow for use novelty. Either an observation is derivable from a set of premises or it is not, regardless of whether anyone has ever actually derived it. Science must be extracted from the contingencies of history and rationally reconstructed entirely in terms of inference. An observation statement follows from a set of premises or it does not, period. Attitudes such as this one are precisely what advocates of the historical school have attempted to combat.

However, when we turn to the work of Lakatos (1970) as a representative of the historical school, his views on this score are not as different from those of Hempel as one might expect.2 Lakatos's methodology of scientific research programmes is "historical" in the sense that the unit of evaluation is a temporal sequence of theories. Such a sequence of theories is theoretically progressive if each new theory in the sequence predicts "some novel, hitherto unexpected fact" (Lakatos 1970, p. 118), but as was his habit, Lakatos immediately transforms this apparently straightforward claim in counterintuitive ways. According to Lakatos (1970, p. 156), when a new research programme explains "old facts" in novel ways, they count as [End Page 222] "new facts"! In addition, all such appraisals are a matter of hindsight. It may take a long time before a research programme can be "seen to produce 'genuinely novel' facts" (Lakatos 1970, p. 156). Unlike Hempel, Lakatos's scientific research programmes have a temporal dimension but one that is crucially retrospective. Even determining which statements are part of a particular scientific research programme is a function of hindsight. Lakatos's method of scientific research programmes may be historical, but it is Whig history.

Before proceeding further, a word must be said about the "rational reconstructions" of which "positivist" philosophers of science are so fond, especially when we have emphasized all the reconstruction that must be carried out if hypotheses are to be tested empirically. The purpose of rational reconstructions is to make logical connections more apparent. Real science tends to be very messy, so messy that its logical structure is obscured. Rational reconstructions lay bare the logical structure by eliminating all historical, psychological, and sociological contingencies. The sort of reconstruction that I discussed previously involves formulating rough-and-ready operational criteria to aid in the testing of more general and abstract hypotheses. In this respect, these reconstructions are the exact opposite of rational reconstructions. They reintroduce the very contingencies that rational reconstructions are designed to exclude.

All of the preceding concerns what the principle of novel prediction actually asserts and whether it should play an important role in science. It is quite another question whether scientists actually do make recourse to this principle. The three authors in Donovan, Laudan, and Laudan (1988, pp. 18-20) who discuss this principle at any length all conclude that it did not play a very significant role in the scientific episodes that they studied (Hoffman 1988, Frankel 1988, and Nunan 1988).3 Hence, the views of "positivists" on this score do not conflict with actual scientific practice. However, such coincidences do not function as confirmation for "positivists" because they do not think that philosophical views such as theirs can be tested empirically in the first place.

Initially, the function of epistemically novel predictions was to guard against the introduction of ad hoc hypotheses. If scientists give no special weight to novel predictions, then perhaps they are not as worried about ad hoc theorizing as they and students of science claim that they should be. [End Page 223] Was the introduction of so many epicycles in Ptolemaic astronomy part of the legitimate articulation of this scientific theory or an instance of blatant ad hocery? How about the introduction of epistatic genes by Mendelian geneticists when transmission patterns did not accord with the basic principles of Mendelian genetics? Was the postulation of epistatic genes ad hoc? More often than not, claims about which hypotheses are ad hoc, and which are not, are largely a function of public relations. "My hypotheses are firmly grounded in careful observation, while yours are invented merely for the purposes at hand."

More fundamentally, claims about ad hoc hypotheses need not turn on the content of these claims but on a commitment to do something with them. A scientist can introduce what looks very much like an ad hoc hypothesis. If that is all that the scientist does with it, then it may well turn out to be ad hoc. However, if this scientist continues to work on the processes and entities postulated, developing and modifying them as he or she continues to work on the overall theory, then they may become a legitimate part of the theory. They may well have been introduced to neutralize a particular problem but were expanded to become an integral part of the theory, implying much more than the phenomenon that they were introduced to handle.

 

Conclusion

The main lessons to be learned from the preceding summaries of three different attempts to test hypotheses about science are that formulating hypotheses about science in ways that can be tested empirically is very difficult and that the necessary clarifications cannot be anticipated prior to our attempts at testing. One reason that novel predictions cannot play the clear-cut epistemological role that some have attributed to them is that the course of science does not run smoothly. On the basis of minimal data (some of which is likely to turn out to be mistaken or misleading) and very hazy theories, tests are run. Before these tests are completed, the scientists running them have already reformulated some of their hypotheses and made adjustments in their investigative techniques. Frequently, early test runs are not worth completing. Instead, new sequences of tests are devised and carried to partial completion, each more sophisticated and determinate than the ones before. Order is introduced only long after the fact, when it comes time to publish the results of one's research.

If scientists already knew everything that they needed to know before they started their empirical investigations, a single experiment might prove conclusive. Of course, scientists do not know in advance everything that they need to know. As a result, sequences of tests need to be run in order to make proper use of the ad hoc hypotheses that scientists formulate, [End Page 224] and in the process some of these ad hoc hypotheses become transmuted into legitimate scientific hypotheses. Lakatos is right about this much. We can decide which hypotheses are ad hoc, and which are not, only in retrospect. In a genuinely historical view of science, many decisions can be made only in retrospect.

Parallel lessons hold for the study of science itself. I did my study on age in my spare time, with the help of two graduate students toward the end. (Incidentally, I did the data collecting because these graduate students did not know enough about the period under investigation to do it properly.) Sulloway was fortunate to be awarded a five-year grant from the MacArthur Foundation. In a paper that Rachel Laudan delivered at the History of Science Society meetings in 1990, she remarked on the inability of the people at Virginia Tech to get adequate funding for their project. The methods that Sulloway and I used did not require large research teams and massive financial expenditures. We could carry out our investigations in relative isolation for very little money, but if science teaches us anything about science, it implies that the investigation of certain problems requires large numbers of people working in close cooperation over many years. Haphazard, local, short-term investigations are not good enough. The people at Virginia Tech learned a lot about their research program as it progressed. By the time that the first cycle was complete, they knew enough to make the second cycle even more determinate. Unfortunately, no second cycle was possible. The funding ceased, and the members of this research team went their separate ways.

The conceptual obstacles confronting science studies are formidable. The disciplinary obstacles are even more formidable. Not only are those of us who study science subdivided into numerous factions, each fighting to maintain its own turf, but also we do not have a tradition of the sort of funding necessary to make headway on the kinds of problems that we are addressing. It also does not help that science studies is seen by many scientists as a threat, a perception that is not totally without foundation. Many of those who study science are more interested in debunking it than in understanding it. The result has been "Science Wars." 4 Relativist students of science present numerous case studies to show how unimportant anything that might be termed evidence is in changing scientists' minds, [End Page 225] but something is desperately wrong with presenting evidence to show how irrelevant evidence actually is. Now that students of science have retraced all the familiar ground that generations of philosophers have trod before them, perhaps we can abandon our intense fascination with philosophical puzzles and return to studying science.

 

 

Northwestern University

 

David Hull is Dressler Professor in the Humanities in the Department of Philosophy, Northwestern University, Evanston, Ill. He is past president of the Philosophy of Science Association, the Society of Systematic Zoology, and the International Society for the History, Philosophy and Social Studies of Science. He has written or edited eleven books, including Darwin and His Critics (1973), Philosophy of Biological Science (1974), Science as a Process (1988), The Metaphysics of Evolution (1989), and Philosophy of Science (1998) with Michael Ruse.

 

Notes

1. The following authors discuss or at least mention Planck's principle: Lavoisier ([1777] 1862, 2:505); Whewell (1837, 2:157); Whewell (1851, p. 139); Comte (1853, 2:152); Darwin (1859, pp. 481-82); Lyell (1881, 2:253); Darwin (1899, 2:85, 218); Huxley (1901); Loewenberg (1932, p. 687); Planck (1936, p. 97); Planck (1949, pp. 33-34); Barber (1961, p. 596); Kuhn (1961, pp. 161-90); Kuhn (1962, pp. 89-90, 150-51); Oakley (1964); Hagstrom (1965, pp. 283, 291); Samuelson (1966, 2:1517-18); Greenberg (1967, p. 45); Scheffler (1967, pp. 18-19); Feyerabend (1970, p. 203); Dolby (1971, p. 19); Cole and Cole (1973, p. 82); Holton (1973, p. 394); Merton and Zuckermann (1973, pp. 497-559); Montgomery (1974, p. 115); Paul (1974, p. 412); Wisdom (1974, 2:829); Bondi (1975, p. 7); Cantor (1975, p. 196); Cole (1975, p. 181); Gunther (1975, p. 458); Brush (1976, p. 640); Chubin (1976, p. 464); Garber (1976, p. 96); Knight (1976, p. 14); Stegmuller (1976, p. 148); Rosenkrantz (1977, p. 251); Blackmore (1978, pp. 347-49); Blua (1978, p. 197); Hull, Tessner, and Diamond (1978, pp. 717-23); McCann (1978, pp. 21, 34, 61-62, 79-80, 90-91, 101, 119); Nitecki, Lemke, Pullman, and Johnson (1978, pp. 661-64); Barash (1979, p. 240); Garvey (1979, p. 15); Hufbauer (1979, p. 744); Mahoney (1979, p. 372); Coats (1980, pp. 190-92); Diamond (1980, pp. 838-41); Newton-Smith (1981, p. 235); Broad and Wade (1982, p. 135); Cock (1983, p. 40); Grayson (1983, p. 208); Messeri (1988, pp. 91-112); Perrin (1988, pp. 105-24); Darwin (1991, p. 279); Stephan and Levin (1992); Margolis (1996, p. 136).

 

2. The classificatory criteria that I am using in placing particular philosophers in particular schools are extremely crude and not very faithful to history. In many important respects, Hempel is anything but a positivist, and Lakatos retains many positivistic elements.

 

3. One feature of empirical testing is that results may conflict with one's own preferences. For example, I think that novel predictions should lend greater support to a hypothesis than the derivation of the commonplace. I also thought that practicing scientists have more sense than their positivist commentators and treat novel predictions as special. Unfortunately, on the basis of these three studies, I am forced to admit that I was wrong.

 

4. Several publications have really heated up the war between the relativist, deconstructionist critics of science and their realist, "positivist" defenders: Gross and Levitt (1994); Gross, Levitt, and Lewis (1996); and Sokal (1996a, 1996b). Gross, Levitt, and Lewis assay what they take to be left-wing, deconstructionist, postmodernist critics of science, while Sokal wrote a parody of the sort of prose that he thought characterizes this same literature and got it published, not as a parody but as a genuine contribution. The reactions have been numerous and heated: see Richards and Ashmore (1996); Fish (1996); Fuller (1997); Dickson (1997a, 1997b); Gottfried and Wilson (1997); Macilwain (1997a, 1997b); Gross (1997); Edge (1997); Forman (1997a, 1997b); Robinson (1997); Levitt (1997); Trefil (1997); Herschbach (1997); Sandler (1997); Ziegler (1997); and Gibbons (1997). One characteristic of the various defenses of the new science studies is that these defenders are rapidly taking back everything that was new about the "new" science studies. No matter how crystal clear their prose may have been, their critics have totally misconstrued them as saying something novel and interesting.

 

References

Barash, D. 1979. The Whisperings Within. New York: Penguin.

Barber, B. 1961. "Resistance by Scientists to Scientific Discovery." Science 134:596-602.

Blackmore, J. 1978. "Is Planck's 'Principle' True?" The British Journal for the Philosophy of Science 29:347-49.

Blau, J. R. 1978. "Sociometric Structure of a Scientific Discipline." Research in Sociology of Knowledge, Science and Art 1:191-206.

Bloor, D. 1997. "Obituary of Thomas Kuhn." Social Studies of Science 27:498-502.

Bondi, H. 1975. "What Is Progress in Science?" Pp. 1-10 in Problems of Scientific Revolutions. Edited by R. HarrŽ. Oxford: Clarendon.

Broad, W., and N. Wade. 1982. Betrayers of the Truth. New York: Simon and Schuster.

Brush, S. G. 1976. The Kind of Motion We Call Heat. New York: American Elsevier.

Burkhardt, F. 1974. "The Learned Societies." Pp. 32-74 in The Comparative Reception of Darwinism. Edited by T. F. Glick. Austin and London: University of Chicago Press.

Cantor, G. N. 1975. "The Edinburgh Phrenology Debate: 1803-1828." Annals of Science 32:195-218.

Chubin, D. E. 1976. "The Conceptualization of Scientific Specialties." The Sociological Quarterly 17:448-76.

Coats, A. W. 1980. "The Historical Context of the 'New' Economic History." The Journal of European Economic History 9:185-207.

Cock, A. G. 1983. "William Bateson's Rejection and Eventual Acceptance of Chromosome Theory." Annals of Science 40:19-60.

Cole, J., and S. Cole. 1973. Social Stratification in Science. Chicago: University of Chicago Press.

Cole, S. 1975. "The Growth of Scientific Knowledge." Pp. 175-220 in The Idea of Social Structure. Edited by L. A. Coser. New York: Harcourt, Brace and Jovanovich.

Collins, H. M. 1985. Changing Order: Replication and Induction in Scientific Practice. London: Sage.

Comte, A. 1853. The Positive Philosophy. Vol. 2. London: John Chapman.

Darwin, C. 1859. On The Origin of Species. London: Murray.

------. 1899. The Life and Letters of Charles Darwin. Vol. 2. New York: Appleton.

------. 1991. The Correspondence of Charles Darwin. Vol. 7. Cambridge: Cambridge University Press.

Desmond, A. 1989. The Politics of Science. Chicago: University of Chicago Press.

Diamond, A. M. 1980. "Age and Acceptance of Cliometrics." Journal of Economic History 40:838-41.

------. 1997. "The Economics of Science." Knowledge and Policy 9:3-142.

Dickson, D. 1997a. "Champions or Challengers of the Cause of Science?" Nature 387:333-34.

------. 1997b. "The 'Sokal Affair' Takes Transatlantic Turn." Nature 385:381.

Dolby, R. G. A. 1971. "Sociology of Knowledge in Natural Science." Science Studies 1:3-21.

Donovan, A., L. Laudan, and R. Laudan, eds. 1988. Scrutinizing Science: Empirical Studies of Scientific Change. Dordrecht: Kluwer Academic Press.

DuprŽ, J. 1990. "Scientific Pluralism and the Plurality of Sciences: Comments on David L. Hull's Science as a Process." Philosophical Studies 69:61-76.

Edge, D. 1997. "Evolution Teaching." Science 274:904.

Fernholm, B., K. Brener, and H. Jšrnvall, eds. 1989. The Hierarchy of Life: Molecules and Morphology in Phylogenetic Analysis. Amsterdam: Elsevier.

Feyerabend, P. 1970. "Consolations for the Specialist." Pp. 197-230 in Criticism and the Growth of Knowledge. Edited by I. Lakatos and A. Musgrave. London: Cambridge University Press.

Fish, S. 1996. "Professor Sokal's Bad Joke." New York Times, 21 May, Opinions and Editorials section.

Forman, P. 1997a. "Assailing the Seasons." Science 276:750-52.

------. 1997b. "Deconstructing Science." Science 276:1955.

Frankel, H. 1988. "Plate Tectonics and Inter-Theory Relations." Pp. 269-87 in Scrutinizing Science: Empirical Studies of Scientific Change. Edited by A. Donovan, L. Laudan, and R. Laudan. Dordrecht: Kluwer Academic Press.

Fuller, S. 1997. "Out of Context." Nature 385:109.

Garber, E. 1976. "Some Reactions to Planck's Law, 1900-1914." Studies in History and Philosophy of Science 7:89-126.

Garvey, W. D. 1979. Communication: The Essence of Science. New York: Pergamon Press.

Gibbons, A. 1997. "Cultural Divide at Stanford." Science 276:1783.

Gottfried, K., and K. G. Wilson. 1997. "Science as a Cultural Construct." Nature 386:545-47.

Grayson, D. K. 1983. The Establishment of Antiquity. New York: Academic Press.

Greenberg, D. S. 1967. The Politics of Pure Science. New York: New American Library.

Gross, P. R. 1997. "Characterizing Scientific Knowledge." Science 275:143.

Gross, P. R., and N. Levitt. 1994. Higher Superstition. New York: Johns Hopkins University Press.

Gross, P. R., N. Levitt, and M. W. Lewis. 1996. The Flight from Science and Reason. New York: New York Academy of Sciences.

Gunther, A. E. 1975. A Century of Zoology. New York: Science History.

Hagstrom, W. 1965. The Scientific Community. New York: Basic Books.

Hempel, C. G. 1966. Philosophy of Natural Science. Englewood Cliffs: Prentice-Hall.

Herschbach, D. 1997. "Deconstructing Science." Science 276:1954.

Hoffmann, J. R. 1988. "AmpŹre's Electrodynamics and the Acceptability of Guiding Assumptions." Pp. 201-17 in Scrutinizing Science: Empirical Studies of Scientific Change. Edited by A. Donovan, L. Laudan, and R. Laudan. Dordrecht: Kluwer Academic Press.

Holton, G. 1973. Thematic Origins of Scientific Thought. Cambridge, Mass.: Harvard University Press.

Hufbauer, K. 1979. "A Test of the Kuhnian Theory." Science 204:744-45.

Hull, D. L. 1985. "Darwinism as an Historical Entity: A Historiographic Proposal." Pp. 773-812 in The Darwinian Heritage. Edited by D. Kohn. Princeton: Princeton University Press.

------. 1988. Science as a Process. Chicago: University of Chicago Press.

Hull, D. L., P. Tessner, and A. Diamond. 1978. "Planck's Principle." Science 202:717-23.

Huxley, T. H. 1901. Life and Letters of Thomas Henry Huxley. New York: Appleton.

Kim, K-M. 1994. Explaining Scientific Consensus: The Case of Mendelian Genetics. New York: Guilford Press.

Knight, D. 1976. The Nature of Science. London: A. Deutsch.

Kuhn, T. 1961. "The Function of Measurement in Modern Physical Theories." Isis 52:161-90.

------. 1962. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.

Ladd, E. C., and S. M. Lipset. 1977. "Survey of 4,400 Faculty Members at 161 Colleges and Universities." The Chronicle of Higher Education, 21 November, 12; 28 November, 2.

Lakatos, I. 1970. "Falsification and the Methodology of Scientific Research Programmes." Pp. 91-195 in Criticism and the Growth of Knowledge. Edited by I. Lakatos and A. Musgrave. Cambridge: Cambridge University Press.

Lavoisier, A-L. (1777) 1862. "RŽflexions sur le Phlogiston." Memoires de l'AcadŽmŽ des Sciences. Reprint, in Oevres de Lavoisier. Vol. 2. Paris: Imprimerie ImpŽriale.

Levin, S. G., P. E. Stephan, and M. B. Walker. 1995. "Planck's Principle Revisited: A Note." Social Studies of Science 25:275-83.

Levitt, N. 1997. "Deconstructing Science." Science 276:1953.

Lightman, B. 1997. Introduction to Victorian Science in Context. Chicago: University of Chicago Press.

Loewenberg, B. 1932. "The Reaction of American Scientists to Darwinism." American Historical Review 38:687-701.

Lyell, C. 1881. Life, Letters and Journals of Sir. Charles Lyell, bart. Edited by K. M. Lyell. London: Murray.

Macilwain, C. 1997a. "Campuses Ring to a Stormy Clash over Truth and Reason." Nature 387:331-33.

------. 1997b. "'Science Wars' Blamed for Loss of Post." Nature 387:325.

Mahoney, M. J. 1979. "Psychology of the Scientist: An Evaluative Review." Social Studies of Science 9:349-75.

Margolis, L. 1996. "Gaia Is a Tough Bitch." Pp. 129-46 in The Third Culture: Beyond the Scientific Revolution. Edited by J. Brockman. New York: Simon and Schuster.

Matthew, P. 1831. On Naval Timbre and Arboriculture. London: Longman, Rees, Orme, Brown and Green.

McCann, H. G. 1978. Chemistry Transformed. Norwood, N.J.: Ablex Publishing Corporation.

Meltzer, L. 1956. "Scientific Productivity in Organizational Settings." Journal of Social Issues 12:32-40.

Mendel, G. 1865. "Versuche źber Pflanzen-hybriden." Verhandlungen des Naturforschenden Veriens Brźn 4:3-47 (appeared in 1866).

Merton, R. K., and Harriet Zuckermann. 1973. "Age, Aging, and Age Structure in Science." In A Sociology of Age Stratification. Edited by M. W. Riley, M. Johnson, and A. Foner. New York: Russell Sage Foundation, 1972. Reprint, Pp. 497-559 in The Sociology of Knowledge, edited by N. W. Storer, Chicago: University of Chicago Press.

Messeri, A. M. 1988. "Age Differences in the Reception of New Scientific Theories: The Case of Plate Tectonics Theory." Social Studies of Science 18:91-112.

Mill, J. S. 1843. A System of Logic. London: Longmans, Green and Co.

Montgomery, W. M. 1974. "German Editions of Important Works on Evolution." Pp. 81-116 in The Comparative Reception of Darwinism. Edited by T. Glick. Austin: The University of Texas Press.

Newton-Smith, W. H. 1981. The Rationality of Science. Boston: Routledge and Kegan Paul.

Nitecki, M. N., J. L. Lemke, H. W. Pullman, and M. E. Johnson. 1978. "Acceptance of Plate Tectonic Theory by Geologists." Geology 6:661-64.

Numbers, R. L. 1998. Darwinism Comes to America. Cambridge, Mass.: Harvard University Press.

Nunan, R. 1988. "The Theory of an Expanding Earth and the Acceptability of Guiding Assumptions." Pp. 289-314 in Scrutinizing Science: Empirical Studies of Scientific Change. Edited by A. Donovan, L. Laudan, and R. Laudan. Dordrecht: Kluwer Academic Press.

Oakley, K. P. 1964. "The Problems of Man's Antiquity." Bulletin of the British Museum (Natural History) 9(5):1-155.

Paul, H. W. 1974. "Religion and Darwinism: Varieties of Catholic Reaction." Pp. 403-36 in The Comparative Reception of Darwinism. Edited by T. Glick. Austin: The University of Texas Press.

Perrin, C. E. 1988. "The Chemical Revolution: Shifts in Guiding Assumptions." Pp. 105-24 in Scrutinizing Science: Empirical Studies of Scientific Change. Edited by A. Donovan, R. Laudan, and L. Laudan. Dordrecht: Kluwer Academic Press.

Planck, M. 1936. The Philosophy of Physics. New York: Norton.

------. 1949. Scientific Autobiography and Other Papers. Translated by F. Gaynor. New York: Philosophical Library.

Popper, K. R. 1962. Conjectures and Refutations. New York: Basic Books.

Price, D. de Solla. 1963. Little Science, Big Science. New York: Columbia University Press.

Rappa, M., and K. Debackere. 1993. "Youth and Scientific Innovation: The Role of Young Scientists in the Development of a New Field." Minerva 31:1-20.

Reskin, B. F. 1977. "Scientific Productivity and the Reward Structure of Science." American Sociological Review 42:491-504.

Richards, E., and M. Ashmore. 1996. "More Sauce Please! The Politics of SSK: Neutrality, Commitment and Beyond." Social Studies of Science 26:219-28.

Robinson, J. D. 1997. "Deconstructing Science." Science 276:1953.

Rosenkrantz, R. D. 1977. Inference, Method and Decision. Dordrecht- Holland: D. Reidel.

Ruse, M. 1996. Monad to Man: The Concept of Progress in Evolutionary Biology. Cambridge: Harvard University Press.

Samuelson, P. 1966. "The General Theory." Pp. 1517-33 in The Collected Papers of Paul Samuelson. Vol. 2. Edited by J. Stiglitz. Cambridge: MIT Press.

Sandler, N. 1997. "Deconstructing Science." Science 276:1954-55.

Scheffler, I. 1967. Science and Subjectivity. New York: Bobbs-Merrill.

Sokal, A. D. 1996a. "A Physicist Experiments with Cultural Studies." Lingua Franca, May/June, 62-64.

------. 1996b. "Transgressing Boundaries: Toward a Transformative Hermeneutics of Quantum Gravity." Social Text 14:217-252.

Stegmuller, W. 1976. The Structure and Dynamics of Theories. New York: Springer-Verlag.

Stephan, P. E., and S. G. Levin. 1992. Striking the Motherlode in Science: The Importance of Age, Place, and Time. New York: Oxford University Press.

Sulloway, F. J. 1996. Born to Rebel: Birth Order, Family Dynamics and Creative Lives. New York: Pantheon.

Trefil, J. 1997. "Deconstructing Science." Science 276:1953-54.

Webster, G., and B. Goodwin. 1996. Form and Transformation. Cambrdige: Cambridge University Press.

Whewell, W. 1834. "Review of Mary Somerville's 'On the Connexion of the Physical Sciences.'" Quarterly Review 51:54-68.

------. 1837. History of the Inductive Sciences. Vol. 2. London: John W. Parker.

------. 1849. Of Induction, with especial reference to Mr. J. Stuart Mill's System of Logic. London: J. W. Parker.

------. 1851. "On the Transformation of Hypotheses in the History of Science." Transactions of the Cambridge Philosophical Society 9:139-47.

Wisdom, J. O. 1974. "The Nature of Normal Science." In The Philosophy of Karl Popper. Vol. 2. Edited by P. A. Schilpp. LaSalle, Ind.: Open Court.

Yoels, W. C. 1973. "On 'Publishing or Perishing': Fact or Fable?" The American Sociologist 8:128-34.

Ziegler, C. A. 1997. "Deconstructing Science." Science 276:1955.